Encyclopedia-of-Chart-Patterns_PDFdrive.pdf

Encyclopedia
of
Chart Patterns

WILEY TRADING ADVANTAGE
Trading without Fear I Richard W. Arms, Jr.
Neural Network Time Series Forecasting ofFinancial Markets / E. Michael Azoff
Option Market Making I Alan J. Baird
Genetic Algorithms and Investment Strategies I Richard J. Bauer, Jr.
Technical Market Indicators I Richard J. Bauer, Jr., and Julie R. Dahlquist
Seasonality /Jake Bernstein
The Hedge Fund Edge I Mark Boucher
Encyclopedia ofChart Patterns I Thomas N. Bulkowski
Macro Trading and Investment Strategies I Gabriel Burstein
Beyond Technical Analysis I Tushar Chande
The New Technical Trader ! Tushar Chande and Stanley S. Kroll
Trading the flan I Robert Deel
New Market Timing Techniques I Thomas R. DeMark
The New Science ofTechnical Analysis / Thomas R. DeMark
Point and Figure Charting I Thomas J. Dorsey
Tradingfor a Living I Dr. Alexander Elder
Study Guide for Trading for a Living I Dr. Alexander Elder
The Day Trader's Manual I William F. Eng
The Options Course I George A. Fontanills
The Options Course Workbook I George A. Fontanills
Pattern, Price & Time /James A. Hyerczyk
Profitsfrom Natural Resources I Roland A. Jansen
The Trading Game I Ryan Jones
Trading Systems <b Methods, Third Edition I Perry Kaufman
Trading to Win I Ari Kiev, M.D.
The Intuitive Trader I Robert Koppel
Nonlinear Pricing I Christopher T. May
McMillan on Options I Lawrence G. McMillan
Trading on Expectations I Brendan Moynihan
Intermarket Technical Analysis I John J. Murphy
The Visual Investor I John J. Murphy
Beyond Candlesticks I Steve Nison
Cybernetic Trading Strategies I Murray A. Ruggiero, Jr.
The Option Advisor I Bernie G. Schaeffer
Fundamental Analysis I'Jack Schwager
Study Guide to Accompany FundamentalAnalysis I'Jack Schwager
Managed Trading I Jack Schwager
The New Market Wizards I Jack Schwager
Technical Analysis /Jack Schwager
Study Guide to Accompany TechnicalAnalysis /Jack Schwager
Schwager on Futures I Jack Schwager
Gaming the Market I Ronald B. Shelton
The Dynamic Option Selection System I Howard L. Simons
Option Strategies, Second Edition I Courtney Smith
Trader Vie III Victor Sperandeo
Campaign Trading/]ohn Sweeney
The Trader's Tax Survival Guide, Revised Edition I Ted Tesser
The Mathematics of Money Management I Ralph Vince
The New Money Management I Ralph Vince
Trading Applications ofJapanese Candlestick Charting I Gary Wagner and Brad Matheny
Trading Chaos I Bill Williams
New Trading Dimensions I Bill Williams
LongTerm Secrets to ShortTerm Trading I Larry Williams
Expert Trading Systems I John R. Wolberg
Encyclopedia
of
Chart Patterns
Thomas N. Bulkowski
John Wiley & Sons, Inc.
New York • Chichester . Weinheim • Brisbane • Singapore • Toronto

This book is printed on acidfree paper. ©
Copyright © 2000 by Thomas N. Bulkowski. All rights reserved.
Published byJohn Wiley & Sons, Inc.
Published simultaneously in Canada.
No part of this publication may be reproduced, stored in a retrieval system or transmitted in
any form or by any means, electronic, mechanical, photocopying, recording, scanning or
otherwise, except as permitted under Sections 107 or 108 of the 1976 United States Copyright
Act, without either the prior written permission of the Publisher, or authorization through
payment ofthe appropriate percopy fee to the Copyright Clearance Center, 222 Rosewood
Drive, Danvers, MA 01923, (978) 7508400, fax (978) 7504744. Requests to the Publisher for
permission should be addressed to the Permissions Department, John Wiley & Sons, Inc.,
605 Third Avenue, New York, NY 101580012, (212) 8506011, fax (212) 8506008,
EMail: PERMREQ @ WILEY.COM.
This publication is designed to provide accurate and authoritative information in regard to the
subject matter covered. It is sold with the understanding that the publisher is not engaged in
rendering professional services. If professional advice or other expert assistance is required, the
services of a competent professional person should be sought.
Library of Congress Cataloging-in-Publication Data:
Bulkowski, Thomas N., 1957
Encyclopedia of chart patterns / Thomas N. Bulkowski.
p. cm.—(Wiley trading advantage)
Includes index.
ISBN 0471295256 (alk. paper)
1. Stocks—Charts, diagrams, etc. 2. Commodities—Charts,
diagrams, etc. 3. Technical analysis. I. Title. II. Series.
HG4638.B85 2000
332.63722—dc21 9915789
To my parents, who continued to love me even after my homemade
rocket set the lawn onfire, and to myfourlegged bestfriend,
Rusty, who saved my life; it grieves me that I couldn't saveyours.
Printed in the United States ofAmerica.
10 9 8 7 6 5 4 3

Preface
When I was a little tyke I decided the easiest way to riches was to play the stock
market. It was, after all, a level playing field, a zerosum game with somebody
winning and somebody losing (hint: The winner is always the broker). All one
had to do to win was pick the stocks that went up and avoid the stocks that went
down. Easy.
I kept this in mind when I graduated from Syracuse University with an
engineering degree and showed up early for my first professional job. Eacl
morning I cracked open the newspaper and plotted my stock picks on a piec
of paper taped to the wall. Bob, my office mate, used the same newspaper to
select his stocks. I chose my selections using strict and exhausting fundamen
tal research, but Bob simply closed his eyes, twirled his hand around, and
plunged his finger into the newspaper. When he opened his eyes and removed
his finger, he announced another pick.
After several months of tracking both our selections, I made a startling
discovery: I was getting creamed. Bob's random selections were beating the tar
out of my carefully researched choices. I also discovered something else: I was
learning a lot by paper trading.
With the hesitancy and distrust inherited from my parents, I studied two
dozen firms before making my final selection and first purchase: I opened a
money market account. The timing was excellent; I was earning over 17% on
my cash. At first glance, the return might imply a very risky investment but it
was not. The prime rate was, after all, at 21%.
Flush with success, I gathered my courage and opened a brokerage
account and began investing with the few pennies I saved. Again, the timing
was excellent as I caught the beginning ofa major bull market. I bought a stock
near 3!
/2 and watched it go to 46l
/i—my first tenbagger.
Lest you think that everything was easy, consider what happened. My
stock portfolio was growing by leaps and bounds, but my professional career
was about to take a turn for the worse. After switching careers more often than
I sometimes like to admit, I landed at a job with a company I could finally call

viii Preface
home—a job that would last a lifetime, or so I thought. Almost six months after
my 10year anniversary with the company, I received a letter from the chair
man. He congratulated me on my decade with the company and looked forward
to even more success for me in the coming years. Six weeks later I was laid off.
I took stock of the situation and decided that, at the age of 36,1 had had
enough. Newspapers term guys like me The Missing Million. We are the ones
who, for whatever reason, leave their jobs and decide not to go back into the
workforce. We retire. Everyone, and I mean everyone (with the notable excep
tion of my cousin Mary Ann—bless her heart), thinks we are nuts. They are
right, of course!
For the longest time, I have been fascinated with technical analysis of
stocks. In the early years, I considered the little squiggles to be nothing short
of voodoo. Still, I was curious as to why the major brokerage houses were hir
ing technical analysts in droves. But I did not dare take my eye off the funda
mentals simply because I did not know anything about the technicals. Then I
discovered Technical Analysis of Stocks and Commodities magazine. During my
lunch hour, I would take the elevator down to the library and read back issues.
Although I had seen chart patterns in the stocks I bought, I never really
attached much significance to them. As some of my selections went sour, I
began to view chart patterns with more respect. The fundamentals always
looked good, but the technicals were signaling a trend change just as I was
about to pull the trigger. The stocks I bought either lost money outright or I
sold them too soon and cut my profits short.
Perhaps this has happened to you. You do your fundamental research on
a stock, then buy it only to watch it go nowhere for a year or more. Even worse,
once you get in, the stock tumbles. Had you looked at the chart the answer was
always there. Prices pierced a trendline, a headandshoulders top appeared out
of nowhere, the relative strength index signaled an overbought situation. In
short, any number of technical indicators were screaming that a buy now
would cost you your shirt. But you never saw the signs because you had your
eyes closed.
You are not alone; I did the same thing for years. I eventually got so frus
trated with the performance of my stock selections that I decided to do my own
research on technical analysis. I went to the library and read the same thing in
many books: A headandshoulders formation works most of the time. What
does that mean? Does it mean they are successful 51 % of the time or 90% of
the time? No one had the answer. I was not willing to risk my hard earned dol
lars on simple bromides. As an engineer I wanted hard, cold facts, not fuzzy
platitudes. So, I wrote this book.
At the back of the book is an Index of Chart Patterns. Ifyou suspect your
stock is making a chart pattern but do not know what to call it, the Index of
Chart Patterns is the first place to look. Page numbers beside each pattern
direct you to the associated chapter.
Preface ix
The chapters are arranged alphabetically, making it easy to locate the
chart pattern of interest. Within each chapter, you are first greeted with a
"Results Snapshot" of the major findings followed by a short discussion. Then,
a "Tour" invites you to explore the chart pattern. "Identification Guidelines,"
in both table form and indepth discussion, make selecting and verifying yo•
choices easier. For simpler chart patterns, the "Tour" and "Identificati
Guidelines" have been combined into one section.
No work would be complete without an exploration of the mistakes, a
the "Focus on Failures" section dissects the cause offailures. The allimport;
"Statistics" section follows. Once you can identify a chart pattern, know how it
is likely to perform, and are alert to possible failure indications, how do you
trade it? That is what the "Trading Tactics" and "Sample Trade" sections
explore. / ''.>
If you have ever worked on a car or done some woodworking, then you
will recognize the importance of selecting the right tool for the job. You would
not want to use a flathead screwdriver when a Phillips works better. Both do
the job but they are hardly interchangeable. Sometimes it is not a screwdriver
you should be using, but a chisel. Selecting the proper tools and knowing how
to use them is half the battle. This book is a lot like that, helping to sort the
wheat from the chaff. Sometimes a chart pattern is frightening enough that y
will want to take profits. At other times, the best trade that you can make is
none at all.
I cannot give you the experience needed to make money in the stock mar
ket using chart patterns. I can only give you the tools and say, "Go to work on
paper first." That is the first step in developing a trading style that works for
you, one you are comfortable with, one that improves as you do. Ifyou review
your paper trades, you will understand why a stoploss order is more than a
necessary evil: It is a useful tool. You will improve your ability to predict sup
port and resistance levels that will, in turn, allow you to tighten your stops and
get out near the top, cut your losses short, and let your profits ride. Simple.
You will discover why the measure rule is so important, especially in tur
bulent markets. Unless you are willing to suffer a 20% drawdown, you will
understand why the average gain quoted so often in this book may be a best
case scenario and will come to grips with why you are still struggling to make
it above the most likely gain. You may discover that your girlfriend loves dia
monds, but as a chart pattern, you cannot seem to make them pay. One word
says it all. Experience.
Good luck.
THOMAS N. BULKOWSKI
December 1999

Acknowledgments
Perhaps several times in your life, something happens that alters the direction
your life is taking. That happened to me several years ago when I brashly sub
mitted my first article to TechnicalAnalysis ofStocks and Commodities. Much to
my surprise and delight, the editor at the time, Thorn Hartle, published the
work. That single event sent me spinning offin a new direction.
Nearly a dozen articles later, I called Thorn and chatted with him about
an idea for a book. He steered me to Pamela van Giessen, senior editor for
John Wiley & Sons, Inc., publisher of this book. A single email ofmy idea to
her put a new set ofwheels in motion.
Simple words cannot express my thanks to these two outstanding indi
viduals. Ofcourse, there are many others such as my younger brother,Jim, the
unsung heroes that sometimes gave me a helping hand, formed my support
group, or gave me a good, swift kick in the butt. They are not forgotten.
T. N. B.

Contents
Introduction
1 Broadening Bottoms
2 Broadening Formations, RightAngled and Ascending
3 Broadening Formations, RightAngled and Descending
4 Broadening Tops
5 Broadening Wedges, Ascending
6 Broadening Wedges, Descending
7 BumpandRun Reversal Bottoms
8 BumpandRun Reversal Tops
9 Cup with Handle
10 DeadCat Bounce
11 Diamond Tops and Bottoms
12 Double Bottoms
13 Double Tops
14 Flags and Pennants
15 Flags, High and Tight
16 Gaps
17 Hanging Man
18 HeadandShoulders Bottoms
19 HeadandShoulders Bottoms, Complex
20 HeadandShoulders Tops
1
12
27
40
5;
72
87
100
119
135
153
165
182
197
213
227
240
252
262
276
290

xiv Contents
21 HeadandShoulders Tops, Complex
22 Horn Bottoms
23 Horn Tops
24 Inside Days
25 Island Reversals
26 Measured Move Down
27 Measured Move Up
28 OneDay Reversals
29 Outside Days
30 Pipe Bottoms
31 Pipe Tops
32 Rectangle Bottoms
33 Rectangle Tops
34 Rounding Bottoms
35 Rounding Tops
36 Scallops, Ascending and Descending
37 Shark32
38 Triangles, Ascending
39 Triangles, Descending
40 Triangles, Symmetrical Bottoms
41 Triangles, Symmetrical Tops
42 Triple Bottoms
43 Triple Tops
44 Wedges, Falling
45 Wedges, Rising
46 Weekly Reversals, Downside
47 Weekly Reversals, Upside
Statistics Summary
Index of Chart Patterns
Subject Index
305
320
332
343
356
369
382
394
404
417
429
439
453
4
6
6
477
487
501
511
529
545
560
576
590
603
617
631
642
654
663
669
Encyclopedia
of
Chart Patterns

Introduction
Jim is struggling.
He is the owner ofJCB Superstores and his competitor across town is
beating him up; there is blood all overJim's ledger. He decides it is time to take
off the gloves: JCB goes public. He uses the money from the initial public
offering to buy his competitor and add a few more stores around town.
With a growing sales base, Jim's clout allows him to negotiate lower
prices for the office supplies he is retailing. He passes on part of the savings to
his customers, while watching his margins widen, and plows the profits back
into building more stores.
Jim calls his friend, Tom, and tells him of his plans to expand the opera
tion statewide. They chat for a while and exchange business tactics on how best
to manage the expansion. When Tom gets off the phone, he decides to con
duct his own research on JCB. He visits several stores and sees the same thing:
packed parking lots, people bustling around with full shopping carts, and lines
at the checkout counters. He questions a few customers to get a sense of the
demographics. At a few stores, he even chats with suppliers as they unload their
wares. Back at the office, he does a thorough analysis of the financials and
looks at the competition. Everything checks out so he orders his trading part
ners to buy the stock at no higher than 10.
When news of the expansion plan hits the wires, the Street panics. It is,
after all, a soft economy and expanding willynilly when a recession looms is
daft, maybe even criminal, according to them. The stock drops below 10 and
Tom's crew makes its move. They quietly buy as much as they can without
raising suspicion. The stock rises anyway. It goes back up to 11, then 12, and
rounds over at 13 before heading back down. i

2 Introduction
Several months go by and the economic outlook is as bleak as ever. The
stock eases down to 9. After Tom checks in withJim for the latest public news,
Tom's team buys more. It is an easy score because investors are willing to
dump the stock especially as yearend tax selling approaches.
Six weeks later the company releases the sales numbers forJCB; they are
better than expected. The stock rises 15% in minutes and closes at 10%. And
that is just for starters. Six months later, it's clear the economy was never
in danger of entering a recession and everyone sees boom times ahead. The
stockhits 20.
Years go by, the stock splits a few times, and the holiday season looms.
Tom interviews a handful of customers leaving JCB Superstores and discovers
that they are all complaining about the same thing: The advertised goods are
missing. Tom investigates further and discovers a massive distribution prob
lem, right at the height of the selling season. JCB has overextended itself;
the infrastructure is simply not there to support the addition of one new store
each week.
Tom realizes it is time to sell. He tells his trading department to dump
the stock immediately but for no less than 281
4. They liquidate about a third of
their large holdings before driving the stock down below the minimum.
Since it is the holidays, everyone seems to be in a buying mood. Novice
investors jump in at what they consider a bargain price. The major brokerage
houses climb aboard and tout the stock, but Tom knows better. When the
stock recovers to its old high, his trading partners sell the remainder of their
holdings. The stock tops out and rounds over. During the next month and a
half, the stock drifts down, slowly, casually. There does not appear to be a rush
for the exits—just a slow trickle as the smart money quietly folds up shop.
Then news ofpoor holiday sales leaks out. There is a rumor about distri
bution problems, merchandising mistakes, and cash flow problems. Brokerage
firms that only weeks before were touting the stock now advise their clients to
sell. The stock plummets 39% overnight.
One or two analysts say the stock is oversold; it is a bargain and investors
should add to their positions. Many bottom fishers follow their brokers' rec
ommendation and buy the stock. Big mistake. The buying enthusiasm pushes
the price up briefly before a new round ofselling takes hold. Each day the stock
drops a bit lower, nibbling away like waves washing against a castle of sand. In
2 months' time, the stock is down another 30%.
The following quarter JCB Superstores announces that earnings will
likely come in well below consensus estimates. The stock drops another 15%.
The companyis trying to correct the distribution problem, but it is not some
thing easily fixed. They decide to stop expanding and to concentrate on the
profitability of their existing store base.
Two years later, Tom pulls up the stock chart. The dog has been flat for
so long it looks as if its heartbeat has stopped. He calls Jim and chats about the
outlookforJCBSuperstores.Jimgushesenthusiasticallyaboutanewretailing
The Database 3
concept called the Internet. He is excited about the opportunity to sell office
supplies online without the need for bricks and mortar. There is some risk
because the online community is in its infancy, butJim predicts it will quickly
expand. Tom is impressed, so he starts doing his homework and is soon buy
ing the stock again.
Investment Footprints
If you picture in your mind the price action ofJCB Superstores, you should
recognize three chart patterns: a double bottom, a double top, and a deadcat
bounce. To knowledgeable investors, chart patterns are not squiggles on a
price chart; they are the footprints of the smart money. The footprints are all
they need to follow as they line their pockets with greater and greater riches.
To others, such as Tom, it is hard work and pavement pounding before they
dare take a position in a stock. They are the ones making the footprints. They
are the smart money that is setting the rules of the game—a game anyone can
play. It is called investing.
Whether you choose to use technical analysis or fundamental analysis in
your trading decisions, it pays to know what the market is thinking. It pays to
look for the footprints. Those footprints may well steer you away from a cliff
and get you out of a stock just in time. The feet that make those footprints are
the same ones that will kick you in the pants, waking you up to a promising
investment opportunity.
This book gives you the tools to spot the footprints, where they predict
the stock is heading, how far it will travel, and how reliable the trail you are fol
lowing really is. The tools will not make you rich; tools rarely do. But they are
instruments to greater wealth. Use them wisely.
The Database
If you want to discover how much you do not know about a chart formation,
try teaching a computer to recognize one. I spent several months doing that
preparing for this book. The program helped me locate, analyze, and log well
over 15,000 formations. It is not a substitute for my eyes or my brain, just
another tool to augment my talent. Consider it another set of dispassionate
eyes, a friend nudging you and saying, "Look at this one here. It's a bumpand
run reversal."
When the starting gun went off, I selected 500 stocks, all with durations
of 5 years (each from mid1991 to mid1996) ofdaily price data on which to col
lect statistics. I included the 30 DowJones industrials and familiar names with
varying market capitalizations. Stocks included in the study needed a heartbeat

Introduction
(theywere not unduly flat over the 5year period) and did not have consistently
large daily price ranges (too thinly traded or volatile).
I usually removed stocks that went below a $1.00, assuming bankruptcy
was right around the corner. Most of the names in the database are popular
American companies that trade on the NYSE, AMEX, or NASDAQ. The
numerous illustrations accompanying each chapter give a representative sam
ple of the stocks involved.
Occasionally a chart formation came along that presented a problem. It
was so scarce that 2,500years (500 stocks times 5 years) ofdailyprice datawere
simply not enough. So I pulled from the database I use on a daily basis. It con
tains about 300 issues and begins where the other one ends.
Stock Performance from 1991 to 1996
Before reading about the various chart patterns in this book, it is wise to review
the performance of the stock market during the period. Figure 1.1 shows a
monthly price chart of the Standard & Poor's 500 stock index. Beginning in
mid1991, you can see that the market hesitated until January 1992. It had a
wild burst upward, perhaps due to the January effect, but trended downward
until May. (In case the January effect is unfamiliar to you, it is commonly
attributed to investors selling their stocks for tax reasons near year end then
buying back during January. The selling may or may not depress prices,
whereas the January buying gives them a temporary lift.) Toward the end of
1992, it looks as if the January effect occurred early, in December, when prices
broke through their malaise of consolidation and reached new highs. Then
it was off to the races, and prices rose on a steady tear until March 1994.
The market stumbled and moved up for 5 months then declined for 4
months. Beginning in 1995, the race resumed, but the pace accelerated. The
slope of the trend tilted upward noticeably until running into some turbulence
in early 1996.
What does all this mean? Viewed as a whole, the market during the 5
years used in my analyses plus the 2 or 3 additional years used sporadically but
not shown in Figure I.I, marks a very bullish environment. While the market
as a whole was going up gangbusters, many individual stocks were not so for
tunate. Some had steady downward trends. Others moved up smartly, rolled
over, and died (check out most semiconductor and semiconductor capital
equipment stocks in 1995).
During a soaring bull market, bullish chart patterns are more successful
by having fewer failures and longer uphill runs. They perform better, chum
ming along on a rising tide that lifts all boats.
Common sense suggests that bearish formations might fail more readily
with stunted declines. More likely, though, is that bearish patterns just disap
Averages and the Frequency Distribution
s & P soo
Figure 1.1 Standard & Poor's 500 stock index from 1991 to 1996.
pear; they never happen. You might think that stocks moving up would form
bearish reversals. Instead, diey just keep moving up, now and again pausing to
catch their breath before continuing the rise.
You can see this trend in the statistics. Bullish formations, those that typ
ically occur after a downward price trend and signal an upward reversal, hap
pen more often than bearish ones. Symmetrical triangles are a good example.
Triangleswithupside, bullishbreakouts occurred 225 times, whereas downside
breakouts happened 176 times. A favoring ofthe bullish trend is also evident in
many paired formations. Consider double bottoms and double tops. There
were 542 bottoms (bullish) and only 454 tops (bearish).
Even the statistics favor a bull market. A stock moving up can advance
50%, 100%, or even 1,000%. The gains can be unlimited, but what of the
declines? A stock can only lose 100%, or all ofits value, and nothing more.
Averages and the Frequency Distribution
The frequency distribution mentioned so often in this book deserves special
attention. Before I discuss it, however, let me explain averages. An average is
the sum ofthe numbers divided by the number ofsamples. Ifyou measure the
returns from five chart patterns and they are 30%, 40%, 50%, 60%, and 120%,
dien the average is 60%. That is the sum of the numbers (300) divided by 5
samples.

This example shows the effect large numbers have on the average. If the
120% gain is not in the series, the average drops to 180/4 or 45%. The single
large gain pulls the overall average upward, distorting the result. This distor
tion is important when discussing bullish formations. A 600% gain in one chart
formation can make a chart pattern appear more successful than it really is.
Instead of dropping off samples (by arbitrarily removing the large returns), I
use a frequency distribution.
The esoteric name frequency distribution is appropriate. To create a fre
quency distribution, find the highest and lowest values to give you the range.
Divide the range by 10 because you want to sort the numbers into 10 bins (10
is arbitrary, but commonly chosen). Then, you do just that—sort the numbers
into one of 10 ranges and place them in the bins. When finished, count how
often the numbers appear in each bin (the frequency). Note that you do not
add up the numbers, you just count how often they appear. It is a lot like see
ing troops on a battlefield. You really do not care how tall each one is, only that
they outnumber you. The results are the same: You wet your pants and run!
An example makes this clear. Look at Table 1.1. Suppose I am studying a
chart formation and have the gains for 50 patterns. For simplicity, suppose the
gains range from 5% to 95%. The first column in the table holds gains less
than 10% and the last column holds gains over 90%. I do not show them all in
the table, but I begin placing die gains into the different bins, and the first 10
gainsare 8%, 35%, 70%, 13%, 95%, 9%, 6%, 33%, 3%, and 63% (seeTable
11). When I finish placing the gains from the 50 formations into the table and
sum the columns, I see which column has the highest frequency. A count of
each column appears as the last row in the table and assumes all 50 formations
were sorted.
From the numbers in the bottom row, we see that the first column has the
highest frequency and represents those formations with gains ofless than 10%.
We might conclude that ifyou invested in a similar chart formation, your gain
is likely to be between zero and 10%, since that is where most (40%) ofdie for
mations reside. The average of the 50 gains will likely be higher than 10%,
especially if the higher ranges show either a large number of entries or repre
sent large gains.
I call the column with the highest frequency "the most likely gain."
Sometimes the sum of the columns are near to one another and so the most
likely gain is a range of values, such as 10% to 20%. Just because a chart pat
tern has a most likely gain of 10% does not mean that you will have a 10% gain
from trading your chart pattern. After all, ifyou trade the pattern well enough
and often enough, you should approach the results represented by the average.
However, I feel that the most likely gain gives the investor a better under
standing of the performance or reliability of the chart pattern.
I use a frequency distribution any time I want to see which range occurs
most often (or any time I think outliers distort the average). It is just another
perspective, a useful tool in the hands of an investor.
6 Introduction

8 Introduction
Investing Using Chart Formations
I could give a dentist's drill to any person walking by, but I would not let him
or her near my teeth. This book is just like that. Itgivesyou the tools to investsuc
cessfully. It suggests which chart patterns work best and which ones to avoid.
Whether you can make money using them is entirely up to you.
I call this book an encyclopedia because that is how I use it. Whenever I
see a chart pattern forming in a stock I own, or am thinking of buying, I read
the applicable chapter. The information refreshes my memory about identifi
cation quirks, performance, and any tips on how I can get in sooner or more
profitably. Then I search for similar patterns in the same stock (using different
time scales), and if that does not work, I search for similar patterns in stocks in
the same industry. I look at them closely to determine if their secrets are
applicable to the current situation. I try to learn from their mistakes.
At the same time, I am paper trading chart formations in the 250 or so
stocks I follow on a daily basis (relax, a review only takes me an hour). Even
though I consider myself an experienced investor (after nearly 20 years, what
do you think?), the constant paper trading keeps me sharp. It has moved
pulling the trigger (buying or selling a stock) from a conscious effort to a rote
reflex. The constant checking on how the chart pattern is faring forces me to
develop an intuitive feel for the formation, the stock, and the market.
Developing an Investment Style
The question I am asked most often is, how do I develop an investment style? It is
usually not asked like that. Most take a more direct approach: How do I make
money trading stocks? When first asked this question, I stumbled over the
answer. I think it is like showing four people the color blue and asking them to
describe it. One person is color blind so you automatically throw out whatever
he says. One says it is solid blue. Another says it is not blue at all but green,
while the third says it looks like a combination: bluegreen. To each individ
ual, blue looks like blue—just do not try to compare answers.
Developing a trading style is a lot like that. It is an individual endeavor
that has a lot in common with experience. I cannot give you experience; I can
only suggest ways to acquire your own.
If you read a chapter on a chart pattern and buy the first stock showing
the pattern, you will probably be successful. The first trade nearly always works
for the novice, maybe even the second or third one, too. Eventually, though,
someone is going to pull the rug out from under you (who knows, maybe it
occurs on the first trade). You will make an investment in a chart pattern and
the trade will go bad. Maybe you will stumble across a herd of bad trades and
get flattened. You might question your sanity, you might question God, but
one thing is for certain: It is not working!
Developing an Investment Style 9
Most people buy stocks like they buy fruit. They look at it, perhaps sniff
it, and plunk down their money. We are not talking about $1.59 here. We are
talking about thousands of dollars for part ownership in a company.
Ifyou have ever been a board member, you know what I am talking about.
You have a fiduciary responsibility to the people who elected or appointed you
to that position. Not only should you study the material handed to you by the
staff, but you have to get out in the field and kick the tires. Do not assume that
what the staff says is always correct or represents the best solution. Question
everything but learn in the process and try to be helpful without being a pest
(I always seem to fall into the pest category). As a shareholder—an owner ofthe
company—should it be any different?
I recently was considering buying a position in a company showing an
upside breakout from a symmetrical triangle. My computer program told me
the company is a member of the machinery industry and further research
revealed that it makes refractory products. I continued doing research on the
company until the message gnawing at me finally sank in. I did not have the
foggiest idea of what a refractory product was. Despite my search for an answer,
I was not getting the sort ofwarm fuzzies I usually get when researching a pos
sible investment. So, I passed it over. I am trading it on paper, sure, but not in
real life. Call it the Peter Lynch Syndrome: Do not invest in anything you can
not understand or explain in a paragraph. Good advice.
Of course, if you blindly invest in chart squiggles and it works for you,
who am I to tell you you are doing it wrong? The fact is, you are not. If you
consistently make money at it, then you have developed an investment style
that fits your personality. Good for you!
My investment style, as you might have guessed, combines fundamental
analysis, technical analysis, emotional analysis, and money management.
Just because I rely on technical analysis does not mean I do not look at the
pricetoearnings, pricetosales, and other more esoteric ratios. Then there
is the emotional element. After going for months without making a single
trade, suddenly a profitable opportunity appears and I will take advantage
of it. Three days later, I will want to trade again. Why? Am I trading
just because it feels good to be finally back in the thick of things? Am I trad
ing just because the single woman living nearby does not know I exist
and I am acting out my frustrations or trying to impress her with the size
ofmy wallet? That is where paper trading comes in handy. I can experiment on
new techniques without getting burned. If I do the simulation accurately
enough, my subconscious will not know the difference and I will learn a lot
in the process.
Once I come to terms with any emotional issues, I look at money man
agement. How much can I realistically expect to make and how much can I
lose? What is the proper lot size to take? When should I add to my position?
How long will it take for the stock to reach my target and should I invest in a
less promising but quicker candidate?

10 Introduction
Investing using chart formations is an exercise in probability. If you play
the numbers long enough, you will win out. Sure, some of your investments
will fail, and you must learn to cut your losses before they get out of hand. But
the winners should serve you well, providing you let them ride. Just do not
make the mistake of watching a stock double or triple only to reverse course
and drop back to where it started. Or worse.
Day Traders, Position Traders, BuyandHold Investors
As I was writing this book, I kept asking myselfwhat is the time horizonfor chart
patterns? Are they best for day traders, position traders, or buyandhold
investors? The answer I kept coming up with is: Yes! Chart formations can be
profitable for day traders—those people who are in and out of a trade during a
single day. Many day traders have trading styles that depend on chart forma
tions, support, and resistance. They concentrate on reliable formations that
quickly fulfill their measure rule predictions.
For position traders, those who hold the trade longer than a day but not
forever, chart patterns offer convenient entry and exit points. I put myself in
this category. Ifthe trade goes bad, I am out quickly. Ifit is profitable, I see no
need to cut my profits short. When the gains plateau, or ifthe stock has moved
about all it is going to, I consider moving on. Like the day trader, I try to max
imize turns by buying formations that promise reliable returns and reach the
ultimate high quickly.
For the longterm investor, chart patterns also signal good entry and exit
points. I recently purchased an oil services company knowing that the invest
ment would not make a significant return for 2 or 3 years (I was wrong: It dou
bled in 3 weeks). It is my beliefthat in 3 years' time, the stock will be in the 3Os,
a sixfold increase from its low. It probably will not qualify for a tenbagger, but
it is not small change either. In the short term, the road is going to be rocky
and I have added to my position as the stock has come down. Since I am in it
for the long term, I have an outstanding order to buy more shares. If this stock
goes nowhere, then my analysis ofthe market trends was wrong, and I will have
learned a valuable lesson.
The Sample Trade
The Sample Trade sections that are included in many of the chapters in this
book are fictitious except for one: the trade I made using a symmetrical trian
gle bottom. Each sample trade uses techniques I wanted to illustrate, incorpo
rating fictitious people in sometimes unusual circumstances. Call it poetic
license, but I hope they give you some ideas on how to increase your profits or
to minimize your losses.
If You Like This Book ... 11
If You Like This Book . . .
When I plunk down my hardearned money for a book, I expect to get a good
value. Many times I have complained that I did not learn anything from a
book. At other times, the information is exciting and new, but I cannot use it
because the tools the author presented are either too esoteric or too expensive.
I vowed to give the reader real value in this book. The information is easy
to find, from the alphabetical chapter layout, to the statistical snapshot at the
start of each chapter, to the advice and suggestions all laid out in easy reference
tables. The chapters are replete with pertinent illustrations. However, I fear
that ifyou try to read this book from cover to cover, it surely will put even the
most hardened insomniac to sleep. Use this book as a reference tool. Refer to
it before you make a trade.
If this book saves you money, gives you the courage to pull the trigger
with a little bit more confidence, or makes you a whopping profit, then I will
have done my job.

1
Broadening Bottoms
RESULTS SNAPSHOT
Upside Breakouts
Appearance
Reversal or
consolidation
Failure rate
Average rise
Volume trend
Percentage meeting
predicted price
target
Surprising finding
See also
Price trend is downward, leading to the formation.
Megaphone appearance with higher highs and lower
lows that widen over time. Breakout is upward.
Shortterm (less than 3 months) bullish reversal
2%
25%, with most likely gain less than 10%
Ragged, but usually follows price: rises as prices rise,
falls when prices fall.
59%
Partial rise at the end of the formation predicts a
downside breakout 67% of the time and partial
declines predict an upside breakout 80% of the time.
Broadening Formations, RightAngled and
Ascending; Broadening Formations, RightAngled
and Descending; BroadeningTops; Broadening
Wedges, Ascending; Broadening Wedges,
Descending
Results Snapshot 13
Downside Breakouts
Appearance
Reversal or
consolidation
Failure rate
Average decline
Volume trend
Percentage meeting
predicted price
target
See also
Price trend is downward, leading to the formation.
Megaphone appearance with higher highs and
lowered lows that widen over time. Breakout is
downward.
Shortterm (less than 3 months) bearish consolidation
6%
27%, with most likely decline between 15% and 20%
70%
Broadening Formations, RightAngled and
Ascending; Broadening Formations, RightAngled
and Descending; Broadening Tops; Broadening
Wedges, Ascending; Broadening Wedges,
Descending
When I compiled the statistics for broadening bottoms, I had to double check
the results because they were unusual. Broadening bottoms with downside
breakouts outperform those with upside breakouts. Bullish formations typically
have gains averaging about 40%; broadening bottoms have gains of just 25%.
Bearish formations decline about 20%, on average, but bearish broadening
bottoms show losses of 27%. This information tells me that even though you
can have an upside breakout, the chart pattern is essentially a bearish one.
Prices do not rise all that far before retreating, and when they do break out
downward, the decline is above average in severity.
The most likely gains—computed using a frequency distribution of the
returns—are about what you would expect: 10% for upside breakouts and a rel
atively high 15% to 20% for broadening bottoms with downside breakouts.
The failure rates are also remarkably small: just 2% and 6% for upside
and downside breakouts, respectively. Anything that is under 20% I consider
acceptable.
One surprising finding concerns partial rises and declines, where prices
begin moving across the formation to the opposite side, reverse course, and
stage a breakout. When prices begin moving down from the top and reverse,
80% of the formations stage an upside breakout. For downside breakouts, the
score is a respectable 67% (two out of three show this behavior).

Tour
You may be wondering what differentiates a broadening bottom from a broad
ening top. A broadening bottom has a price trend leading down to the start of
the formation; a broadening top has prices trending up. This differentiation is
an arbitrary designation I made to separate the two formation types. I could
have used their location in the 12month price range (those located in the
upper half are tops, the rest are bottoms). However, this methodology poses a
problem when the formation is near the center ofthe yearly price range: Is it a
top or a bottom?
Using a price trend leading to a formation is no surefire solution either.
If the price trend is nearly horizontal or changes abruptly just before the for
mation starts, then I pretend I am a moving average. Would a 90day moving
average be trending up or down? Once you know the trend, you can then fig
ure out whether you are dealing with a broadening top or bottom.
Some maintain that a broadening bottom does not exist. They simply
lump every broadening pattern into the broadening top category. I decided to
separate the two on the off chance that their performance or behavior differs.
You may want to combine the statistics or do your own research.
Figure 1.1 is an example of a broadening bottom. This particular one is
called a fivepoint reversal because there are five alternating touches, two
minor lows and three minor highs. A fivepoint reversal is also rare: I located
Bane One Corp. (Bank, NYSE, ONE)
Aug 94
Figure 1.1 A broadening bottom formation, specifically a fivepoint reversal, so
called because of the two minor lows (the even numbers) and three minor highs
(the odd numbers).
Identification Guidelines 15
only 5 in the 77 broadening bottoms I examined. The price trend begins mov
ing down in late August and reaches a low 2 days before the formation begins.
Yes, prices do move up for several days, leading to the first touch of the top
trendline, but I still consider the overall price trend to be moving down to the
formation.
This particular chart pattern shows the partial decline I mentioned ear
lier. Prices move down from 26 to 24/2, then reverse course and shoot out the
top. The stock reached a high of 38'/2 just over a year later.
Identification Guidelines
Table 1.1 lists the identification guidelines for broadening bottoms. As men
tioned earlier, a declining price trend precedes a broadening bottom. Even if
prices rise just before the formation begins, ignore it. It is still a bottom. This
arbitrary designation also makes intuitive sense: A bottom should appear at the
end of a downtrend, not when prices are climbing to the moon.
The shape of the formation is distinct. It reminds me of chaos theory
where small disturbances oscillate back and forth, then sometimes grow
unbounded, wreaking havoc. In the stock market, prices reach new highs then
cross over and make new lows. When you draw a trendline across the minor
highs and another connecting the minor lows, the formation looks like a
megaphone.
The two trendlines drawn across the minor highs and lows are important.
The top trendline should slope up; the bottom one should slope down. The
diverging trendlines distinguish the broadening bottom from other types of
Table 1.1
Identification Characteristics of Broadening Bottoms
Characteristic Discussion
Pricetrend The intermediateterm price trend should be downward leading
to the formation. ,
Shape Megaphone shape with higher highs and lower lows.
Trendlines Prices follow two trendlines: The top one slopes up and the
bottom one slopes down.
Touches Should have at least two minor highs and two minor lows, but
not necessarily alternating touches.
Volume Irregular but usually rises as prices rise and recedes as prices fall.
Breakout The breakout can occur in either direction and, in some cases,
prices move horizontally for several months before staging a
definitive breakout.

formations, such as the rightangled broadening formation (which has one
horizontal treiidline) or the broadening wedge (both trendlines slope in the
same direction). So it is important that both trendlines have a slope that is
opposite each other (that is, the top slopes up and the bottom slopes down).
A broadening bottom needs at least two minor highs and two minor lows
to be a valid formation. Anything fewer means you are incorrectly identifying
the formation. What is a minor high or low? A minor high is when prices trend
up, then drop back down, leaving a clearly defined peak. A minor low is just the
same thing flipped upside down: Prices move lower, then head back up leaving
a clearly defined valley. Figure 1.1 shows five minor highs or lows, labeled by
numbers. The odd numbers tag the minor highs and the even numbers are the
minor lows. Let me stress that the minor highs and lows need not be alternat
ing, as in Figure 1.1. Just as long as you can count at least two peaks and two
valleys—wherever they may appear—that is fine.
There is nothing magical in the volume trend. I performed linear regres
sion from die start of each formation to the end point (not the breakout point
that is usually a month beyond the end of the formation) and found that vol
ume rises about 58% or 59% (upside and downside breakouts, respectively) of
the time. That is just a little better than a coin toss, certainly not strong enough
to make a definitive statement.
Ifyou look closely at most broadening bottoms, you will find that volume
usually follows price. In Figure 1.1, the price decline between peak 1 and
trough 2 shows a receding volume trend. When prices head up from point 2 to
point 3, so does volume. One thing is certain: Volume is irregular and the ris
ingfalling trend is only a general guideline often broken. When selecting a
broadening bottom, I ignore the volume pattern.
The breakout point is difficult to identify in a broadening formation as it
is developing. In retrospect, it is easier. I look for the place where prices pierce
the up or down trendline or make an extended move. Ifprices pierce the trend
line, then the penetration point becomes the breakout point. Ifprices move up
and follow along the top trendline without piercing it, then I backtrack to the
prior minor high and draw a horizontal line forward in time until prices cross
it. When that happens, that is the breakout point.
Let me give you an example. Consider the broadening bottom shown in
Figure 1.2. The price trend over the preceding month leading to the formation
is downward. The two trendlines outline a widening price pattern as you would
expect from a broadening formation. There are more than two minor highs and
two minor lows pictured, meeting another criterion mentioned in Table 1.1.
Where is the breakout? This formation is particularly easy. Ifyou extend
the top trendline upward, you find that prices rise well above the line, signal
ing an upside breakout. Then it is just a matter of backtracking to the highest
minor high and drawing a horizontal line to determine the actual breakout
price. Point A marks the highest high in die formation.
Focus on Failures 17
Standard Microsystems Corp. (Computers ft Peripherals, NASDAQ, SMSC)
Jul95
Figure 1.2 A breakout from the broadening bottom occurs when prices rise
above the highest high in the formation, shown as point A.
This formation is typical ofbroadening bottoms. The breakout is upward
and occurs at a price of 18. Soon, die stock moves up to 23^, a rise of23% or
nearly die 25% average rise for broadening bottoms with upside breakouts.
Focus on Failures
The good news is that with only three formation failures there is litde to worry
about. The bad news is that with only three failures there is not much to learn.
Figure 1.3 shows one of the three broadening bottom failures. Prices head
down and appear to suffer a deadcat bounce lasting from April to August. I do
not recommend taking a position in any stock that shows a deadcat bounce
regardless of how attractive the formation looks. Obey this recommendation
for 6 months to a year while the stock recovers and management gets its house
in order (or solves the cause ofwhatever is ailing the stock).
In the 3 weeks before the formation appeared, prices were heading higher
in reaction to the deadcat bounce. In June diey moved horizontally from the
formation top for over a month before easing down. It was during this time
that prices rose above the high ofthe formation (see point A).
I do not consider prices to break out above or below a formation until the
closing price moves beyond the formation high or low, which is the case with
pointA. It is not an upside breakout because the close is at 337
/s, well below the

Figure 1.3 This broadening bottom forms as part of the recovery process from a
deadcat bounce. When prices close below the formation low, a downside break
out occurs. Point A shows where prices move above the high but do not close
higher. The formation is a failure because prices do not move down by more than
5% below the breakout point before reversing.
formation high of 34*4. Two days later, it peaks above the high, but the close
is also below the formation high.
However, look what happens when prices begin sinking in midJuly.
They drop below the formation and close even lower. The price needs to drop
below 30%. At its lowest point, it closes at 297
/s. That is just fifty cents below
the low, but it is enough to signal a downside breakout. Within a week ofmov
ing below the formation low, prices shoot to 33 and continue up using a slower
trajectory.
Figure 1.3 represents what I call a 5% failure. Prices break out lower but
fail to continue moving in the breakout direction by more than 5% before
heading back up. The reverse is also true for upside 5% failures: Prices move
up by less than 5% before turning around and tumbling.
Statistics
Table 1.2 shows general statistics, which I separated into two types: upside and
downside breakouts. Since there is a dearth ofbroadening bottoms in my usual
database of 500 stocks over 5 years, I searched the database that I use on a daily
Statistics 19
Table 1.2
General Statistics for Broadening Bottoms
Description Upside Breakout DownsideBreakout
Number of formations: 35
in 500 stocks from 1991 to
1996; 42 in about 300 stocks
from 1996; to 1999
Reversal or consolidation
Failure rate
Average rise/decline of
successful formations
Most likely rise/decline
Of those succeeding, number
meeting or exceeding price
target (measure rule)
Average formation length
Partial rise but ended down
Partial decline but ended up
Percentage of time there
was a trend reversal within
3 months
45
45 reversals
1 /45 or 2%
25%
10%
26 or 59%
2 months (61 days)
12/18 or 67%
16/20 or 80%
48%
32
32 consolidations
2/32 or 6%
27%
15% to 20%
21 or 70%
2 months (57 days)
12/18 or 67%
16/20 or 80%
52%
basis. About 3 years long, it covers approximately 300 stocks and picks up from
where the other database ends, so there is no overlap in dates. It is noteworthy
that I uncovered more formations (42 versus 35) in the most recent 3 years
(900 years of daily price data) than in the prior 5 years (2,500 years of daily
price data).
Broadening bottoms with upside breakouts act as reversals of the pre
vailing trend, whereas those with downside breakouts act as consolidations.
This observation makes sense when coupled with the provision that the trend
leading to the formation must be downward. Under those circumstances,
an upward breakout will be a reversal, whereas a downward breakout is a
consolidation.
The failure rate is very low: 2% and 6% for the two breakout types. I
think the reason for this occurrence is that at its widest point, a broadening for
mation represents a strong trend as prices move from one side ofthe formation
to the other. Once this momentum gets under way, it seems likely to continue,
and not falter after a breakout occurs (leading to a 5% failure). As the saying
goes, a trend in motion tends to remain in motion.
The average rise and decline is 25% for upside breakouts and 27% for
downside ones. Both statistics are unusual. The upside breakout is below the
usual 40% or so for wellbehaved bullish formations. The 27% decline is well

20 Broadening Bottoms I '
above the usual 20% norm for bearish formations. The numbers suggest die
broadening bottom is predominantly a bearish formation, resulting in short
upside gains or extended downside losses.
The most likely rise or decline is about average: 10% for upside breakouts
and a stronger than normal 15% to 20% for downside breakouts. Figure 1.4
shows the results from a frequency distribution of die gains and losses. I call
the tallest columns the most likely gain or loss because they have the highest fre
quency (the most formations in a given percentage range). It is the return an
investor is likely to experience most often.
The figure looks quite irregular with returns forming two humps: one
from 10% to 25% and a second from 35% on upward. A small sample size is
probably the reason, with just 45 or 32 formations to divide between 10 cate
gories, so view the results with skepticism.
I explain the measure rule in the Trading Tactics section, but it involves
computing the height of the formation and adding or subtracting it from the
breakout price. The result gives the target price to which the stock will move.
For upside breakouts, prices reach the target just 59% of the time, whereas
downside breakouts score much better, at 70%. Still, the values are a bit shy of
the 80% benchmark I consider a minimum for reliable formations.
The average formation length is remarkably close for both types ofbreak
outs: about 2 months. Since this is an average, the actual lengths can range all
over the place. If you can state one thing about broadening formations, it is
that they take time to form. The oscillating movements from one side of the
formation to the other do not happen overnight.
Figure 1 .4 Frequency distribution of returns for broadening bottoms.
Statistics 21
An interesting anomaly I noticed when scanning broadening formations
is die partial rise or decline. Figure 1.1 shows a good example of what I am
talking about. Prices begin to move down across the formation to the opposite
side, turn around, and break out. When a partial rise occurs, a downside break
out follows 67% of the time—that is two out of three. A partial decline does
even better: 80% ofthe formations showing a partial decline break out upward.
So, ifyou see a partial rise or decline in a broadening bottom, you might want
to jump in and trade the stock with the expectation that a breakout will follow.
Some have said that when a broadening formation has an upside breakout
(usually a bullish scenario), then the ultimate high is not far away. Soon, they
maintain, the stock will reach its high and make an extended downward move.
I tested this premise and found that it basically is not true. Only 48% of the
formations have die ultimate high (a trend reversal) appear widiin 3 months of
the breakout point. I think anything more than 3 months places a company
into another fiscal quarter, and a different dynamic is probably responsible for
any downturn.
Table 1.3 shows breakout statistics for broadening bottoms. There were
45 formations with upside breakouts and 32 with downside breakouts. Once a
formation ends, the actual breakout occurs about a month later. As explained,
I consider a breakout to occur when prices pierce one of the trendlines (and
closes outside the trendline) or continue moving along a trendline for an inor
dinate amount oftime. Once a breakout occurs, it takes 4 months for those for
mations with upside breakouts to reach their ultimate high and 3 months for
chart patterns widi downside breakouts to reach their ultimate low.
Where in the yearly price range do breakouts occur? To find die answer,
I divided the yearly price range into thirds and sorted each formation into the
appropriate range. Those formations with upside breakouts appear most often
in the center or lower third ofdie price range, suggesting that die chart pattern
itself resides rather low in the price range. This occurrence makes sense
Table 1.3
Breakout Statistics for Broadening Bottoms
Description
Number of breakouts
Formation end to breakout
For successful formations,
days to ultimate high/low
Percentage of breakouts
occurring near 12month
low (L), center (C), or high (H)
Percentage gain/loss for each
12month lookback period
Upside Breakouts
45 or 58%
28 days
4 months (123 days)
L40%, C42%, HI 9%
L26%, C23%, H26%
Downside Breakouts
32 or 42%
36 days
3 months (95 days)
L78%, C15%, H7%
L30%, C28%, H22%

because a requirement of a broadening bottom is that prices trend downward
to the start of the formation. Since the breakout point is at the top of the for
mation, it sometimes pokes the breakout point into the center third of the
yearly price range. For downside breakouts, the breakout point is usually in the
lowest third of the yearly price range.
Mapping the performance ofdie chart patterns over the same price range
shows that those with upside breakouts split evenly: The lowest and highest
thirds of the yearly price range score best with gains averaging 26%. For
downside breakouts, those in the lowest third of the yearly price range have the
highest average decline: 30%. This percentage tapers offwith those formations
in the highest third of the range showing the worst returns at 22%.
Table 1.4 shows the last group of statistics—all the formations broken
down by alternating touches. This is not the same as the requirement of hav
ing two minor highs and two minor lows in each broadening bottom. Two
minor highs, for example, can occur without prices declining fully to the oppo
site side. I counted the number of alternating touches for each formation to see
if there is a pattern to the number of touches and the breakout point. As you
can see in Table 1.4, formations with four alternating touches break out most
often, 13 each. However, formations with upside breakouts commonly range
from three to five alternating touches per chart pattern.
What does this mean? Ifyou find a broadening bottom in a stock you own
or are considering purchasing, you might try counting the number of alternat
ing touches. If the stock has four touches, then it is more likely to break out on
the next crossing of the formation. The table suggests that the likelihood of an
upside break out after four alternating touches is 52%, and a downside break
out is 57% (the side of the first touch gives the breakout direction: If the first
touch is on the top, then an upside breakout is likely after four touches).
Table 1.4
Frequency Distribution of Successful Formations by Number of Alternating
Touches with Cumulative Percentage of Total
Number of
Alternating Touches
3
4
5
6
7
8
9
Number with
Upside Breakouts
10: 23%
13: 52%
10: 75%
6: 89%
3: 95%
1: 99%
1 : 1 00%
Number with
Downside Breakouts
4: 13%
1 3: 57%
8: 83%
2: 90%
3: 100%
0: 100%
0: 1 00%
Trading Tactics 23
To count the number of touches, refer to Figure 1.1 where I labeled the
touches. I do not consider the initial entry point to be a touch. In the figure,
the entry point is on the bottom of the formation at about 23'/s. Prices move
to the top and reverse course. That is the first touch. Then prices descend to
the opposite side, making another touch. You can see on the fourth touch that
prices touch the bottom trendline twice. Between the two touches is a minor
high, but since prices did not touch or near the top trendline, no additional
touch scores.
Trading Tactics
Table 1.5 shows trading tactics for broadening bottoms. The first tactic is to
determine how much money you are likely to make in a trade. The measure
rule helps with the prediction. Subtract the highest high from the lowest low
in the formation to give you the formation height. Then add the value to the
highest high to get the target price for upside breakouts and subtract the height
from die lowest low for downside breakouts.
Figure 1.5 makes die computation clear. Point A shows die highest high
in the chart pattern at 14H. The lowest low is point B at 12. The formation
Table 1.5
Trading Tactics for Broadening Bottoms
Trading Tactic Explanation
Measure rule
Co long at the low
Long stop
Go short at the high
Short stop
Move stops
Other
Compute the difference between the highest high and the
lowest low in the formation. Add or subtract this value from
the most recent minor high or low, respectively. The result is
the target price for upside and downside breakouts.
Once recognizing a broadening formation, buy after the stock
makes its turn at the low.
Place a stoploss order below the minor low to protect
against a trend reversal.
Sell short after prices start heading down at the top.
Place a stop '4 above the minor high to protect against an
adverse breakout. Cover the short when it turns at the
trendline and starts moving up. For a downside breakout,
cover as it nears the target price or any support level.
Raise or lower the stop to the next closest minor low or high
once prices pass the prior minor high (for long trades) or low
(for short sales).
If a broadening bottom shows a partial decline or rise, trade
accordingly (on a partial decline, go long; on a partial rise,
short the stock).

Acuson Corp (Medical Supplies, NYSE, ACN)
Jan 93 Feb Mar Apr May |un |ul
Figure 1.5 A broadening bottom with five alternating touches. Expect a down
ward breakout because a partial rise appears.
height is the difference between the two or 2 Vs. Add the value to the high to
arrive at the upside price target. This turns out to be 16'4. I compute the
downside target by subtracting the height from the lowest low (that is, 12 2l
/s
or 97
/s). You can see in Figure 1.5 that the price never quite reaches the down
side price target. For downside breakouts, prices fulfill the measure rule 70%
of the time but only 59% ofthe time for upside breakouts. Both values are shy
of die 80% that I like to see for reliable formations.
Once you have uncovered a broadening bottom, with two minor highs
and two minor lows, you can think about trading it. When the price bounces
off the lower trendline, buy the stock. Sell when it turns down. The downturn
may occur as a partial rise partway across the formation or prices may cross
completely to the other side, touch the top trendline, and head down. Remem
ber, the formation may stage an upside breakout, so do not sell too soon and
cut your profits short.
In a rising price trend, place a stoploss order l
/s below the minor low.
Should the stock reverse and head down, you will be taken out with a small
loss. As the stock rises to the opposite side of the formation, move your stop
upward tol
/s below the prior minor low. The minor low may act as a resistance
point, so you will be giving the stock every opportunity to bounce off the resis
tance level before being cashed out.
The trading tactic for downside breakouts is the same. When prices touch
the top trendline and begin moving down, short the stock. Place a stoploss
order VB above the highest high in the formation, then pray that prices decline.
Sample Trade 25
Ifluck is on your side and die stock heads down, move your stop lower. Use the
prior minor high—place the stop H above it.
If the stock makes a partial rise or decline, consider acting on it. This is a
reliable breakout signal. For partial rises, die signal is right 67% of the time,
and for partial declines, it works 80% of the time. Take advantage of it but
make sure you place a stoploss order Vg beyond the nearest resistance point in
case the trade goes bad.
Once prices break out and leave die broadening pattern, consider selling
if the price nears the target. There is no guarantee that die price will hit or
exceed die target, so be ready to complete die trade, especially if there is a
resistance level between the current price and the target. The stock may reach
the resistance point and turn around.
Sample Trade
Susan likes to think ofherself as die brains in die family. While her husband is
suffering in foul weather as a carpenter, she is toiling away at her keyboard, a
slave to her computer masters. She is an active position trader who is not afraid
to short a stock, given good profit potential and an especially weak fundamen
tal or technical situation. It is a stressful life but making money often is.
When she spotted die broadening bottom shown in Figure 1.5, she began
her analysis. The stock reached a high of 3 7% in early November 1991 and has
been heading down ever since. Now, widi die formation trading at 14, she
wondered how much downside remained. She drew die two trendline bound
aries and counted die number of alternating touches (in Figure 1.5, diree are
labeled as numbers and Point A is die fourth touch).
Since most broadening formations tend to break out after four alternat
ing touches and since the price was near die top of die formation heading
down, she guessed diat die stock would break out downward on the next cross
ing. So she sold the stock short and received a fill at 137
/s. It was a gamble, sure,
but one she was comfortable making. In any case, she immediately placed a
stop at 1454, or H above the high at point A.
Susan was overjoyed to see the stock plummet 2 days later and race across
to die other side of die formation, touching the bottom trendline at point B.
Usually, her trades are not that easy. She decided to protect her profit and low
ered the stop to the nearest minor high, shown as point C, at 13% or !
/g above
the high. Then she waited.
The stock bounced off die lower trendline instead of busting through as
she hoped. She decided to be patient and see what die stock did next. Widi her
stoploss order in place at the breakeven price, she felt protected and com
fortable in letting the trade ride.
The stock bounced off die 12'/s support level and did a partial rise before
it met resistance and headed back down. Two days after cresting, she made die

determination that on the next touch, the stock would pierce the lower trend
line and continue down. She doubled her stake by selling more stock short at
123
4. She was wrong. The stock continued down 1 more day before moving up
again. Susan adjusted her stoploss order to include the additional shares, but
kept it at the same price level (13%). Again she waited. The stock slowly
climbed and reached a minor high of 13'/8 before heading down again. This
time the decline was swift enough to punch through the resistance zone at the
lower trendline.
When the stock descended below point B, Susan lowered her stoploss
order to V« above that point or I2l
/s. Then she looked at the measure rule for
the price target. She calculated a target of97
/8 and wondered ifthe stock would
really reach that point. To be safe, she decided to cash out if the stock reached
lO'/s, or '/e above a common support price of 10 (a whole number typically
shows support).
When the stock plunged to 10% on high volume, she wondered ifshe was
looking at a oneday reversal chart pattern. With those formations, it is diffi
cult to be sure if prices would reverse or not. She decided to hold on to her
original target.
Two days later, prices zoomed upward and her stop closed out the trade
at 12'/s. She did not make much money (about 9% with a hold time ofjust over
a month), but she gained experience and a few pennies to put in the bank.
2
Broadening Formations,
RightAngled and
Ascending
RESULTS SNAPSHOT
Appearance
Failure rate
Failure rate if waited
for breakout
Average decline
Volume trend
Fullbacks
Percentage meeting
predicted price target
Horizontal bottom with higher highs following an
upsloping trendline
Shortterm (up to 3 months) bearish reversal
34%
9%
18%, but most likely decline is about 10%
Irregular
72%
43 %; using half of formation height gives 91%
success ratin?
Before I began studying this formation, I assumed prices would climb away
from it, simply because the word ascending is in the title. However, that is not
how the formation performs. It is a bearish reversal of the shortterm price
trend. This is not a novel finding, as others have discussed the bearish behav
ior of ascending broadening formations. The word ascending in the title refers
27

28 Broadening Formations, RightAngled and Ascen_..ig
to the minor highs that rise over time. The base of this formation is flat but the
tops widen out, generally following an upsloping trendline.
There are a few surprises highlighted by the Results Snapshot. The fail
ure rate falls from 34% to 9% if you wait until after a downside breakout
before buying the stock. An improvement is not unusual but such a large one
is. The large gain is because I ignore all upside breakouts, and only a few down
side breakouts fail, leaving a small failure rate.
The other interesting statistic is the number of pullbacks to the formation
base. This number is due, in part, to the messy looking breakout that seems to
be quite common with this reversal. After a breakout, prices move horizontally
and bounce around a bit before continuing down. The decline is sometimes
over quickly as a 10% to 20% decline is easy to erase, fostering a 72% pullback
rate.
Tour
Rightangled ascending broadeningformations: What does the name mean? Right
angle implies that it is a member of the triangle family. A horizontal base with
an upsloping hypotenuse forms a right triangle. The third side drops down
from the hypotenuse to the base and intersects it at a 90 degree angle, forming
the socalled right angle. Ascending means that the hypotenuse ascends over
time as contrasted with descending broadening formations. Broadeningforma
tion means that prices make higher highs. Ascending and descending triangles,
in contrast, have narrowing price movements.
Figure 2.1 puts the formation in perspective. There are two formations
shown in the chart. The first one is somewhat illformed but better perform
ing than the second. Both formations have a base outlined by a horizontal
trendline connecting the minor lows. The upsloping trendline skirts the tops
of the minor highs. The result is a triangleappearing formation with prices
that broaden out, but do not let the ascending price pattern fool you. This
formation is bearish: Prices plummet through the base of the formation most
of the time.
Why do rightangled ascending broadening formations form? Consider
Figure 2.2. The rise began in midDecember 1991 on volume that was higher
than anything seen in almost 2 months. By late February, the stock had reached
a new high and was rounding over after meeting selling resistance at 14. The
stock returned to the 12'A level where it found support. At that point, it
paused for about 2 weeks and established the base on which a horizontal trend
line appears.
The reason for the horizontal trendline is one of perceived value. As the
stock approached the $12 level, more investors and institutional holders pur
chased the stock. The desire to own the stock at what they believed a good
Edwards, A. C. Inc. (Securities Brokerage, NYSE, ACE)
Apr 92 May 'un
)ul A
"9 Sep Oct Nov Dec
Figure 2.1 Two rightangled ascending broadening formations bounded by a
horizontal base and upsloping trendline. Prices decline after a downside breakout.
Baker). Inc (Shoe, NASDAQ, JBAK)
IS
|an92
Figure 2.2 A pullback to the base of the formation. Pullbacks occur often in
ascending broadening formations.
29

30 Broadening Formations, RightAngled and Ascen
value outweighed the reluctance of sellers to part with their shares. The
demand halted the decline in the stock and eventually sent it skyward again.
This happened in midApril as volume spiked along with the price. The enthu
siasm caused the stock to reach a new high. Momentum was high enough so
that the next day, prices rose even further before closing lower. With the sec
ond peak, a tentative trendline drawn along the tops of the formation sloped
upward and gave character to the broadening formation.
The stock moved rapidly back down even as volume increased. This
decline stopped before it reached the lower trendline, signaling continued
enthusiasm. Prices pushed higher and reached a new high, this one at 15l
/2 on
May 6. The upsloping trendline resistance area repelled any further advance.
The stock simply did not have enough upward momentum to push through the
selling pressure at the new level.
The next day volume dried up, but there was enough momentum remain
ing for another try at the summit. When the attempt failed, the smart money
headed back for base camp and volume receded even further. As prices col
lapsed, other investors joined in the retreat and volume moved up. In less than
2 weeks, prices were back at the lower trendline.
Another feeble attempt at a new high floundered on unremarkable vol
ume. The stock moved horizontally and stalled out—a partial rise that often
spells trouble for a stock. On June 4, prices dropped on high volume and
returned to the horizontal trendline. The stock paused there for just over a
week before moving down and punching through the support level at 12'/4.
A pullback is quite common for ascending broadening formations, so it is
no surprise that the decline quickly faded. After a rapid 13% retreat, the stock
turned around and pulled back to the base of the formation. Although it is not
shown in Figure 2.2, the stock continued moving up until it began forming
another ascending broadening formation in late October with a base at 16'/2.
The ascending broadening formation represents the desire of investors
and traders to own the stock at a fixed price, in this case about 12 '/4. Their buy
ing enthusiasm pushes prices higher until mounting selling pressure causes a
halt to the rise and sends the stock tumbling. With each attempt, fewer people
are left willing to sell their shares until they receive an even higher price, so a
broadening range ofprices appears at the top. Eventually, the buying enthusi
asm at the base of the formation collapses and removes the support for the
stock. When that happens, the stock punches through the support level and
declines. It continues moving down until reaching a point where other
investors perceive significant value and buy the stock.
What are die characteristics ofan ascendingbroadening formation? To answer
the question, peruse the selection guidelines outlined inTable 2.1. While con-
Table 2.1
Identification Characteristics of RightAngled Ascending Broadening Formations
Shape
Horizontal bottom support line
Upsloping top trendline
Volume
Premature breakouts
Price action before breakout
Downside breakout
Support and resistance
Looks like a megaphone with the base of the
formation horizontal and bounded on the top by
an upsloping trendline.
A horizontal, or nearly so, trendline that connects
the minor lows. Must have at least two distinct
minor lows before drawing a trendline.
An upsloping trendline bounds the expanding
price series on the top. Must have at least two
minor highs to create a trendline.
Irregular with no consistent pattern.
Very rare. A close below the horizontal trendline is
most likely a genuine breakout.
Prices sometimes move horizontally for many
months before moving outside the formation high
or low. After a breakout, expect a pullback to the
base of the formation.
Prices drop below the horizontal trendline usually
accompanied by a surge in volume.
Follows the two trendlines into the future.
sidering the table, look at Figure 2.3, an ascending broadening formation on a
weekly scale. The overall shape of the formation looks like a megaphone with
one side horizontal. The bottom of the formation follows a horizontal trend
line, while an upsloping trendline bounds the top side. The top trendline
touches at least two minor highs. A minor high refers to a distinct peak that is
clearly visible and well separated from other peaks on the chart. The horizon
tal trendline also shows two minor low touches as prices descend to the trend
line. The phrase minor low refers to valleys separated and distinct from other
troughs. The various touch points help define the boundary of the formation.
As you can see from Figure 2.3 and the preceding charts, the volume pat
tern is irregular. However, in a majority of cases, volume picks up after the
breakout. Although this formation fails to descend, you can still see the volume
rise in early 1993.
I define premature breakouts to be prices that close outside the formation
boundary but return before the formation ends. Premature breakouts for this
formation are rare enough that they should not be of concern.
In some ascending broadening formations, prices make higher highs and
form a solid, horizontal base at the start but then move sideways for many

32 Broadening Formations, RightAngled and Ascending
Parker Drilling Co. (Oilfield Svcs/Equipment, NYSE, PKD)
8
6
92 F M A M | | A S O N D 9 3 F M A M ) J A S O N D 9 4 F M A M J | A S O N D95 FM A M J
Figure 2.3 Support and resistance areas on a weekly time scale. They appear
along the trendline axis and can extend far into the future, as in this case.
months. Eventually, prices rise above the formation top or slide through the
bottom trendline and stage a breakout. Once a breakout occurs, typically
downward, expect a pullback. A pullback is when prices move lower, then turn
around and touch the bottom trendline. Prices may continue moving up but
they usually bounce offthe trendline and continue back down. A pullback gives
investors another opportunity to short the stock or add to their short position.
Before investing, however, make sure the pullback is complete and prices are
declining once again.
I chose Figure 2.3 because it shows the two common areas ofsupport and
resistance. These areas follow the trendlines. Along the base of the formation
projected into the future, the support area repels the decline over 2 years after
the formation ends. The rising trendline tells a similar tale; it repels prices
three times nearly a year later. The implications of this observation can be pro
found. If you own a stock and it is breaking out to new highs, it would be nice
to predict how high prices will rise. One way to do that is to search for forma
tions such as this one. Many times, extending the trendline into the future will
predict areas of support and, in this case, resistance.
Although the trendline did not predict the absolute high, it did suggest
when prices would stall. The resistance area turned out to be a good opportu
nity to sell the stock.
Focus on Failures
What can we learn from a review of the failures of this formation? Figure 2.4
shows two broadening formations, the one on the left fails to descend but the
one on the right makes up for it. The figure makes one lesson clear: Always
wait for a confirmed breakout before taking a position in a stock; that is, wait
for prices to fall below the lower trendline before selling your long position
or selling short. Even though most ascending broadening formations break out
downward, the failure rate is too high to hazard an investment before knowing
the outcome. Had you sold the stock short during the first formation, your
position would not have made money for almost half a year. Look back at
Figure 2.3. A short position in the stock at the low would have lost money
for years.
Selling a stock prematurely is just as bad. If you held a long position in
the stock shown in Figure 2.4 but sold it duringJune, you would have regret
ted your trade until December when the footwear company slipped. Had you
waited for a downside breakout, you would have remained in the stock as it
ascended. Once the second broadening formation took shape, a sale after
prices pierced the horizontal trendline would have gotten you out at a better
price.
Figure 2.4 Two broadening formations. The formation on the left fails to descend
below the lower trendline. You should wait for the breakout before investing in
ascending broadening formations.

34 Broadening Formations, RightAngled and Asce
Statistics
Table 2.2 shows general statistics.Just like other broadening formations in this
book, I did not feel comfortable basing the statistics on my 5year database
alone, so I incorporated my more recent database for 35 additional forma
tions. This gives a total of 216 formations, making ascending broadening for
mations one of a rare breed. An examination of the formation reveals that 81
are consolidations of the prevailing trend, but the vast majority are reversals,
with 135 falling into that category.
I measure the failure rate in two ways. Since I expect a downward break
out, I counted the number of formations in which that is not the case. There
are 74, giving a failure rate of 34%. What if the investor waits for a downside
breakout? This is called breakout confirmation. It lowers the failure rate to just
9%, well below the 20% maximum I consider reliable formations to possess.
That is how I suggest you trade this formation: Wait for a confirmed downside
breakout before selling short.
Table 2.2
General Statistics for RightAngled Ascending Broadening Formations
Description Statistic
Number of formations in 500 stocks from
1991 to 1996
1996 to 1998
Failure rate
Failure rate if waited for downside
breakout
Average decline of successful formations
Most likely decline
Average rise for failed formations
Most likely rise for failed formations
Of those succeeding, number meeting
or exceeding price target (measure rule)
Use measure rule based on half height
Days to ultimate low
181
35
81 consolidations, 135 reversals
74/216 or 34%
13/151 or 9%
18%
10%
32%
20%
60 or 43%
125 or 91%
3 months (86 days)
3 months (81 days)
Note: Only two out of three ascending broadening formations work as expected and the
most likely decline is meager at 10%.
Statistics 35
What is the average decline for formations with downside breakouts?
The average is 18%, but the most likely decline is less than 10%. I measure this
by sorting the percentage losses into 10 bins and counting the results. The
resulting frequency distribution reveals that the most likely loss is narrower
than the average, due to a few large declines and several in the 15% to 20%
range. Together, they skew the average upward.
I use a similar method for rises from failed formations. A failed formation
is one with an upside breakout or downside breakout that fails to continue
moving down by more than 5%. The average gain is 32%, with the most likely
rise being about 20%. Over a quarter of the formations with upside breakouts
(27%) rise over 50%.
The measure rale, which predicts a target price, is disappointing for
ascending broadening formations. With only 43% of the formations meeting
or exceeding the target, I decided to compute a new measure rule that gives a
higher success rate. I computed the formation height by subtracting the low
est low from the highest high and then dividing by two. The target price is the
height subtracted from the breakout price. Dividing the height by two is the
only change in the formula, and it results in 91% of the formations meeting
their price targets.
The average formation length is just under 3 months, long enough to be
visible on weekly charts. I also computed the average duration from the end of
the formation to the ultimate low. This turns out to be about 3 months.
Table 2.3 shows statistics related to the breakout. There are only 13 for
mations that break out downward but fail to continue moving down by more
than 5% (the socalled 5% failure). This statistic coupled with only three for
mations breaking out upward and then moving lower suggests that once the
formation breaks out, it will likely continue in the breakout direction. For
investors, this is worth knowing. Simply trade with the trend.
Almost a quarter (24%) of the formations break out upward, 6% have
horizontal breakouts, and the remainder are downside ones (70%). I define a
breakout either as prices closing below the lower trendline and moving down
or as rising above the highest high in the formation and continuing up. Often
prices just meander between the two points for several months before finally
staging a breakout.
Throwbacks are prices that break out upward and return to the top trend
line. They occur 44% of the time for those formations with upside breakouts.
The average time to complete a throwback is less than 2 weeks (11 days). That
is die time it takes to flip around and touch the top trendline. I exclude any
throwback taking longer than 30 days. If it takes over a month for prices to
return, I consider it normal price action and not due to a throwback.
Fullbacks are more prevalent. A pullback is when prices break out down
ward and quickly return to the formation base. Seventytwo percent ofthe for
mations with downside breakouts experience pullbacks. The average time to

36 Broadening Formations, RightAngled and Ascend
Table 2.3
Breakout Statistics for RightAngled Ascending Broadening Formations
Description
Downside breakout but failure
Upside breakout but failure
Upside breakout
Horizontal breakout
Downside breakout
Throwbacks
Average time to throwback completion
Fullbacks
Average time to pullback completion
For successful formations, days to ultimate low
For failed formations, days to ultimate high
Percentage of downside breakouts occurring near
the 12month price low (L), center (C), or high (H)
Percentage loss for each 12month lookback period
Volume for breakout day and next 5 days
compared with day before breakout:
Percentage of successful breakouts occurring on
high (H) or low (L) volume
Percentage of failed breakouts occurring on
Statistic
1 3 or 9%
3 or 6%
52 or 24%
1 3 or 6%
1 51 or 70%
23/52 or 44%
11 days
109/151 or 72%
12 days
2 months (69 days)
3 months (96 days)
L22%,C42%, H36%
L17%, C18%, HI 8%
157%, 170%, 122%,
107%, 99%, 102%
H74%, L26%
H80%, L20%
Note: Downside breakouts experience pullbacks 72% of the time.
complete a pullback to the formation base is 12 days. Again, I remove any pull
back over 30 days.
Ascending broadening formations reach their ultimate low or high
quickly, in about 2 or 3 months, respectively. The abruptness of the decline is
due in part because these formations do not decline very far (the most likely
decline is just 10%), so it takes less time to reach the ultimate low.
Most ascending broadening formations occur near the middle of the 12
month price range, as measured from the base of the formation. The largest
declines split evenly at either 17% or 18% throughout the various yearly price
ranges. In essence, it does not matter where in the yearly price range the for
mation occurs; the performance is the same (unlike some other formations
that show definite trends).
Although volume appears irregular throughout the formation, I did
examine the volume surrounding the breakout day. When compared with the
day before the downside breakout, the day after the breakout typically shows
Trading Tactics 37
the largest volume. It measures 70% above the benchmark but recedes as the
week wears on. This pattern is not unusual as investors seem to sell once they
realize a confirmed breakout is underway.
I next wanted to know if there is a relationship between high and low
breakout volume and the success or failure of the formation. In both upside and
downside breakouts, high volume is present. Thus, breakout volume, by itself,
is not a key to the success or failure of a particular formation. Put another way,'
just because you see a lowvolume downside breakout is no reason to suspect
prices will soon recover.
Trading Tactics
Table 2.4 lists trading tactics. The measure rule predicts the price to which the
stock will decline. Compute the difference between the highest high and the
horizontal trendline in the formation. Subtract this value from the value of the
horizontal trendline, and the result is the target price. The target should serve
as a minimum price move to expect, but with ascending broadening forma
tions, prices usually miss the target (only 43% of the time is the target met).
For a more conservative approach, try calculating the formation height
and dividing by 2, then subtract the value from the horizontal trendline. Prices
reach the nearer target almost all the time (91%). The closer target value also
serves as a wakeup call indicating that the formation is probably not worth
trading—at least on the downside.
Table 2.4
Trading Tactics for RightAngled Ascending Broadening Formations
Measure rule
Waitfor confirmation
Buy upside breakouts
Ignore downside breakouts
Compute the formation height from highest high to
the horizontal trendline. Subtract the height from the
value of the horizontal trendline and the result is the
target price. More accurate targets use a formation
height divided by 2.
Since this formation has a comparatively high failure
rate (34%), you should always wait for the breakout
to drop (close) below the horizontal trendline or
above the upsloping trendline.
Once a breakout occurs, prices continue in the
direction of the breakout. Buy an upside breakout
and expect a 20% gain.
Most breakouts occur downward but the resulting
loss is about 10%. Such a small decline is usually not
worth the risk of a short sale.
Note: The best approach is to buy after an upside breakout.

38 Broadening Formations, RightAngled and
Figure 2.5 Ascending broadening formation. Predicted price targets using half
and full formation heights. A broadening top formation appears in late October.
Figure 2.5 makes the measure rule clear. The height of the formation is
the difference between the highest high (34H) and the trendline price (29^4), or
47
/s. Subtract the result from the trendline price, giving a target price of 243
/s.
Since prices only reach the target 43 % of the time, I show a second one. The
nearer target uses half the formation height, or 2.44, to give a price target
of 26.81. Prices reach the closer target 91% of the time after a downside
breakout.
With an overall failure rate of 34%, there is a high likelihood of an
adverse breakout from this formation. Therefore, an investor should always
wait for a breakout before making a trade.
Although die formation usually breaks out downward, you can try buying
the stock on an upside breakout. This approach allows you to go with the usual
trend in prices (up), and the most likely gain is about 20%. Again, be sure to
wait for the upside breakout (when prices close above the upsloping trendline)
since only one in three breakouts is upward.
I do not suggest trying to capitalize on a downward breakout by shorting
the stock. Although the likelihood of a decline is good, the most likely decline
is only 10%, hardly enough to warrant the extra risk of a short sale. If you
already own the stock and do not want to experience a 10% or perhaps larger
decline, then you can either wait for a confirmed downward breakout (wait for
prices to close below the lower trendline) or sell the stock as it nears or touches
the upper trendline and begins heading down.
Sample Trade 39
Sample Trade
Palmer is a wiry sort of guy, one who acts as if he has swallowed too much caf
feine. I am sure you have met the type. Faced with the situation shown in Fig
ure 2.5, he took swift, decisive action. At point A, where the stock touched the
top trendline, he quickly sold it short and received a fill at 333
/8. He placed a
stop at 34 in case the trade went against him. Then he waited.
It did not take long for the stock to cross the formation and reach the hor
izontal trendline. Unfortunately, Palmer did not use an order to automatically
cover his short at 293
/s (the value ofthe trendline). So when prices bounced off
the low, he covered his short the following day, shown as point B, at 30'/2.
Immediately, he went long and bought the stock at the same price.
Palmer placed a stoploss order just below the horizontal trendline, at
291
/4, just in case. Then he extended the top trendline but worried that the
stock might not reach its predicted high. He opted to put a target price at !/8
below the old high at point A. In less than a week, the stock reached his target
and sold at 33'/2 (point C). Since the stock was still showing an upward bias, he
laid back for a bit and waited for the trend to reverse. Three days later he sold
the stock short again at 33. This time, he put a sell order at '/s above the lower
trendline at 29'/2. The trade went against him. It rose to 34 and oscillated up
and down for nearly 3 weeks, never quite reaching his stoploss point of 343
/8.
Then the stock plunged and zipped across the formation. It hit his target price
at point D, and he covered his short.
Sensing a shift in the investment winds, he went long on the stock at the
same price but put a stop loss '/8 below the lower trendline. The following day
prices hit his stop at 29'/4 and he took a small loss. For some unexplained rea
son, Palmer walked away from the stock at this point. Perhaps it was the small
loss he incurred on his last trade, or perhaps he was just running low on caffeine.

3
Broadening Formations,
RightAngled and
Descending
RESULTS SNAPSHOT
Appearance
Reversal or
consolidation
Failure rate ifwaited
for downside
breakouts
for upside
breakouts
Average rise
Average decline
Volume trend
Fullbacks
Throwbacks
Percentage meeting
predicted price
target
Horizontal top with lower lows following a down
sloping trendline
Shortterm (up to 3 months) reversal
3%
19%
27%, but most likely rise is between 20% and 30%
19%, but most likely decline is about 10% to 15%
Irregular
33%
23%
69% for successful downside breakouts, 89% for
successful upside breakouts
40
( Tour 41
Before I begin studying a particular formation, I review the available literature
and determine in which direction the breakout is likely to occur. For the gen
eral class of broadening formations, the breakout pattern is said to be down:
Broadening formations are apparently bearish. After I completed the statistics
for this formation, however, I surprisingly found that more rightangled
descending broadening formations broke out upward than downward. At first
I logged the upside breakouts as failures since I expected the correct breakout
to be down. I was wrong. I reexamined the literature and discovered that
descending broadening formations could break out either way. So I reworked
the statistics and scanned all the formations again to make sure that they agreed
with the new methodology.
The first thing you may notice in the Results Snapshot is that the rever
sal or consolidation line does not say whether the formation is bullish or bear
ish. That is because it depends on the breakout direction. If the breakout
direction is up, which it is 57% ofthe time, the formation is bullish. For down
side breakouts, which occur 37% of the time, the formation is bearish. The
balance have horizontal breakouts or no specific breakout direction.
Only those formations that break out and continue in the opposite direc
tion classify as failures (the socalled 5% failures). I consider failure rates below
20% acceptable, so the rates for both breakout directions score well.
The average rise (27%) or decline (19%) are both below par. Bullish for
mations usually have gains of about 40% and bearish ones decline about 20%,
on average. However, the most likely rise for upside breakouts is reassuring, at
20% to 30%. Since the tabulation uses a frequency distribution of gains, the
numbers imply that the returns distribute evenly. In other words, there are few
large gains to skew the average upward. Unfortunately, with a small sample
size, the numbers are suspect. I believe the most likely rise is probably in the
10% to 15% range, as it is for most other bullish formations.
The measure rule works out better for this formation than for the ascend
ing broadening formation. For downside breakouts, almost 7 in 10 (69%) meet
or exceed their price targets, whereas 89% of upside breakouts reach theirs.
The first statistic is borderline, but the second is reassuring. The meaning is
also clear: Wait for the breakout, then place a trade in the direction of the
breakout.
Tour
What do descending broadening formations look like and why do they form?
Figure 3.1 is an example of the chart pattern. The characteristic flat top and
downsloping bottom are apparent in the figure. These are the two key ingre
dients. Prices at the top of the formation reach the same price level before
declining. Over time, a horizontal trendline can be drawn connecting them.

42 Broadening Formations, RightAngled and DescerT .d
Applebees (Restaurant, NASDAQ, APPB)
Figure 3.1 Descending broadening formation. A horizontal trendline along the
top and a downsloping trendline connecting the minor lows is characteristic of
this chart pattern. The extended, downsloping trendline shows future support and
resistance zones. A oneday reversal appears on November 3 when prices pushed
above the formation top on high volume, but closed at the low for the day.
Along the bottom of the formation, the minor lows touch a downsloping
trendline before prices rebound. Eventually, prices break out of the formation
by either closing above the top trendline or below the bottom one.
In Figure 3.1, the breakout is downward since prices close below the lower
trendline. I require prices to close outside the trendline so that is why the peak
on November 3 does not classify as an upside breakout. On that day, prices close
at 19, the low for the day, and below the top trendline value of about 191
/2.
I mentioned in the Results Snapshot discussion that most descending
broadening formations have upside breakouts. Figure 3.2 shows an example.
The top of the formation is well formed with several minor peaks reaching the
same price level. However, three oneday touches compose the lower trend
line. A trendline touch is a trendline touch regardless ofwhether it is composed
of oneday spikes or many days of consecutive touches.
Figure 3.2 shows a broadening formation with an upside breakout provid
ing a 10% rise in just over 2 weeks. During May 1996, the stock reached 29, for
a 25% gain. The figure also shows one of the few throwbacks to the top of the
formation. This one occurs almost 4 weeks after the breakout. I consider throw
backs or pullbacks that occur later than 30 days to be just normal price action,
not due to the throwback or pullback. This one just makes the cut at 27 days.
Valero Energy Corp. (Petroleum (Integrated), NYSE, VLO)
Jun 95
Figure 3.2 Another descending broadening formation but this time the breakout
is upward. Almost 4 weeks after the breakout, prices throw back to the formation
before ultimately moving higher.
Figure 3.3 Two descending broadening formations. The first formation shows a
trendline rebound resulting from an earlier support zone. The second formation
shows a partial rise that often precedes the ultimate breakout. Shown are two resis
tance areas that parallel the trendlines.
43

44 Broadening Formations, RightAngled and DesceC ;g
Why do these chart patterns form? Look at Figure 3.3. During 1993, the
stock entered the first formation in early April and moved higher on moderate
volume until it reached about 35. There, investors selling the stock matched
buyers eager to own the security and the rise stalled. It traveled sideways until
May 10 when it moved below the prior minor low. As the stock approached the
31 level, it entered a support zone set up by the retracement in midMarch.
The decline stalled and moved sideways for several days. Due to the support
level, many investors believed that the decline was at an end and the stock
would move higher. It did. As volume climbed, the price gapped upward and
quickly soared back to the old high.
The stock ran into selling pressure from institutions and others trying to
sell a block of shares at a fixed price. The available supply halted the advance.
Prices hung on for a few days, moved a bit lower, and paused before beginning
a rapid decline to a new minor low.
As volume climbed, the stock declined until it touched the lower trend
line, a region of support. Suspecting an oversold stock, investors bought and
forced it higher again. When the stock reached the old high, there were fewer
shares available for purchase. Apparently, those investors and institutions who
were trying to get 35 a share for their stock sold most oftheir shares in the pre
ceding months. Soaking up the available supply, the stock gapped upward and
closed above the old high. An upside breakout was at hand.
The stock moved higher but soon formed another descending broaden
ing formation. This one was compact and tight but had bearish implications.
When the stock tried to reach the top trendline but could not, the partial rise
foretold the coming decline. The stock plunged through the lower trending in
late September and continued lower.
Ifyou look at both formations, their stories are nearly the same. There is
a supply of stock available at a fixed price. After exhausting the supply, prices
either rise above the top trendline or decline below the lower one. The deter
mination on which way things will go is not clear. Sometimes the supply over
whelms buyers and the stock declines, unable to recover as it pierces the lower
trendline. At other times, the supply gives out and enthusiastic buyers jump in
and push the price higher.
Are there some guidelines that can assist in identifying descending broadening
formations? Yes, and Table 3.1 outlines them. The shape of the formation
looks like a megaphone with the top held horizontal. Prices climb until they
touch the top trendline, then reverse direction. On the lower edge, prices
decline making a series of lower lows until they touch the lower trendline.
When two minor highs achieve the same, or nearly the same, price level,
you can draw a horizontal trendline connecting them. The same applies to the
Table 3.1
Identification Characteristics of RightAngled Descending Broadening Formations
Shape
Horizontal top resistance line
Downsloping trendline
Volume
Premature breakouts
Breakout
Partial rise or decline
Support and resistance
Looks like a megaphone, tilted down, with the top of
the formation horizontal and bounded on the bottom
by a downsloping trendline.
A horizontal line of resistance joins the tops as a
trendline. Must have at least two distinct touches
(minor highs) before drawing a trendline.
The expanding price series is bounded on the bottom
by a downsloping trendline. Must have at least two
distinct minor lows to create a trendline.
Irregular with no consistent pattern.
Very rare. A close outside the trendline is most likely a
genuine breakout.
Prices can break out in either direction, usually
accompanied by a rise in volume that soon tapers off.
For an established formation, when prices climb
toward the top trendline or decline toward the lower
one but fail to touch it, prices often reverse direction
and break out of the formation.
Follows the two trendlines into the future but is
sporadic.
downsloping trendline: It requires at least two distinct touches before draw
ing the trendline. There is usually ample time to recognize a broadening for
mation, and many times there are more than two touches of each trendline.
There is no consistent volume pattern for this formation. Sometimes vol
ume tapers off, then explodes on the breakout day, just like its triangle cousins.
At other times, volume starts slowly and rises as the breakout nears. Ofthe two
scenarios, the first is slightly more likely than the second, at 53% versus 47%.
Since the numbers are so close, I attach no significance to them.
A partial rise, as shown in Figure 3.3, or a partial decline is often a clue to
the ultimate breakout direction. When prices curl around on a partial rise or
decline and return to the trendline, they usually break out immediately (that is,
without crossing the formation again). We will see in the Statistics section that
this behavior is more reliable for upside breakouts than downside ones.
The trendlines, when projected into the future, can sometimes act as
areas of support or resistance, depending on which side prices are approaching
(Figures 3.1, 3.3, and 3.7 show examples). Sometimes the support or resistance
level is active for months or even years at a time.

46 Broadening Formations, RightAngled and Desi ji
ding
Pacific Telesis Group (Telecom. Services, NYSE, PAC)
|ul 91 Aug Sep Oct Nov Dec Jan 92 Feb Mar Apr
Figure 3.4 A descending broadening formation with prices that fail to continue
moving up. The partial decline suggests the ultimate breakout will be upward, but
the rise falters and prices move downward instead.
Focus on Failures
Since descending broadening formations can break out in either direction, I
show both views of failed breakouts. The first one, Figure 3.4, is characterized
by the telltale partial decline in late November. From there, the stock climbs
and eventually pierces the top trendline. Once prices close above the trendline,
you would expect them to throw back to the formation top then continue
higher or simply move upward from the start. In this situation, prices stall at 45
and return to the formation proper—a classic throwback. Unfortunately,
instead of rebounding and heading higher like a typical throwback, the stock
continues down. It does some more work inside the formation before shooting
out the other side in a straightline run.
Had you purchased after the upside breakout, you would have seen the
stock decline from a purchase point of about 44'/2 to a low of 367
/s. Even a stop
at the lowest point of the formation would have gotten you out at 39, still a
hefty decline. However, if you held onto the stock (not recommended, by the
way), it would have been rewarding. The low occurred on April 8 (not shown),
and it turned out to the be the lowest price reached during the next 2 years.
The stock hit its peak in early November 1993 at a price ofnearly 60.
Healthcare Compare (Medical Services, NASDAQ, HCCC)
Figure 3.5 A downside breakout failure. Prices decline by less than 5%, turn
around, and eventually hit 42. Such failures are rare, but they do occur, so stoploss
orders are always important. A broadening top formed in early November.
Figure 3.5 shows a more harrowing tale because it involves a short sale.
Investors watching the sharp 2day decline beginning October 14, 1994, would
be tempted to short the stock the next day. Had they done so, or even waited
a few days, they would have bought near the low. From that point on, the stock
moved higher before it pulled back into the formation where it meandered
before ultimately soaring out the top. Ifyou were a novice investor and had not
placed a stop on your short sale, your loss would have taken you from a low of
243
/s to 53, where it peaked near the end of the study.
Figure 3.5 represents a failure type I call 5%failures. That is when prices
break out in a given direction and move less than 5 % before moving substan
tially in the direction opposite the breakout. It is the type of failure that can
turn a small profit into a large loss if stops are not used.
If there is a bright side to the situations shown in Figures 3.4 and 3.5, it is
that failures do not occur very often. The statistics follow, but for now let me
point out that 8 of every 10 formations continue moving in the direction ofthe
breakout. The two figures should also provide a warning to make sure you use
stops to limit your losses. Even if you choose to hold a mental stop in your
head, be sure to pull the trigger once things begin to go bad.

48 Broadening Formations, RightAngled and Deso
Statistics
Table 3.2 shows the general statistics for the descending broadening forma
tion. Of the broadening formations studied so far, this one is the most rare. I
used two databases and still found only 82 formations in over 3,000 years of
daily price data. Rare indeed!
There are 6 more reversals (44) than consolidations (38) of the trend.
Like most formations, I compare the trend before entry to the chart pattern to
the trend after the breakout.
A formation classifies as a failure ifprices do not continue in the breakout
direction by more than 5% before reversing course. Table 3.2 provides the
failure rate for the two breakout directions. Upside breakouts show failure
Table 3.2
General Statistics for RightAngled Descending Broadening Formations
Description
1991 to 1996
1996 to 1998
Failure rate for upside breakouts
Failure rate for downside breakouts
Average rise for successful upside breakouts
Most likely rise
Average decline of successful downside
breakouts
Most likely decline
Of those succeeding, number meeting or
exceeding price target for upside breakouts
(measure rule)
exceeding price target for downside
breakouts (measure rule)
Days to ultimate high, upside breakouts
Days to ultimate low, downside breakouts
Success rate of partial rises
Success rate of partial declines
Statistic
69
13
9 or 19%
1 or 3%
27%
20% to 30%
19%
10% to 15%
34 or 89%
20 or 69%
3 months (88 days)
5 months (148 days)
3 months (86 days)
7/12 or 58%
18/23 or 78%
Note: This rare formation does better on an upside breakout, scoring an average 27% gain.
Statistics 49
rates of 19%, whereas downside breakouts fail 3% of the time. I consider val
ues below 20% to be acceptable, so both directions score well.
The average rise from an upside breakout is 27%, but the most likely rise
is between 20% and 30%. Since I base the most likely rise on a frequency dis
tribution of the gains for each formation, I must caution you not to place too
much emphasis on this particular statistic. There are simply too few samples in
die study on which to base any firm conclusions. The highest frequency, for
example, is seven hits in the 30% range followed by five hits in the 20% cate
gory. A frequency of 30 or higher, by comparison, is reliable. Other formation
types typically have likely rises in the 10% to 15% range, and, if enough sam
ples were present in this formation, that is the performance I would expect.
The average decline from a downside breakout is 19%. This figure is
close to the range (20%) of most bearish formations. The most likely decline
is in the 10% to 15% range, and the same warning applies to the sample size.
Even though the samples may be few, the results are typical and there appear
to be no surprises.
Use the measure rule to predict the target price for the formation after a
breakout. The Trading Tactics section of this chapter explains the calculation
more thoroughly, but the measure rule simply computes the height of the for
mation and adds or subtracts the result from the breakout price. Often, the tar
get price serves as a minimum expected price move. For this formation, upside
breakouts hit or exceed their targets 89% of the time and downside breakouts
reach or decline below their targets 69% of the time. I like to see values above
80%, so the downside target performance is low.
The average formation length is about 3 months (88 days), and takes
between 3 months (for downside breakouts) and 5 months (for upside break
outs) to reach their ultimate price values. The longer time it takes to reach the
ultimate high makes sense since the percentage gain (27%) is larger than the
decline resulting from downside breakouts (19%). In short, it takes longer
to go further.
A partial rise, where prices begin heading up from the lower trendline and
approach but do not touch the top trendline before turning down, results in a
downside breakout 58% of the time. That is not much higher than guessing
the result of a coin toss. However, partial declines—which is the same curling
action only from the top of the formation—result in an upside breakout 78%
ofthe time. That is high enough on which to base a trading strategy: Ifyou see
a partial decline that begins heading up, buy the stock.
Table 3.3 outlines the breakout statistics. Since this formation can break
out either way, the failure statistics appear in two categories: upside and down
side failures. Only 19% of the formations breaking out upside fail to continue
moving up by more than 5%. For downside breakouts, only one formation fails
to continue moving down. Remember, the sample size is only 47 and 30 sam
ples for upside and downside breakouts, respectively.

50 Broadening Formations, RightAngled and DescCfling
Table 3.3
Breakout Statistics for RightAngled Descending Broadening Formations
Upside breakout
Horizontal breakout
Downside breakout
Throwbacks
Fullbacks
Percentage of upside breakouts occurring near
12month price low (L), center(C), or high (H)
Percentage gain for each 12month lookback period
Percentage of downside breakouts occurring near
12month price low (L), center (C), or high (H)
Volume for breakout day and next 5 days compared
with day before breakout
Percentage of successful breakouts occurring on
Percentage of failed breakouts occurring on
9 or 19%
Ior3%
47 or 57%
5 or 6%
30 or 37%
II or23%
11 days
10 or 33%
14 days
L0%, C26%, H74%
L0%, C41%, H25%
L40%, C40%, H20%
L31%, C14%, 1110%
143%, 105%, 96%,
92%, 83%, 71 %
H76%, L24%
H70%, L30%
Note: For a traditionally bearish formation, there are more upside breakouts than downside
ones—57% versus 37%.
There are three types ofbreakouts: up, horizontal, and down. Most ofthe
formations (57%) break out upward. Downside breakouts are next at 37%,
with the remainder breaking out horizontally. If you are wondering what a
horizontal breakout looks like, see Figure 3.6. As shown, after the formation
ends, the stock moves horizontally for about 6 months before closing above the
top of the formation. Even so, in about 4 months, prices dip below the forma
tion top again. Not until July 1994 do prices stage a strong rally and move
decidedly higher.
Throwbacks and pullbacks are rare, being found in only 23 % and 33 % of
the formations, respectively. The duration of pullbacks, at 14 days, is slightly
above normal (10 to 12 days) for all formation types. I ignore any throwback or
pullback beyond 30 days because I consider it normal price action and not due
to a throwback or pullback.
Statistics 51
Avery Dennison Corp (Chemical (Specialty), NYSE, AVY]
92 M A M | J A S O N D 93F MA M J ] A S O N D 94F MA M| j A S O ND 95 FM AM)
Figure 3.6 A horizontal breakout. Prices do not rise above or fall significantly
below the formation for months on end, shown on a weekly scale.
Most of the upside breakouts (74%) occur near the yearly high. This
observation suggests descending broadening formations appear at the end of
an upward trend. Note that there are no formations appearing near the yearly
low. Since I use the top of the formation as the benchmark, this result is not
unusual.
The best performance from upside breakouts occurs when the breakout is
in the middle third of the yearly price range; why is unclear. In other forma
tions, the tendency is for the upper range to perform best, possibly due to
momentum players taking hold of the stock and bidding it higher after an
upside breakout. Since this formation broadens downward, it might scare off
investors even after an upside breakout.
Downside breakouts tell a different tale. Most breakouts split between the
lower and center third of the yearly price range. The highest price range gar
ners only 20% of the formations since I use a breakout below the lowest for
mation low in the computation. True to form, formations breaking out
downward in the lowest third of the yearly price range perform best, with an
average decline of 31 %. This performance suggests that it is better to short
stocks making a new yearly low than a yearly high.
Breakout volume is surprising in that it peaks then quickly recedes. The
breakout volume is, on average, 43% above the prior day (or 143% of total
volume) and diminishes rapidly and steadily to 71 % a week later. I did not sep

52 Broadening Formations, RightAngled and Deseed,ng
arate the volume numbers into the two breakout types; the numbers apply to
all descending broadening formations.
The breakout volume for successful breakouts is indistinguishable from
unsuccessful ones. Both occur on high volume about threequarters of the
time. So, if you see weak volume on an upside breakout, do not be too con
cerned about an impending failure.
Trading Tactics
Table 3.4 outlines trading tactics for descending broadening formations. Fig
ure 3.7 illustrates the computation of the measure rule. Compute the forma
tion height by first taking the difference between the highest high (49'/>) and
the lowest low (43'/2). Add the result (6) to the value of the horizontal trendline
to get a target price of 55l
/2. Prices reach this target during midMarch 1996 as
the stock climbs on its way to 60.
Table 3.4
Trading Tactics for RightAngled Descending Broadening Formations
Trading Tactic
Wait for confirmation
Stops
Intraformation trading
Partial decline
Explanation
Measure rule Compute the formation height by taking the difference
between the horizontal top and the lowest low in the
formation. For upside breakouts, add the result to the value
of the horizontal trendline. For downside breakouts,
subtract the value from the lowest low. The result is the
expected target price.
It is unclear which way prices will break out, so it is best to
wait for prices to close outside the trendlines. Once they do,
expect prices to continue moving in the direction of the
breakout. Place your trades accordingly.
Once a breakout occurs, consider the opposite side of the
formation as the stoploss point. However, in many cases
you will want something closer to your purchase price so
look for nearer support or resistance zones. Once the stock
moves substantially, advance the stop to the breakeven
point.
Once you recognize a broadening formation, consider
placing a trade as prices reverse course at the trendline. Co
long at the bottom and short at the top but be sure to use
stops to protect against an adverse breakout.
Buy a stock if you see a partial decline once prices curl
around and begin heading back up.
Note: Always wait for the breakout, then trade with the trend.
Trading Tactics 53
Ifthe stock breaks out downward, the measure rule computation is nearly
the same. Subtract the formation height from the lowest low giving a target
price of 37'/2. Be aware that upside breakouts are more likely to reach their tar
gets (89%) than downside ones (69%).
Once you know the target price, you can make a profit and loss assess
ment for the trade. What is the likely downside move compared with the tar
get price? Does the potential profit justify the risk of the trade? For Figure 3.7,
there is support in the 46 to 47 area. Examining the peaks and valleys of the
prior price action determines support and resistance levels. In March 1995
(not shown in Figure 3.7), there is an area of congestion bounded by a sym
metrical triangle with an apex at about 46. Additional resistance appears inJuly
and October, as shown in Figure 3.7. Together, the 46 to 47 area makes a good
location for a stoploss order.
Let us say the stop is 453
/4, just below the bottom of the support area. If
the trade happens at 50'/2, which is the close the day after the upside breakout,
that gives a potential loss ofless than 10%. With a target price of 55'/2, or 10%
upside, the win/loss ratio is an unexciting one to one. In such a situation, you
Figure 3.7 An upside breakout from descending broadening formations. To com
pute the measure rule for upside breakouts from descending broadening forma
tions, find the difference between the high and low in the formation, denoted by
points A and B. Add the result to point A to get the target price. It took almost 7
months for prices to exceed the target. A small symmetrical triangle appears at
point C.

54 Broadening Formations, RightAngled and Descenv...ig
nv...ic
could either tighten your stop by moving it higher (and risk getting taken out
by normal price action) or look elsewhere for a more profitable trade. Remem
ber there is no rule that says you have to place a trade.
The Statistics section of this entry introduces you to partial declines—
78% of the time an upside breakout follows. That is high enough to risk a
trade. If you see a partial decline occur (and it really does not matter how far
down it dips, so long as it is not touching or coming too close to the lower
trendline) and it begins heading back up, buy the stock. With any luck, it will
shoot out the top of the formation and continue higher. As always, be sure to
place a stoploss order and raise it as prices climb.
Sample Trade
Ralph is a formation trader with a measure of experience milking chart patterns
for all they are worth. When he noticed what he thought was either a descend
ing broadening wedge or a rightangled descending broadening formation, he
bought the stock. His order, placed at point C in Figure 3.7 (463
/8); was just
after the stock bounced off the lower trendline.
He monitored it closely, and watched the stock move up the very next
day, then ease lower. After a few days, Ralph saw a symmetrical triangle form
and he became worried. Those formations, he reasoned, usually follow the
trend and the trend was downward. When the stock moved below the lower
triangle trendline, Ralph sold the stock and received a fill at 461
/2.
When he erased the lines from his computer screen and looked at the
fresh price chart, he knew he had made the right decision because a partial rise,
such as where the triangle formed, usually portends an immediate, downside
breakout.
Sure enough, the following day prices dropped even further, tagging the
broadening formation trendline again. Then they rebounded. In the coming
days, he watched as prices surprisingly zipped across the formation and
touched the top trendline. Ralph took a small loss after factoring in commis
sions. Did he sell too soon or was he just being cautious, and what lessons did
he learn? Spend some time searching for the answers in your own trades and
you will rapidly become a better investor.
4
Broadening Tops
R E S U L T S SNAPSHOT
Upside Breakouts
Appearance
Failure rate
Average rise
Volume trend
Percentage meeting
Surprising finding
Synonyms
See also
Price trend is upward, leading to the formation.
lows that widen over time. Breakout is upward.
Shortterm (less than 3 months) bullish consolidation
4%
34%, with most likely gain between 10% and 15%
75%
Partial rise at the end of the formation predicts a
downside breakout 65% of the time and a partial
decline predicts an upside breakout 86% of the time.
Expanding triangle, orthodox broadening top, and
fivepoint reversal
Broadening Bottoms; Broadening Formations, Right
Angled and Ascending; Broadening Formations,
RightAngled and Descending; Broadening Wedges,
Ascending; Broadening Wedges, Descending
55

56 Broadening Tops Tour 57
Downside Breakouts
Appearance
Failure rate
Average decline
Volume trend
Percentage meeting
See also
Price trend is upward, leading to the formation.
lows that widen over time. Breakout is downward.
Shortterm (less than 3 months) bearish reversal
4%
64%
Broadening Bottoms; Broadening Formations, Right
Angled and Ascending; Broadening Formations,
RightAngled and Descending; Broadening Wedges,
Ascending; Broadening Wedges, Descending
Broadening tops, not surprisingly, act a lot like broadening bottoms. What
separates a top from a bottom is the price trend leading to the chart pattern.
For tops, the intermediateterm price trend is upward; for broadening bot
toms, it is downward. This is an arbitrary distinction I made just to see if the
two formations act differently. In answer to the question you have probably
posed right now: The two formations act similarly, but their performance dif
fers slightly.
Broadening tops divide into two types: those with upside breakouts and
those with downside ones. You can see in the Results Snapshot that the failure
rates are the same for both: 4%. That is a very good number. I consider failure
rates below 20% to be acceptable.
The average gain for upside breakouts is subpar at 34%. Wellbehaved
bullish formations usually score about a 40% rise. Downside breakouts from
broadening tops perform better than normal with a 23% loss. The usual loss
for all bearish chart patterns is about 20%.
The broadening top formation performs better than some chart patterns
regarding the measure rule. The rule estimates a target price for the stock.
Three out of every four broadening tops (75%) with upside breakouts meet or
exceed their price targets, whereas 64% of those formations with downside
breakouts meet or exceed theirs. Still, I consider values above 80% to be reli
able, so the formation comes up a bit short.
This formation does excel in predicting the breakout direction. When a
partial rise or decline occurs—that is, when prices move toward the opposite
side, then reverse before touching the trendline—it is a breakout signal. For
partial rises, the breakout direction is downward. A downside breakout follows
those formations showing a partial rise 65% of the time. Partial declines score
even better when they break out upward. Almost all the formations (86%)
showing a partial decline correctly predict an upside breakout.
Tour
Broadening formations come in a variety of styles and names. There are the
broadening tops and bottoms, rightangled ascending and descending, expand
ing triangle, orthodox broadening top, and fivepoint reversal. The last three,
expanding triangle, orthodox broadening top, and fivepoint reversal, are syn
onyms of the broadening top formation, with the last two being based on five
turning points.
For a quick tour of the formation, look at Figure 4.1. The stock began an
uphill run in December 1994 and paused for about 2 months in May and June.
Then it continued its climb and reached a high in midSeptember at a price of
533
/4. Holders of the stock, witnessing the long run, decided to sell their shares
and the stock headed lower. On September 25, 1995, volume spiked upward
and halted the decline. Investors, seeing a 40% retrace of their gains from the
Beneficial Corp. (Financial Services, NYSE, BNL)
Double Top
Jun 95 |ul Aug Sep Oct Nov Dec |an96 Mar
Figure 4.1 A double top changed into a broadening formation. The oneday
reversal appeared as the third peak after an unsustainably quick price rise. The
broadening top formation marked a struggle between eager buyers and reluctant
sellers at the lows and the quicktotakeprofit momentum players at the peaks.

58 Broadening Tops
June level, apparendy thought the decline overdone and purchased the stock,
sending prices higher.
Prices peaked at a higher level, 54V:, on October 19. Many diligent in
vestors probably suspected that a double top was forming and promptly sold
their holdings to maximize their gains, sending the price tumbling. Prices con
firmed the double top when they fell below the confirmation point, or the
lowest low between the two peaks, at 483
/4.
Volume picked up and the struggle between supply and demand
reasserted itself. The decline stalled as traders willing to buy the stock over
whelmed the reluctance to sell. The stock turned around and headed higher.
By this time, chart followers could draw the two trendlines—one across the
twin peaks and another below the two valleys. The broadening top formation
was born. Astute traders probably jumped on the bandwagon at this point and
purchased the stock. They wanted to play the anticipated rise as the formation
broadened out. The stock cooperated and moved higher, reaching the top
trendline once again at a new high of 55l
/2.
The steepness of the ascent in the latter stages was unsustainable. The
peak looked like a oneday reversal, with a close near the low of the day and a
wide daily price swing. However, volume was unconvincing. It was higher that
day than during the prior week, but it certainly was not of the caliber ofthe late
November spike. In any case, the stock tumbled downward and soon reached
a new low of 43'/2, stopping right at the downsloping trendline. Once the
stock began moving higher, the momentum players jumped on board and vol
ume increased along with the price. Buying enthusiasm and rising momentum
pushed the stock higher, climbing through the top trendline. An upside break
out occurred.
Throughout the various peaks and troughs of this formation, there was a
struggle between buyers and sellers. Near the lows, the buyers believed the
stock was oversold and they eagerly bought it. At the top, they quickly sold
their shares and pocketed their handsome profits. This selling, of course, sent
the stock back down.
Some investors, seeing the stock decline below their purchase price and
still believing that the stock had value, bought more. The behavior also helped
turn the stock around at the lows and probably explained their heightened
nervousness at the tops. They wanted to keep their gains this time, instead of
watching them evaporate should the stock decline again. Once a higher high
was evident and the stock turned lower, they sold, forcing down the stock more
quickly this time and on higher volume. You can see this on the declines after
the second and third peaks.
The formation in Figure 4.1 also makes evident that identifying the ulti
mate breakout is exceedingly difficult. It appears that each new high or new
low may be the final breakout. Only when prices move in the opposite direc
tion is it clear that prices will not break out. We explore ways to profit from
this behavior in the Trading Tactics section.
Table 4.1
Identification Characteristics of Broadening Tops
Pricetrend The intermediateterm price trend leading to the formation
should be up.
Shape Megaphone shape with higher highs and lower lows. Five
point reversals have three peaks and two troughs.
Trendlines Prices are bounded by two trendlines: The top one slopes up
and the bottom slopes down.
Touches Should have at least two minor highs and two minor lows,
but not necessarily alternating touches.
Volume Irregular but usually rises as prices rise and recedes as prices
fall.
Breakout The breakout can occur in either direction and, in several
cases, prices move horizontally for several months before
staging a definitive breakout.
Table 4.1 shows the identification guidelines for the broadening top forma
tion. The first criterion is the price trend leading to the formation. This price
trend is what differentiates a broadening top from a broadening bottom. For a
broadening top, the price trend should be leading up to the formation, not
down as in its bottom counterparts. This is just an arbitrary designation I have
chosen to distinguish the two formations.
Trendlines drawn across the peaks and valleys resemble a megaphone.
Higher highs and lower lows make the formation obvious to those versed in
spotting chart patterns. The slope of the trendlines is what distinguishes this
formation from some others. The top trendline must slope up and the bottom
one must slope down. When one of die two trendlines is horizontal or nearly
so, die formation classifies as a rightangled ascending or descending broaden
ing formation. When the two trendlines slope in the same direction, die for
mation is a broadening wedge.
There should be at least two minor highs and two minor lows before die
chart pattern becomes a broadening top. A minor high is another name for a
distinct price peak. A minor low refers to the valley pattern as prices descend
to a low then turn back up. Again, die minor low should be a distinct trough
diat is easily recognizable. Figure 4.1 shows diree minor highs touching die
top trendline and four minor lows either nearing or touching die bottom
trendline. The minor highs and minor lows need not alternate as prices criss
cross die formation.

60 Broadening Tops
Linear regression on the volume trend shows it splitting evenly between
trending up and trending down. If you look closely at most broadening top
chart patterns, you will see that volume typically follows price. When prices
rise, so does volume; when prices fall, volume recedes. In Figure 4.1, volume
rises as prices climb into the beginning of the formation and round over at the
peak. Then volume declines, following prices lower. However, I attach no real
significance to volume in a broadening top formation.
A breakout happens when prices move outside the trendline boundaries
or follow a trendline for an extended time. In Figure 4.1, ifyou extend the top
trendline upward, it will intersect prices at about 58. Prices push through this
level and move higher. When a breakout occurs, I consider the actual breakout
price to be the value of the highest peak in the formation. In Figure 4.1, for
example, the breakout price is 55'/2, or the high at the early December peak.
For an example of how to apply the various guidelines, consider the
broadening top shown in Figure 4.2. At first glance, it looks like a large mega
phone with price trends that generally follow two sloping trendlines. The top
trendline slopes upward and the bottom one slopes downward, each intersect
ing the minor highs or lows at least twice. Prices, over time, form higher highs
and lower lows until they breakout of the formation, generally moving beyond
the line of trend before retracing.
The volume pattern is irregular but generally rises as prices move up and
recedes as prices move down. Figure 4.2 shows this quite clearly. During the
rise in midNovember, for example, volume jumped upward as prices peaked,
then just as quickly receded as prices declined.
Where in the yearly price range does the formation appear? Broadening
tops, as you would expect, appear near the top of a price range or near the top
of an upward trend.
The only characteristic not discussed so far is that the formation is usu
ally horizontal and symmetrical. The formation does not appear as an ascend
ing or descending broadening wedge, with both trendlines either sloping up or
both sloping down. In broadening formations one trendline must slope up
while the other must slope down.
Orthodox broadening tops and fivepoint reversals describe the same type
of formation. They are simply broadening tops that have three minor highs
and two minor lows. Figure 4.3, for example, falls into this category. Other
than the name, I found no substantial difference between broadening tops and
orthodox broadening tops or fivepoint reversals.
Some analysts say fivepoint reversals are bearish indicators, that the for
mation predicts a downward breakout. My statistics, admittedly on only 30 for
mations, suggest this is untrue. Sixteen break out upward and the others break
out downward. The sample size is too small to make a definitive statement, but
the numbers do reflect the general trend of upside breakouts for all broaden
ing top formations, that is, slightly more than half (53%) break out upward.
ASA Ltd (Investment Co. (Domestic), NYSE, ASA)
6
Nov Dec Jan 92 Feb Mar Apr May Jun
Aug91 Sep Oct
Figure 4.2 The broadening top has higher highs and lower lows as the price
action widens over time.
Albertsons Inc. (Grocery, NYSE, ABS)
|un 95 |ul Aug Sep Oct Nov Dec Jan 96 Feb Mar Apr
Figure 4.3 A weak example of a fivepoint reversal or orthodox broadening top.
It has three minor highs and two minor lows composing the five turning points.
61

62 Broadening Tops
Focus on Failures
What does a failure look like? Look at Figures 4.3 and 4.4, two examples of
broadening patterns that fail to continue in the expected direction. Figure 4.3
shows a sharp downward thrust that pierces the trendline on high volume.
Since this is clearly outside the lower trendline, and coupled with the failure of
prices to attain the upper trendline, a downside breakout is at hand. But the
downward movement stalls on very high volume, turns around, and moves
higher. This is an example of a 5% failure, that is, prices break out then move
less than 5% in the direction of the breakout before heading substantially in
the other direction. In this case, the breakout direction is downward, but prices
recover before moving lower than 5 % below die breakout point.
Contrast the behavior shown in Figure 4.3 with that shown in Figure 4.4.
I include this chart because I have noticed that a large number of broadening
formations act this way. Instead of making a clear up or down thrust that
pierces the trendline, prices move horizontally for months on end before
finally moving above or below the formation highs or lows.
In the case of Figure 4.4, prices decline below the low in early July and
halt. They climb for a bit then recede again and reach a new low in early
August. Another recovery sees prices rise no higher than 44 for about half a
year before finally staging an upside breakout.
Arco Chemical Co. (Chemical (Basic), NYSE, RCM)
Apr May Jun Dec Jan 94 Feb Mar
Figure 4.4 Prices in this broadening top moved horizontally for 6 months before
staging an upward breakout. This is a common occurrence with broadening tops.
Statistics 63
Even Figure 4.3 shows a consolidation pattern for 6 weeks after prices
pierce the trendline. Prices do not move very high before stalling and essen
tially travel sideways for an extended period. Then a new trend sets in and
prices finally break out in the genuine direction.
The last point I want to make is that failures are rare. Only 8 occurred in
nearly 200 formations. Once a broadening formation breaks out, it continues
moving in the same direction.
Statistics
As I completed my first pass through my main database, it became clear that I
was not finding many broadening formations. Only about 25% of the stocks I
examined had usable formations. So I expanded the sample size by adding the
database that I use on a daytoday basis. This latter database is smaller, with
nearly 300 stocks, but it is only about 3 years long. This compares with the 5
year, 500 stock database used throughout this book. Together, there still were
not that many formations, only 189, but the sample size is sufficiently large
enough to draw valid conclusions. Table 4.2 shows the general statistics.
Does the formation perform as a reversal or consolidation of the existing
trend? Just over half (100) are continuations of the trend (consolidation) and
Table 4.2
General Statistics for Broadening Tops
Description Upside Breakout Downside Breakout
Number of formations: 132 in
500 stocks from 1991 to 1996,
57 in about 300 stocks from
1996 to 1999
Failure rate
Partial rise but ended down
Partial decline but ended up
Percentage of time there was
a trend reversal within 3 months
100
100 consolidations
4/100 or 4%
34%
10% to 15%
72 or 75%
2.5 months (72 days)
45/69 or 65%
25/29 or 86%
41%
89
89 reversals
4/89 or 4%
23%
10% to 20%
54 or 64%
2 months (67 days)
45/69 or 65%
25/29 or 86%
47%

64 Broadening Tops
the others (89) are reversals. The formation's behavior depends on the direc
tion ofthe breakout. Formations with upside breakouts act as consolidations of
the trend because one ofthe criteria for a broadening top is that the price trend
leading to the formation is upward. Thus, they continue the upward trend;
broadening tops with downside breakouts act as reversals.
Some analysts say that broadening top formations are inherently bearish
reversals, but I found little to support that claim. I prefer to think of them as
weakly performing bullish patterns (those with upside breakouts) or mildly
strong bearish patterns (downside breakouts).
The failure rate is not easy to determine since the formation has no indi
cation in which direction it will ultimately break out, so I used a 5% price
movement as the benchmark. This breakout means that ifprices decline below
(or rise above) the lowest (highest) point in a formation by less than 5%, then
turn around and head in the opposite direction, the formation is a failure.
The reasoning for this approach is simple. If you purchased the stock
after an upside breakout but the stock failed to continue rising, then you would
be disappointed and may even take a loss. Given such constraints, out of 189
formations, only 8 fail to continue moving in the direction of the breakout.
That is a good score.
Once a breakout occurs, what is the average move? For formations with
upside breakouts, the average rise is 34%, whereas formations with downside
breakouts decline by 23%. The average gain is a bit under the 40% that bull
ish formations exhibit, but the average decline is above the 20% loss for bear
ish chart patterns.
Large percentage moves can skew the overall average, so I computed the
most likely rise and decline using a frequency distribution. Figure 4.5 shows a
graph of the results. Large gains are the case with upside breakouts as you can
see in the figure. Twentyone percent of the formations with upside breakouts
have gains over 50%. Ifyou ignore this column, then the next highest one is for
returns in the 10% to 15% range. I consider this range to be the most likely
return that investors can expect. Ifyou are lucky, you will have a formation that
does considerably better, just do not bet that farm that it will happen.
For downside breakouts, the most likely loss ranges between 10% and
20%, which includes the three tallest columns in the graph; they are near
enough to one another that I lump them together. Over half the formations
(56%, or the sum of die first three columns) with downside breakouts have
returns less than 20%.
The measure rule applies just as it does for most formations. Find the
formation height and add the result to the highest point in the formation for
the upside target and subtract the result from the formation low for the down
side target.
You can see in Table 4.2 that 75% ofthe upside breakouts reach the tar
get price. However, downside breakouts fare much worse, with 64% reaching
or exceeding their targets. I consider values above 80% to be reliable, so the
Statistics 65
10 15 20 25 30 35
Percentage Cain or Loss
LTH Cains • Losses
40 >50
Figure 4.5 Frequency distribution of returns for broadening tops. The most likely
gain for upside breakouts is 10% to 15%, and the loss is likely to be in the 10% to
20% range after a downside breakout.
broadening top does not quite make the grade. I explain the measure rule fur
ther in the Trading Tactics section.
The average formation length is nearly the same for both of the breakout
types at 72 and 67 days. The formation is long enough to show up on the
weekly charts.
Partial rises and declines are a reliable way to determine the breakout
direction and they allow an investor to take a position in the stock early. Fig
ure 4.3 shows a partial rise that correctly predicts a downside breakout. For
broadening tops that show a partial rise, where prices leave the lower trend and
rise up but do not come close to the top trendline, a downside breakout follows
65% of the time. The results are even better with partial declines, which are
dips from the top trendline that move downward without coming close to the
lower trendline. Partial declines correctly predict an upside breakout 86% of
the time.
Some analysts suggest that broadening formations are bearish formations.
Even when they do not work, analysts maintain, a trend reversal soon appears.
I scanned my database and looked at all formations to see if this is true. Only
41% of die formations widi upside breakouts reach their ultimate high in less
than 3 months. This means tJiat less dian half suffer a trend reversal in under
3 months. Downside breakouts perform marginally better, at 47%, but still not
high enough to support other analysts' premise. It is my belief, supported by
the statistics, that even when things do not work out as expected, a broadening
top does not presage the end of the trend.

66 Broadening Tops
Table 4.3 shows breakout statistics. Figure 4.4 is a good example of the
distance between where the formation ends and where the breakout occurs.
For both breakout types, it takes over a month to reach the breakout point, on
average. I consider the breakout to be where prices either rise above the high
est high in the formation or drop below the lowest low.
In the 100 formations I examined with upside breakouts, it takes nearly
half a year (166 days) to reach the ultimate high. Downside breakouts, on the
other hand, reach their ultimate low much more quickly, in about 4 months
(129 days). This makes sense because it takes longer to rise 34% than to fall by
23% (the average gain or loss from Table 4.2). Both durations measure from
the breakout point to the ultimate high or low point (a point where a signifi
cant trend change occurs).
Where in the prior 12month price range do breakouts occur? Most
(94%) upside breakouts happen in die upper third of the price range. Although
this may sound suspicious, it is reasonable. Consider that upside breakouts
occur along the top of the formation placing them closer to the top of the
yearly price range. Since broadening tops occur near the end of an upward
trend, it should come as little surprise that most (49%) downside breakouts
from broadening tops appear in the center third of the yearly price range. The
highest third follows with 40%.
Substituting the gains or losses for prices shows interesting results.
Upside breakouts from broadening tops climb 35% on average when they
occur in the upper third of the yearly price range. Downside breakouts split
evenly with about 22% in each category (with a slight edge, 24%, going to the
center third).
Table 4.3
Breakout Statistics for Broadening Tops
Description Upside Breakouts Downside Breakouts
Number of breakouts
Formation end to breakout
For successful formations,
days to ultimate high/low
low (L), center (C), or
high (H)
Percentage gain/loss for
each 12month lookback
period
100 or 5 3%
46 days
5.5 months (1 66 days)
L0%, C6%, H94%
L0%, C24%, H35%
89 or 47%
34 days
4 months (129 days)
L10%,C49%,H40%
L22%, C24%, H22%
Note: The best performing broadening tops with upside breakouts are those that appear
in the upper third of the price range.
Statistics 67
Table 4.4 shows statistics for alternating touches. I wanted to determine
if there is a relationship between the number of alternating touches and the
breakout. A frequency distribution of the number of alternating touches
reveals that most formations have four alternating touches before staging a
breakout (either upward or downward). Some, however, go on to cross the for
mation several more times before ultimately breaking out.
How did I count the number of alternating touches? Look back at Figure
4.2. The numbers represent the alternating touches. I usually tag the first
touch on the side opposite the entry point. In Figure 4.2, the entry is from the
bottom, but it almost makes its way to the top, so I decided to call the first
alternating touch on the bottom. Figure 4.3 shows an easier example of count
ing alternative touches.
Returning to Table 4.4, turning the numbers into percentages gives an
investor a better feel for how likely a breakout will be. For example, ifyou have
a formation with the first touch on the top and a total of five alternating
touches, then the likelihood of a downside breakout (since the last touch is also
on the top) is 95% should prices cross the formation again. That is a safe bet,
but it does not mean that the formation will break out downward. Since prices
are already at the top, they could do a partial decline, turn around, and break
out upward.
There is also the belief among some technical analysts that the breakout
will occur opposite the side with the most touches. For fivepoint reversals
(which have three touches on top and two on the bottom), the breakout should
be downward. Of the 30 fivepoint reversals in my database, only 47% break
out downward.
What about inverted fivepoint reversals? Such a formation has two
minor highs and three minor lows. The statistics do show that these formations
break out upward twice as often as downward (68% versus 32%). However, I
Table 4.4
Frequency Distribution of Successful Formations by Number of Alternating
Touches with Cumulative Percentage of Total
Note: The table indicates that most successful broadening tops have four alternating
touches before breaking out.
Number of
Alternating
Touches
3
4
5
6
7
Number with
Upside
Breakouts
8: 8%
42: 52%
31: 84%
12: 97%
3:100%
Number with
Downside
Breakouts
8: 9%
35: 51%
21: 75%
17: 95%
4:100%

68 Broadening Tops
hasten to add that there are 36 additional formations that have more than 5
touches. Some fivepoint reversals, inverted or not, go on to bounce off the
opposite side several more times. Once that happens, of course, they are no
longer fivepoint reversals. My conclusion is that there is nothing magic about
fivepoint reversals or their inverted brethren. They are all broadening forma
tions and there is really nothing special about them that I have been able to
determine.
Trading Tactics
Table 4.5 outlines trading tactics for broadening top formations. The first
thing to consider about trading tactics is the measure rule. The measure rule
predicts the price to which the stock will move. For many formations, one sim
ply computes the height of the formation and adds the result to the breakout
price. Broadening formations are not much different. Consider Figure 4.6.
The height of the formation is the difference between the highest high (12H)
and the lowest low (10), or 2l
/s. For upside breakouts, add the height to the
Table 4.5
Trading Tactics for Broadening Tops
Measure rule
Co long at the low
Long stop
Go short at the high
Short stop
Move stops
Other
Compute the height between the highest high and the
lowest low in the formation. Add or subtract the height
from the highest high or lowest low, respectively. The
results are the target prices for upside and downside
breakouts.
Once a broadening top appears, buy after the stock
makes its turn at the low.
Place a stoploss order below the minor low. Should
the stock reverse course, you will be protected.
Sell short after prices start heading down at the top.
Place a stop above the minor high to protect against
an adverse breakout. Cover the short when it turns at
the bottom trendline and starts moving up. For a
downside breakout, cover as it nears the target price or
any support level.
Raise the stop to the next lowest minor low (for prices
moving up) or next highest minor high (for prices
moving down) as prices cross the formation.
Go long if a broadening top shows a partial decline.
Consider adding to your position once it makes an
upside breakout.
Trading Tactics 69
highest high in the chart pattern giving a minimum price move of 14'/4, as
shown in the figure.
How do you make use of the measure rules? Imagine that you are con
sidering purchasing the stock. Since it is never clear which way a broadening
formation will ultimately break out, it is difficult to pick a good longterm
entry point. The easiest way to invest using the formation is to buy just after
the stock turns at the bottom trendline.
Since a broadening formation requires two points along the top trendline
and two along the bottom before the formation appears, point A in Figure 4.6
shows one likely investment location. Before placing the buy order, compute
the target price using the measure rule. The target price will help you deter
mine ifthe potential gain is worth the risk. In the example shown in Figure 4.6,
the purchase price is about 103
/8 and the target price is 14'/4, a 37% move. The
stop loss should be 97
/8, for a potential loss of 5%, which gives a rewardtorisk
ratio of 7 to 1, more than enough to risk a trade.
On an upside breakout, prices reach the target 75% of die time, on aver
age. This means you need not feel pressured to sell too early. Let the stock
wind its way upward while you watch your profits grow.
Buy the stock as soon after it touches the lower trendline and moves
higher. Place a stoploss order % below the lowest low (V8 below point A).
Should prices drop, your position will likely be sold before a large loss occurs.
Esterline Technologies (Precision Instrument, NYSE, ESL)
^ Double Top _
Jul 91 Aug Sep Oct Nov Dec |an 92 Feb Mar Apr
Figure 4.6 Use the measure rule to compute the target price. First, compute the
formation height from the highest high to the lowest low then add or subtract the
height from the highest high or lowest low, respectively. Depending on the break
out direction, the result is the expected target price.

70 Broadening Tops "
If the stock fails to break out upward, perhaps you can capture the ride up
to the upper trendline for a 20% rise. Buy when prices turn upward at the
lower trendline, then be ready to sell once prices reach the old high. Prices may
pause for a bit before moving higher and tagging the top trendline or they may
reverse at this point. Make sure your stops have been raised to protect some of
your profits.
What about the measure rule for downside breakouts? Again, the forma
tion height is 2'/s. Subtract 2'/g from the lowest low (10) to arrive at the target
price of 77
/8. If prices break out downward, they should reach 77
/8. However,
the reliability of the target price for downside breakouts is just 62%—not very
reassuring.
For short positions in broadening tops, open the short after the price
touches point B and begins heading down. Place a stop Vs above the highest
high (12'A in this case) to limit your losses. Lower your stop to the next minor
high or apex of the broadening top (either 1 l7
/s or 11 l
/s in this case) once the
stock nears point A. Sometimes the stock will not make it down to the trend
line before beginning to move up. At other times, there is a lengthy pause
before prices turn around or continue down. A lower stoploss point helps you
achieve at least some measure of profit.
The last trading tactic is to look out for partial rises or declines, which
occur when prices begin to cross the formation but do not come close to the
opposite side. Instead, prices reverse course and break out soon after. When
you see a partial rise or decline, place a trade once the stock reverses course. If
a breakout happens, then consider adding to your position.
Sample Trade
Sandra has always taken a liking to the stock market but never had enough
money to jump into the game. Still, she paper traded stocks just to get a feel for
the markets and dreamed ofone day trading for real. Then her parents died in
a tragic car accident with a drunk driver.
The year that followed was tough for Sandra because she was close to
them and missed them dearly. She bought a dog to help fill the void in her life,
but it was not the same. Fortunately, her parents had insurance and a few sav
ings, all of which she inherited. After paying taxes to the government, she sud
denly realized there was no need to continue working. "Why wait till I'm older
when my health might be gone or I might die young like my parents?" She
retired at 29.
Sandra knew that if she cut her expenses to the bone she could live off the
savings. She paid off the mortgage, the car loan, credit card balances, and
stopped going out to eat in restaurants. Her lifestyle changed to accommodate
the limited income but one thing she would not compromise: her paper trading.
Sample Trade 71
After opening a brokerage account, she waited for the perfect trade and
finally found it; the one shown in Figure 4.6. She saw the broadening top for
mation early enough to buy into the stock before the breakout. Two days after
the stock reached point A, she placed her order and received a fill at lO'/j.
Immediately, she placed a stop '/s below the lowest low, or 97
/s for a potential
loss of 6%.
Sandra applied the measure rule and was looking at an upside target of
14'/4. If everything worked as expected, that would give her a return of over
35%. After she placed the trade, she sat back and waited but kept an eye on the
price action. When prices paused at the high of the formation, she wondered
if the trend was going to reverse course. She considered taking her profits and
running but decided against it. After a few days, she recognized a flag forma
tion and hoped that it represented a halfmast move (the flag being a halfway
point in the upward move). If that was true, she could expect a climb to 13%
(that is the distance from the top of the flag (12%) to the start of the move at
10 projected upward using the lowest low in the flag at llH).
A few days later, the stock not only fulfilled the measure rule for the flag
formation, but for the broadening top as well. Did Sandra sell? No. Since the
stock was moving up, she wisely decided to let her profits ride. However, she
did raise her stop to Il7
/s, or H below the formation high. She viewed this
point as a support zone and hoped that should the stock retreat to that level, it
would rebound before taking her out.
In midFebruary, just after the second peak around 1 7 appeared, she rec
ognized a double top formation. She moved her stop to Vg below the confir
mation point, or 14'/4. About two weeks after raising her stop, her position sold
when the stock plunged from the prior close at 1 5s
/8 to 1 2 Vs. After commission
costs, she made 3 3 % in less than 4 months

5
Broadening Wedges,
Ascending
Tour 73
RESULTS SNAPSHOT
Appearance
Reversal or
consolidation
Failure rate
for downside
breakout
Average decline
Volume trend
Fullbacks
Throwbacks
Percentage meeting
predicted price
target
Surprising finding
Price action follows two upsloping trendlines that
broaden out.
24%
6%
20%, with most likely decline being 10%
Slight tendency to increase over time
21%
7%
61%
A downside breakout follows a partial rise 84% of the
time.
I call this formation an ascending broadening wedge because it resembles a ris
ing wedge formation with a broadening price pattern. I first noticed it when
searching for the symmetrical variety of broadening formations. Symmetrical
broadening formations have a price pattern that revolves about the horizontal
axis—prices form higher highs and lower lows and two sloping trendlines out
line the formation; one trendline slopes up and the other slopes down.
The ascending broadening wedge is different from a rising wedge. Both
ascending broadening wedge and rising wedge trendlines encompass price
action that slopes upward. However, that is where the similarity ends. The top
trendline of ascending broadening wedges slopes upward at a higher rate than
the bottom one, giving the appearance of a broadening price series.
Volume of ascending broadening wedges is also opposite that of rising
wedges: Volume for the ascending broadening wedge is slightly higher as the
breakout nears, but in a rising wedge volume typically recedes.
The Results Snapshot shows some statistics for ascending broadening
wedges. This chart pattern sports a relatively high, 24%, failure rate that
reduces to 6% if one waits for a downside breakout. Failure rates below 20% I
consider to be acceptable, so the rule is, wait for a downside breakout before
selling this one short.
Fullbacks (21%) and throwbacks (7%) are rare, which is somewhat
unusual for broadening formations. The numbers suggest that once a breakout
occurs, there is no looking back; prices break down cleanly and continue down.
This is not strictly true as the chart patterns accompanying this entry attest.
There is some hesitancy when the stock nears the lower trendline.
The 61 % result for meeting die predicted price target is lower than I like
to see in a formation (which is at least 80%). It is even weaker after revealing
the computation. The computation is not based on the formation height as you
would expect. Rather, the target is the lowest point in the formation. The mea
sure rule is similar to the wedge family. Prices decline to, or below, the start of
the formation about twothirds of the time.
A surprising finding for ascending broadening wedges is the partial rise.
A partial rise is when prices lift off the lower trendline but curl around and
head down before coming near the top trendline. When a partial rise occurs, a
downside breakout follows 84% of the time.
Tour
What does an ascending broadening wedge look like? Consider the chart pat
tern in Figure 5.1. The first thing you will notice is the two sloping trendlines;
the top one has a slightly steeper slope than the bottom one. Together, the two
trendlines spread out over time but both slope upward. Once prices pierce the
72

Air Express Intl. Corp. (Air Transport, NASDAQ, AEIQ
Jul92
Figure 5.1 An ascending broadening wedge. Two upsloping trendlines contain
prices that broaden over time.
Barrick Cold (Cold/Silver Mining, NYSE, ABX)
Partial Rise
Tour 75
15
Sep 91 Oct Dec |an 92 Feb Mar Apr May
Figure 5.2 An ascending broadening wedge. The broadening feature is clear in
this chart. The partial rise and failure to touch the upper trendline is a signal of an
impending trend change.
bottom trendline, they drop rapidly. The chart looks like a pieshaped wedge
that slopes uphill. That is why it is called an ascending broadening wedge. The
price action alternates and is contained by the two trendlines. The two trend
lines are not parallel. If they were parallel, you would have a channel or rec
tangle (when both trendlines are horizontal).
Figure 5.2 shows a better example of a broadening wedge, with two up
sloping trendlines where the slope ofthe top trendline is much steeper than the
bottom one.
Figures 5.1 and 5.2 both show a similar situation. The formation appears
at the end of a rising price trend and signals a reversal. Although a reversal is
not always the case, nor is the formation required to be at the end of a rising
price trend, both situations occur more often than not.
Figure 5.2 also shows an interesting pattern that is key in identifying the
start of a new price trend: the partial rise. After touching the lower trendline,
prices again move up but fail to touch the higher trendline. As prices descend,
they pierce the lower trendline and continue moving down. The chart also
shows a similar situation that occurs earlier in the price pattern, around the
start of the new year. There is a rising price trend that fails to touch the upper
trendline. In this case, prices return to the lower trendline, then rebound and
zoom up again to touch the higher trendline. The partial rise fails to predict a
change in trend. We further explore partial rises later.
Why does a broadening wedge form? Pretend for a moment that you are
the head of an investment conglomerate that has big bucks to spend and wants
to buy shares in another company. When the price is low, you instruct your
trading department to begin its buying spree. The sudden buying demand
forces the price to climb even though the trading department spreads its orders
over several days and through several brokers. The trading department tries to
keep its buying quiet, but the word gets out that you are in the market. The
momentum players jump on your coattails and ride the stock upward by buy
ing into the situation. This sends the stock higher than you expected, so your
trading department stops buying for the moment.
Value investors, sensing an overbought situation developing, are willing
to sell their shares at the higher price. Soon the stock is moving down again.
But before it can reach its old low, the buythedip crowd jumps in and halts
the decline. Your trading department, seeing a higher low form, jumps back in
and buys while the price is still reasonable. Some new value investors also
decide the stock is worthy of a flyer and add to the buying pressure. The com
pany itself gets into the act and buys shares authorized by the board of direc
tors as a sharebuyback program announced long ago. The buyback program
is nearing the expiration date and the company feels it is the right time to buy
to complete their promise to shareholders.
The stock makes a new high. When it climbs high enough, die selling
pressure overwhelms the buying demand and sends the shares lower, but the

74 Broadening Wedges, Ascending Tour 75
Air Express Intl. Corp. (Air Transport, NASDAQ, AEIC)
Jul 92 Aug
Figure 5.1 An ascending broadening wedge. Two upsloping trendlines contain
prices that broaden over time.
Barrick Gold (Cold/Silver Mining, NYSE, ABX)
Partial Rise 15
Sep 91 Oct Nov Dec Jan 92 Feb Mar Apr May |un
Figure 5.2 An ascending broadening wedge. The broadening feature is clear in
this chart. The partial rise and failure to touch the upper trendline is a signal of an
impending trend change.
bottom trendline, they drop rapidly. The chart looks like a pieshaped wedge
that slopes uphill. That is why it is called an ascending broadening wedge. The
price action alternates and is contained by the two trendlines. The two trend
lines are not parallel. If they were parallel, you would have a channel or rec
tangle (when both trendlines are horizontal).
Figure 5.2 shows a better example of a broadening wedge, witii two up
sloping trendlines where the slope ofdie top trendline is much steeper than the
bottom one.
Figures 5.1 and 5.2 both show a similar situation. The formation appears
at the end of a rising price trend and signals a reversal. Although a reversal is
not always the case, nor is the formation required to be at the end of a rising
price trend, both situations occur more often than not.
Figure 5.2 also shows an interesting pattern that is key in identifying the
start of a new price trend: the partial rise. After touching the lower trendline,
prices again move up but fail to touch the higher trendline. As prices descend,
they pierce the lower trendline and continue moving down. The chart also
shows a similar situation that occurs earlier in the price pattern, around the
start of the new year. There is a rising price trend that fails to touch the upper
trendline. In this case, prices return to the lower trendline, then rebound and
zoom up again to touch the higher trendline. The partial rise fails to predict a
change in trend. We further explore partial rises later.
Why does a broadening wedge form? Pretend for a moment that you are
the head of an investment conglomerate that has big bucks to spend and wants
to buy shares in another company. When the price is low, you instruct your
trading department to begin its buying spree. The sudden buying demand
forces the price to climb even though the trading department spreads its orders
over several days and through several brokers. The trading department tries to
keep its buying quiet, but the word gets out that you are in the market. The
momentum players jump on your coattails and ride the stock upward by buy
ing into the situation. This sends the stock higher than you expected, so your
trading department stops buying for the moment.
Value investors, sensing an overbought situation developing, are willing
to sell their shares at the higher price. Soon the stock is moving down again.
But before it can reach its old low, the buythedip crowd jumps in and halts
the decline. Your trading department, seeing a higher low form, jumps back in
and buys while the price is still reasonable. Some new value investors also
decide the stock is worthy of a flyer and add to the buying pressure. The com
pany itself gets into the act and buys shares authorized by the board of direc
tors as a sharebuyback program announced long ago. The buyback program
is nearing the expiration date and the company feels it is the right time to buy
to complete their promise to shareholders.
The stock makes a new high. When it climbs high enough, die selling
pressure overwhelms die buying demand and sends die shares lower, but the

price will not drop far—not with everyone trying to buy at a good price. What
you have then are much higher highs from the unbridled buying enthusiasm
and higher, but more sane, lows as your conglomerate and the company itself
try to buy near a fixed low price. You never quite succeed and pay higher and
higher prices as the minor lows move up.
Soon, however, the stock is too pricey even for your tastes. You may even
decide it is time to sell some, or all, ofyour holdings. Meanwhile, the momen
tum players send the stock coasting higher, but this time the price does not
come close to the upper trendline.
Everyone has his or her ear to the ground listening, trying to figure out
what all the buying enthusiasm is about. In the distance, a rumble sounds. The
samestore sales numbers are going to be lower this quarter, the shorts say.
This time the rumor finds sympathetic ears. The rumble heard earlier is the
stampede of die smart money running for the exits. The price drops quickly. It
may hover for a bit around support zones while novice investors, who have not
gotten the word, buy the stock. When they finish placing their trades, the
meager buying demand abates and the stock crashes through the lower trend
line and heads down further.
There are a number of identification guidelines, outlined in Table 5.1, that
make this formation unique. As I discuss the different guidelines, consider the
ascending broadening wedge depicted in Figure 5.3. This formation is differ
ent from Figures 5.1 and 5.2 in that it is born from a region of consolidation.
From the beginning of the study in July 1991, prices move generally horizon
tally and do not fall much below the 153
/g level (Figure 5.3 shows only a por
tion of the prior price action).
The situation changes just before the new year. Prices start moving up on
December 23. They reach a new high in midJanuary but soon move down. At
that point, two tentative trendlines connect the highs and lows. Although it is
too early to form a definitive conclusion, a broadening wedge appears to be
taking shape. After prices move up and touch the upper trendline then pull
back to the lower one again, the broadening wedge formation is clearly visible.
At the start of March, prices move higher but quickly stall, turn around, and
pierce the lower trendline. The partial rise and trendline penetration suggest a
change in trend is at hand.
It is a false breakout. Prices travel higher for 3 days in a tight, narrow pat
tern then zoom upward and touch the upper trendline. The bottom trendline
has to be redrawn to accommodate the slight decline below the old trend
line. Clearly, prices have more work to do before declining below the lower
trendline.
* Table 5.1
Identification Characteristics of Ascending Broadening Wedges
Shape
Trendlines
Touches
Volume
Premature breakouts
Breakout
Partial rise
Looks like a megaphone, tilted up, with price action that
outlines two upsloping, trendlines.
The top trendline has a steeper upward slope than the lower
one and neither is horizontal.
There should be at least three distinct touches (or near
touches) of the trendlines on each side. This helps assure
proper identification and performance of the formation.
Irregular with a slight tendency to rise over the length of the
formation.
Very rare. A close below the lower trendline is usually a
genuine breakout.
The breakout direction is downward the vast majority of
times, but an upside breakout is not unheard of.
Prices touch the lower trendline, climb toward the top
trendline, but fail to touch it. Prices reverse direction and
break out downward from the formation.
Fleetwood Enterprises (Manuf. Housing/Rec. Veh., NYSE, FLE)
Nov 91 Dec |an 92 Feb Mar Apr May Jul
Figure 5.3 A broadening formation that fails to continue moving down, requir
ing a redraw of the formation boundaries. The internal partial rise is rare, occurring
in just 18% of the formations.
77

A month later, prices return to the lower trendline and move higher for a
few days. However, the rise stalls and prices pierce the lower trendline. Like a
replay of the price action a month earlier, prices return to the trendline and
start moving higher. However, this rise falters on low volume and quickly
returns to the lower trendline. When prices gap down on April 16, the genuine
breakout occurs. In rapid fashion, prices plummet to below 123
/4, nearly a 50%
decline from the high.
Looking back at Figures 5.1, 5.2, and 5.3, there are several characteristics
that ascending broadening wedges have in common. The overall shape appears
as a megaphone. This appearance is not uncommon for broadening formations
except that this megaphone tilts upward: both trendlines slope higher. The
upper trendline has a higher slope than the lower one, giving the formation a
broadening appearance. The lower trendline also slopes upward separating
the formation from the rightangle and symmetrical varieties of broadening
formations.
In my studies of ascending broadening wedges, I select formations that
have at least three touches of each trendline (or at least come close). The three
touch minimum helps remove consideration given to normal price action and
helps identify reliable chart patterns.
The volume pattern is irregular but generally rises as prices move up and
recedes as prices decline. Although it is not clear from the charts, volume tends
to rise over time. However, this tendency is slight when considering all ascend
ing broadening wedges and is not a mandatory selection guideline.
Figure 5.3 shows an exceptional premature breakout. Usually, prices fol
low the lower trendline and do not penetrate it until very near the breakout.
When prices do break out, the price action can be messy, as shown in the fig
ure. Sometimes the price runs straight through the lower trendline without
pausing, and sometimes it weaves around the trendline before finally continu
ing down. In either case, the breakout is usually downward.
The partial rise, already mentioned, occurs in nearly half of the ascending
broadening wedges. Since it usually occurs just before a breakout, it is an
important trend change indicator. In a partial rise, prices start moving up, after
having touched the lower trendline, then stop before touching (or coming
close to) the upper one. Prices return to the lower trendline and usually head
lower, staging a genuine breakout. Note that a partial rise must begin from the
lower trendline, not as an upward retrace from the top trendline.
Focus on Failures
Figure 5.4 shows what a failure looks like. Although prices break out down
ward, they fail to continue moving down by more than 5%. The breakout
occurs at a price of 50'/s and prices move to a low of 485
/s about 2 weeks later,
resulting in a 3% decline — too small to register as a success.
Statistics 79
Eaton Corp. (Auto Parts (OEM), NYSE, ETN)
Oct Dec Jan 94 Feb
Figure 5.4 An ascending broadening wedge that fails to continue moving down.
Prices decline less than 5% below the breakout point before moving higher.
Only 7 out of 126 or 6% of the formations breaking out downward fail to
continue moving down by more than 5%. This is an exceedingly low figure.
However, it does not mean that failures do not occur. If you blindly invest in
an ascending broadening wedge, there is a 76% likelihood that the formation
will break out downward. This is a reasonably good number and it improves if
you wait for an actual breakout. Although 76% may head down, that still leaves
24% that either move up or move horizontally. A failure rate of 24% is not
something that you should take lightly. As with most formations, it pays to wait
for a confirmed breakout before investing.
Statistics
Table 5.2 shows general statistics for ascending broadening wedges. The for
mations are more plentiful than other types of broadening chart patterns. I
located 157 in 500 stocks over a 5year period. Out ofthose formations, 118 or
75% act as reversals of the prevailing trend.
The failure rate is comparatively high at 24%, meaning that one out of
every four formations has an upside breakout. If you wait for the breakout to
occur, the failure rate drops to a more palatable 6%. Only 7 out of 126 forma
tions fail to continue moving down by more than 5% after a downside break
out. These 5% failures are rare and not wordi worrying about.

Table 5.2
General Statistics for Ascending Broadening Wedges
Number of formations in 500 stocks
from 1991 to 1996
Failure rate
Failure rate if waited for downside breakout
Most likely decline
Of those succeeding, number meeting or exceeding
price target (measure rule)
Partial rise followed by a downside breakout
Partial rise as a percentage of all formations
Partial rise not at formation end
Percentage with rising volume trend
157
39 consolidations, 118
reversals
38/157 or 24%
7/126 or 6%
20%
10%
73 or 61 %
4 months (116 days)
3 months (95 days)
63/75 or 84%
48%
29 or 18%
59%
Note: Most formations break out downward and a partial rise is often a clue to a downside
breakout.
How far will a stock decline after a downside breakout? On average, the
decline is 20%. However, a frequency distribution ofthe declines tell a slightly
different tale (see Figure 5.5). The vast number ofhits occur in the 10% cate
gory, suggesting that the most likely decline is about 10%. The 15% to 25%
categories have a rather steady but declining trend, leading me to believe the
likely decline is probably understated at 10%. The likely decline is probably
higher, perhaps in the 15% to 20% range.
For a bearish formation, the measure rule usually is the height of the for
mation subtracted from the breakout point. However, for wedges, the price
target is the lowest low in the formation, which is what I use for this formation
too. I compare the ultimate low price with the formation start. Only 61% of
the time is the ultimate low price below the formation low. This is not a very
reassuring figure; I consider values above 80% to be reliable.
I compute the average formation length by subtracting the date of the
formation end from the beginning. On average, the formation is about 4
months long (116 days), and it takes almost as long (95 days) to reach the ulti
mate low.
A frequency distribution of the days to the ultimate low helps determine
in which investment category the formation belongs: short, intermediate, or
Statistics 81
Figure 5.5 Frequency distribution of declines for ascending broadening wedges.
The most likely decline is 10%, followed by the 15% to 25% range.
long term. There are 81 formations with durations less than 3 months and the
remainder of the formations split evenly, at 19 each, for intermediate and long
term, suggesting that the investment implications of this formation are short
term. If you short a stock showing this formation, the odds favor you closing
out your position in less than 3 months.
One set of exciting statistics my research reveals is the behavior ofpartial
price rises. A partial rise, such as that shown in Figure 5.2, occurs nearly half
the time (48%) but when it does occur, it is followed by a downside breakout
84% of the time. In only 18% of the formations does a partial rise within a for
mation occur, such as that shown in Figure 5.3. If you choose not to take
advantage of the partial rise, you are still being alerted to a possible trend
change. Sometimes that in itself is worth knowing, especially ifyou are about
to buy a stock.
The volume trend, as measured by the slope of a line found using linear
regression, indicates 59% of the formations have an increasing volume trend.
This result is weak and suggests volume is probably not meaningful as an indi
cator of the validity of this formation. Observation ofvolume indicates that the
volume series is irregular in appearance.
Table 5.3 lists statistics pertaining to breakouts. Out of 29 upside break
outs, only 1 formation breaks out upward but fails to move higher than 5%.
Downside breakouts (126) are much more plentiful. Even so, there are only
seven failures of downside breakouts. These are 5% failures, meaning that the
stock fails to continue moving down by more than 5% after a downside break
out before recovering substantially. The remainder (2 or 1%) of the formations
have horizontal breakouts.

Description
Table 5.3
Breakout Statistics for Ascending Broadening Wedges
Statistic
Upside breakout
Horizontal breakout
Downside breakout
Throwbacks
Fullbacks
Percentage of breakouts occurring
near the 12month price low (L),
center (C), or high (H)
Percentage loss for each 12month
lookback period
compared with day before breakout
1 or 3%
7 or 6%
29 or 18%
2 or 1%
126 or 80%
2 or 7%
7 days
26 or 21%
9 days
19%, C27%, H64%
L25%, C21%, H19%
165%, 1 34%, 116%, 96%,
114%, 109%
Throwbacks and pullbacks are rare for ascending broadening wedges.
Since there are so few upside breakouts, it is no surprise that there are only two
throwbacks to the top of the formation. The time between a breakout and
completion of a throwback is just a week. Other formations usually have
throwback durations of about 10 to 12 days. Again, the short throwback time
is probably due to the low number of throwbacks.
Pullbacks, at 26 or 21%, are more plentiful but still rare when compared
with other types of formations. The pullback completion time is just 2 days
longer (9 days total) than its throwback brethren. For this formation, the num
bers imply that throwbacks and pullbacks are not reliable and should not be
depended on to initiate a trade. Ifyour stock does pull back, consider adding to
your position once prices begin falling again.
For downside breakouts, I examined where the breakout occurs over the
prior 12month period. The formation appears most often (64%) within one
third of the yearly price high. This may sound surprising since the breakout is
on the bottom of the formation. However, since the wedge rises over time, the
breakout point lifts into the higher ranges. In essence, you will find ascending
broadening wedges breaking out near the yearly high, probably near the end of
an uptrend.
I divided the yearly price range into thirds and distributed the percentage
declines over the range, depending on where the breakout occurs. The largest
" Trading Tactics 83
declines occur in the lower third of the price range with an average decline
measuring 25%. This is larger man normal, but the largest decline occurring
near the yearly low is not surprising. Other bearish formations show this ten
dency. This behavior reinforces the belief that you should short a stock mak
ing new lows, not new highs.
Table 5.3 shows the volume for the week after a breakout as compared
with the day before the breakout. There is high volume on the breakout day,
as expected, and it measures 65% above the prior day (or 165% of the prior
day's total). Throughout the following week, volume remains relatively high,
although inconsistent.
Trading Tactics
Table 5.4 lists trading tactics. The measure rule for this formation is different
from most other formations in that it is based on the lowest daily low, not on
the height of the formation.
Figure 5.6 shows two ascending broadening wedges and application ofthe
measure rule. Both formations are well formed, but the first one has a tendency
to rise slightly above the top trendline before beginning its downhill run. The
chart marks the lowest low in each formation. The low serves as the expected
minimum price move. The formation on the left shows prices reaching the tar
get price in midNovember just as prices turn around and rebound. The for
mation shown on the right has prices hitting the target when they plummet.
The reason for choosing the lowest low in the formation is simply for
performance. Using the formation low as the target price allows 61% of the
formations to achieve the target. This means that almost two out ofevery three
formations decline below the formation low. I like to see performance numbers
Trading Tactic
Table 5.4
Trading Tactics for Ascending Broadening Wedges
Explanation
Measure rule
Partial rise
Use the lowest price in the formation as the minimum
price move to expect. Do not use the formation height
subtracted from the breakout point as the result
overstates the anticipated decline.
Although this formation breaks out downward 80% of
the time, it is best to wait for a downside breakout before
shorting the stock.
If a stock shows a partial rise and begins to head back
toward the lower trendline, consider selling short. A
downside breakout follows a partial rise.
Note: Wait for the breakout then trade with the trend.

Centex Corp. (Homebuilding, NYSE, CTX)
Double Top
Figure 5.6 The measure rule as it applies to two ascending broadening wedges.
Astute investors will recognize the twin peaks as a double top.
above 80%, so the 61% value is poor. As you look at the formation on the
right, you can see how close the price target really is to the breakout price.
If you were to compute the formation height and subtract it from the
breakout point, there would be even fewer formations meeting the price target.
Only 27% would meet the new, lower, price target, which is very different
from 80%. What about computing a target price that is nearer the breakout,
say, using half the formation height subtracted from the breakout? This com
putation only results in a 64% success rate. The number is certainly better than
27% but is not much of an improvement over the 61% number using the low
est formation low.
This formation has a good track record ofdownside breakouts. However,
20% of the time prices either move horizontally or break out upward. If you
wait for a breakout before investing, you substantially increase your chances of
a profitable trade. Once prices decline below the lower trendline, sell the stock
short. Be prepared to cover the short as prices near the target, especially if the
price approaches an area of support.
The most likely decline from an ascending broadening wedge is just 10%,
but it could be substantially higher. In computing the most likely decline, you
can use the measure rule. However, consider that support areas are probably
better places to close out the trade. In Figure 5.6, the lowest formation low on
the left is also a support point. Prices decline to the low in early August then
head up and create the formation. Several months later, prices decline to that
Sample Trade 85
level and turn around. As prices decline after the second formation, the support
level at 36 changes into resistance. During March, prices try to rebound but
turn away near the 36 level.
An exception to the waitforbreakoutconfirmation rule is if a partial
rise occurs. Looking again at Figure 5.6, you might conclude that there is a
partial rise in the right formation. You would be wrong. I define a partial rise
as when prices touch the lower trendline, move up, then return to the lower
trendline. The figure shows prices starting from the top trendline, not the
lower one. Figure 5.2 shows a properly identified partial rise.
If you detect a partial rise, consider shorting the stock. In 84% of the
cases, a downside breakout follows a partial rise. Since you are getting a quicker
jump on the stock, your profits should be larger. When the stock declines to
the lower trendline, move your stoploss order to breakeven. If the stock
should turn around at the trendline and head up, consider closing out your
position.
Sample Trade
Curtis works the night shift at a large bakery near his home. Working at night
frees up the daylight hours for other activities, such as sleep. Occasionally, he
spots a situation such as that shown in Figure 5.6, one that makes the morning
sun seem even brighter.
Each day before he hit the sack, he plotted the stock and watched with
amusement the first broadening wedge form. When die second one appeared,
he wiped the sleep from his eyes and took notice. It was not so much the broad
ening wedge that excited him; it was the wedge coupled with the double top.
Together, they spelled an especially bearish situation, one that he was willing
to shell out his hard earned money to trade.
The day after the stock closed below the lower trendline, Curtis sold the
stock short and received a fill at 39/2. He used the double top measure rule to
estimate his target price. With a top at 453
/t and a valley low of 36'/4, the target
turned out to be 26% (that's 453
/4 36'/4). He decided to place an order to cover
the short at 27'/8, just above the whole number and just above where everyone
else was likely to place theirs. Then he went to bed.
Each day, before he closed the curtains to get some sleep, he would check
on his stock. To him, it was pleasing to see the stock begin moving down
immediately and sailing below the nearest broadening wedge target price
(387
/8). He lost some sleep worrying about the upward retrace in March, and
wondered if the party were over.
Curtis hung in there and the stock eventually pierced the resistance zone
and kept moving down. He hoped that the March resistance zone was just the
corrective phase of a measured move down, which would place the target price
at 221
/2, well below his target at 27'/s. He decided not to be greedy and lower

86 Broadening Wedges, Ascending "
his target. Instead, he moved his stoploss price down to just '/s above the resis
tance zone high at 363
/8.
On April 18 he was rudely awakened from his REM state by a phone call
from his broker. The short was covered at his target price. He got up and
started his computer and checked out the situation. All told, he made about $12
a share. That put a smile on his face and he went back to his dream of telling
his boss what he could do with the night shift
6
Broadening Wedges,
Descending
RESULTS SNAPSHOT
Appearance
Reversal or
consolidation
Failure rate
Average rise
Volume trend
Throwbacks
Percentage meeting
Surprising finding
Price action follows two downsloping trendlines that
broaden out.
Longterm (over 6 months) bullish consolidation
37%
46%, with most likely rise being 20%
Usually increases over time
40%
81%
An upside breakout follows a partial decline 76% of
the time.
What a surprise! When I discovered this formation, I thought it would act like
a falling wedge, but it did not. Falling wedges are reversal formations, whereas
descending broadening wedges act as a consolidation of the prevailing trend.
The volume pattern is also different from falling wedges. In the descending
broadening wedge formation, the volume tends to increase over time but with
falling wedges, it decreases.
87

During my search through the various stocks in the database, I noticed
that this formation appears most often after an uptrend. Prices move down in a
broadening wedge for several months then return to their original direction: up.
The pattern is representative of a trend consolidation, reaffirming the statistics.
How do you measure the failure rate of this formation? Once I knew that
the formation acts as a consolidation of the trend, then trend reversals become
failures. Also, formations that are consolidations but fail to move more than
5% in the breakout direction (after a breakout) are also failures. Altogether,
37% of the formations fail. That is well above the maximum 20% that I con
sider reliable formations to possess.
Being a consolidation in a bull market, the average rise is a very high
46%. The most likely rise is 20% and an astounding 40% of the formations
have gains above 50%! With bullish numbers like these, it makes the failure
rate seem tolerable.
The measure rule stacks up well, but that is no surprise considering the
performance is so good. Unlike ascending broadening wedges, this formation
uses the formation height added to the breakout price to predict the target.
Over 80% of the formations with upside breakouts meet or exceed their price
targets.
One surprise is really not a surprise at all for broadening formations: the
partial decline. When prices move down from the top trendline then rebound,
an upside breakout occurs 76% of the time.
Tour
What does the formation look like? Figure 6.1 shows a wellformed descend
ing broadening wedge. The stock begins rising in June 1994 and rounds over
at the top a year later, in August. In September, the stock starts down in tight
oscillations that broaden over time. A month later, two trendlines drawn across
the highs and lows make the wedge shape clear.
Volume at the start of the formation is well below normal. As the forma
tion develops, volume is erratic, but trends higher. Computing the slope ofthe
volume line using linear regression confirms the result; the slope is positive,
which indicates volume is getting heavier.
In midOctober, prices gap up and shoot above the upper trendline. A
breakout occurs. Volume spikes upward also and continues to be heavy for sev
eral days as prices climb.
As you look at die chart, you may make an interesting observation. The
price trend has three stages: The first stage is the long bullrun fromJune 1994
to August 1995, leading to a consolidation or retrace for 2 months (second
stage), then prices move higher (third stage). In Figure 6.1, the broadening for
mation is a consolidation of the upward trend. Taken as a whole, it looks like
the corrective phase of a measured move up chart pattern.
Cognex (Precision Instrument, NASDAQ, CCNX)
May 95 Jun Nov Dec
Figure 6.1 The descending broadening wedge acts as a consolidation of the
upward trend. Two downsloping trendlines outline price action that broadens
out. Volume usually increases over time.
Canandaigua Brands Inc. A (Beverage (Alcoholic), NASDAQ, CBRNA)
19
May 92 Jun Jul Aug Sep Oct Nov Dec |an 93 Feb
Figure 6.2 The descending broadening wedge formation acts as a reversal of the
intermediateterm downward trend.
89

Contrast Figure 6.1 with Figure 6.2, where a descending broadening
wedge acts as a reversal of the intermediateterm price trend. Prices peak in
May 1992 and head lower. During August, prices begin to broaden out as they
continue their downward spiral. By midSeptember, a descending broadening
wedge forms.
Volume is low at the start of the formation but does have a few spikes.
Into September, volume moves up and becomes even more irregular. At
the start of October, as prices begin moving up, volume recedes. Prices pierce
the top trendline on negligible volume and head higher. A trend reversal
is at hand.
Up to October, prices have been trending down rather steadily, then
prices reverse course and start climbing. ByJanuary prices reach the old high.
During lateJanuary and early February, prices pierce the old high and record
a new one. A year later, prices soar to a high of 32, almost triple the low
achieved during the formation (103
/4).
Table 6.1 outlines the identification guidelines for the formation, and Figure
6.3 shows another example of a descending broadening wedge. Figure 6.3
shows two downsloping trendlines that encompass a series of oscillating
prices. The two trendlines look like a megaphone, tilted down. The chart pat
Table 6.1
Identification Characteristics of Descending Broadening Wedges
Shape The formation looks like a megaphone tilted down. Two
downsloping trendlines outline the price action.
Downsloping Both trendlines slope downward, with the lower
trendlines trendline having a steeper slope. Thus, the two lines
broaden out over time. Neither trendline is horizontal.
Trendline touches The formation requires at least two distinct touches of each
trendline.
Volume Usually rises over the length of the formation. However, the
volume pattern is not a prerequisite.
Breakout Since the formation acts as a consolidation of the trend, if
prices were moving up, they usually continue moving up
after the breakout. If the price trend was down, prices
usually continue down after a downside breakout.
Partial decline For a partial decline, prices must touch the top trendline
and move down, turn around, then head higher without
coming close to the lower trendline. An upside breakout
usually follows a partial decline.
tern is narrow at the start but gets wider over time. Neither trendline is hori
zontal, which is a key consideration since it differentiates this formation from
other types of broadening formations.
There are a number of touches of the minor highs against the top trend
line and the minor lows against the bottom one. There should be at least two
distinct touches—two minor highs and two minor lows—of each trendline to
correctly define a broadening formation.
The slope of the volume trend is usually upward, unlike formations from
the narrowing wedge family: They have volume trends that recede over time.
The increasing volume pattern seems to catapult prices higher, sending them
out the top of the formation. Volume at the breakout is usually high but need
not be. As long as demand exceeds supply, prices will rise.
Since the formation represents a consolidation of the prevailing trend,
prices continue in the same direction they were traveling before the formation
began. As shown in Figure 6.3, if prices are heading up, they will continue
moving up after the formation completes.
Sometimes, however, the formation acts as a reversal of the trend. Figure
6.2 is an example of a reversal. There is no easy way to differentiate a budding
consolidation from a reversal. Both often occur at the end of a rising price
trend. In a consolidation, the breakout is usually upward and prices resume
climbing. In a reversal, the breakout is downward and prices head lower. Only
the breakout direction decides whether the formation represents a consolida
tion or a reversal of the prior price trend.
Alaska Air Group Inc. (Air Transport, NYSE, ALK)
Partial Decline
Followed by
Upside Breakout
Sep 95 Oct
Figure 6.3 A descending broadening wedge as a consolidation of the rising
trend. Volume moves higher even as prices head down. A partial decline signals
that an upside breakout is coming.

The partial decline, such as that shown in Figure 6.3, often indicates an
impending upside breakout. This works quite well for consolidations or rever
sals. However, the success rate appears much lower for downside breakouts
that follow partial rises. I discuss statistics in more detail later, but there are too
few samples to really make a definitive statement about partial rises.
Why do these formations form? The chart pattern, as do many forma
tions, illustrates the struggle between supply and demand. In Figure 6.3, after
attempting to close the September gap in early December, the stock stalls.
Buying enthusiasm dries up and the stock heads lower.
Volume sometimes rises as prices near whole dollar amounts and at 40%,
50%, or 60% retraces of the prior rise or fall. At those points, prices are some
what more likely to stage a rebound. That is what occurs in Figure 6.3. Prices
rise from a low of 13% to a high of 187
/8. A 40% retrace ofthis range takes prices
back to 163
/4. This is quite close to 17, and you can see some hesitation in the
stock at that level. However, once the stock approaches the 60% retrace level
(15%), the smart money knows the jig is up. They start buying heavily for the 5
days surrounding the newyear. Prices halt their decline and move higher. How
ever, sellers are not sitting by idly. They sell into the rally and prices eventually
stall and drift lower, forming the partial decline highlighted in Figure 6.3.
As it happens, volume dries up as prices trace a Vshaped pattern. Low
volume before an upside breakout reminds me of the calm before an approach
ing storm. Formations such as ascending and descending triangles commonly
have low volume just before a breakout. Volume spikes upward the day prices
close above the prior minor high (where the top trendline ends in January).
The momentum players take the upper hand, and prices surge higher on very
high volume.
Focus on Failures
I classify a failure two ways. First, when the formation acts as a reversal—that
is, a failure of the formation to continue the direction of the prior trend. Since
descending broadening formations usually act as consolidations of the prevail
ing trend, a reversal is a failure. Figure 6.2 shows this type of failure.
Figure 6.4 shows a second way for a failure to occur, a formation that acts
as a consolidation of the trend. Before the formation begins, prices are rising.
Immediately after the breakout, prices are also rising. The process starts on
September 3 when prices punch through the top trendline. However, they do
not travel very far before returning to the trendline. I have extended the top
trendline in Figure 6.4 to make the throwback clear. You see that prices ride
along it until they gap down (a breakaway gap) on September 17.
This formation is what I call a 5% failure, that is, when prices breakout
and move less than 5% before reversing course. Failures to act as a consolida
tion of the trend are common; failures to continue moving by more than 5%
Statistics 93
United Technology (Diversified Co., NYSE, UTX)
, Breakout
|un92
Figure 6.4 This formation is a failure according to the 5% rule. Prices fail
away from the formation by more than 5% before moving down.
to move
are not. Only four formations, or 3%, do not pass the 5% rale. The lessons
from this analysis suggest that it is imperative that you wait for a breakout
before placing a trade. Once a breakout takes place, prices will continue mov
ing in the direction of the breakout.
Statistics
The descending broadening wedge is a rare formation, being found in only 101
unique stocks (117 formations) over a 5year span. Of those formations, 66%
act as consolidations ofthe current trend. Table 6.2 shows general statistics for
this formation.
As explained earlier, trend reversals count as failures of the formation to
consolidate, so the failure rate is high at 37%. Also included in the rate are four
5% failures, where prices move less than 5% after a breakout before reversing.
Upside breakouts have an average gain of 46%, which is very high, but it
is not so surprising when you consider that this formation acts as a consolida
tion of the trend. In a bull market, the trend is inherently upward, and this for
mation simply underscores that fact. Still, many bullish formations have gains
of about 40%, so this one scores well.
The most likely rise from this formation is 20%, suggesting that there
is an inordinate number of large gains. The number derives from a frequency
distribution and a review confirms this suspicion. Forty percent of the chart

Table 6.2
General Statistics for Descending Broadening Wedges
1991 to 1996
Failure rate
Average rise of successful formations
Most likely rise
exceeding price target (measure rule)
For successful formations, days to ultimate high
Volume trend
Partial decline followed by upside breakout
117
34 or 37%
46%
20%
47 or 81%
64% have a rising volume trend
35/46 or 76%
Note: The formation acts most often as a consolidation of the longterm bullish trend.
patterns with upside breakouts have gains over 50%. Since large gains tend to
pull the overall average up, I ignore the category and choose the bin with the
next highest total. That turns out to be the 20% category.
I computed the price target using several different rules and discovered
that the traditional way works best for upside breakouts. The Trading Tactics
section explains this more thoroughly, but it involves computing the formation
height and adding the result to the breakout price. This results in 81% of the
formations meeting their target prices (for successful, upside breakouts only).
I consider formations with values above 80% to be reliable.
The average formation length is about 2 '/2 months and results in a long
term move (6 months or longer). This formation is one of the few that have
longterm, bullish implications. The longterm category is a reinforcement of
the tenet that large rises take longer than shorter ones.
The upward volume trend is stronger in this formation than its brother,
the ascending broadening wedge. Almost two out of three formations (64%)
experience heavier volume over time. The slope of a line found using linear
regression on volume determines whether volume is trending up or down over
the length of the formation.
Partial declines and rises are important clues to the performance of this
formation. An upside breakout follows a partial decline 76% of the time (Fig
ure 6.5 shows a good example ofthis).
A partial decline must always begin after touching the top trendline. Prices
then head down toward the lower trendline but do not touch it (or come that
close). The stock curls around and moves back up and pierces the top trend
line, usually immediately. The piercing results in an upside breakout.
Statistics 95
United Illuminating Company (Electric Utility (East), NYSE, UIL)
Partial Decline Followed by Upside Breakout
|an 92
Figure 6.5 A descending broadening wedge with an upside breakout follows a
partial decline 76% of the time. The stock must always touch the top trendline
before beginning the partial decline and must not come too close to the lower
trendline before reversing.
Downside breakouts also follow a partial rise. However, due to the scarcity
ofdownside breakouts and even fewer partial rises (there are only 5 out of 22),
only two partial rises have downside breakouts. The sample size is just too
small on which to base any conclusions.
Table 6.3 shows breakoutrelated statistics. Premature breakouts are
when prices close outside the formation boundary (that is, a close above or
below the trendline) and return to the formation before the genuine breakout.
Due to the broadening nature of these formations, you would expect few pre
mature breakouts. That is exactly what happens. There are only seven prema
ture breakouts, six of them occurring as premature upside breakouts.
Thirtyfour formations with upside breakouts fail. Most of these failures
are due to the formation acting as a reversal instead of a consolidation of the
price trend. Ifthe price trend is downward and the breakout is upward, for exam
ple, then the chart pattern is a failure because it does not act as expected. There
are nine formations with downside breakouts that fail in a similar manner.
Most ofthe formations (79%) have upside breakouts, 19% have downside
ones, and the remainder are horizontal (no breakout for months).
Throwbacks, when compared to pullbacks, are much more numerous at
37 versus 8 simply because there are more upside breakouts. On a percentage
basis, throwbacks and pullbacks are about even at 40% and 36%, respectively,
for upside and downside breakouts. The average time to complete a throwback

Description
Table 6.3
Breakout Statistics for Descending Broadening Wedges
Statistic
Premature upside breakouts
Premature downside breakouts
Upside breakout
Horizontal breakout
Downside breakout
Throwbacks
Fullbacks
near the 12month price low (L), center (C),
or high (H)
Percentage gain for each 12month
lookback period
6 or 5%
Ior1%
34 or 37%
9 or 41 %
92 or 79%
3 or 3%
22 or 19%
37 or 40%
II days
8 or 36%
14 days
L16%, C45%, H39%
L51%, C53%, H44%
136%, 152%, 101 %, 103%, 89%, 109%
Note: Upside breakouts are the most plentiful, occurring 79% of the time.
to the formation top is just 11 days. Fullbacks take 2 weeks, on average, to
return to the lower trendline axis.
Most formations (45%) have breakouts in the center third of the yearly
price range. To determine this, remove any formation with a breakout less
than a year from the start ofthe study. Then sort the breakout price (using the
daily low) into three categories, low, center, and high, according to where it
appears in the yearly price range. The results indicate where in the yearly price
range the breakout is most likely to occur.
Since descending broadening wedges trend downward over time, it
should come as no surprise that relatively few breakouts occur in the highest
third ofthe price range. A formation probably has its start in the highest range
but trends down and moves into the center third ofdie range before the break
out. Mapping performance over the yearly price range shows where the best
performing breakouts occur. The gains are highest in the center third of die
range with scores averaging 53%. This compares with a 46% gain for all
descending broadening wedges with upside breakouts. So ifyou have a choice
between two chart patterns, one with a breakout near the yearly high and one
with a breakout in the center of the yearly price range, go with the center one.
Chances are, it will perform better.
Trading Tactics 97
Volume on the day ofthe breakout averages just 36% above the prior day
(136% of the volume figure). However, a day later, the volume rises to 52%
above the day before the breakout. We have seen this pattern before. Presum
ably, once investors recognize a breakout, they jump on the bandwagon. That
is why the day after a breakout sometimes has a higher average volume.
Trading Tactics
Table 6.4 outlines trading tactics for descending broadening wedges. Use the
measure rule to predict the minimum price to which the stock will move. First,
subtract the lowest price in the formation from the highest. This gives the
height of the formation. Once prices break out upward, add the result to the
breakout price to arrive at the target price.
Figure 6.6 makes die computation of the measure rule clear. The highest
price in the formation is at the start: 471
A. The lowest low is near die forma
tion end at 42]
/s. Add the difference, 5l
/a, to the low of the breakout price
(441
/2) to get the target of 495
/g. Consider the target a minimum price move.
The daily highs and lows used in die height calculation represent the widest
points. I use the daily low at the breakout as a conservative measure.
Table 6.4
Trading Tactics for Descending Broadening Wedges
Measure rule
Partial decline
Trade the trendiines
Stops
Compute the formation height by taking the difference
between the highest high and the lowest low in the
formation. For upside breakouts, add the result to the
breakout price to achieve the target price.
This formation has too many failures (trend reversals) to
risk taking a position before a breakout. Wait for prices to
dose beyond the trendiines before placing a trade.
If a stock shows a partial decline from the top trendline
and begins to head back up, consider going long. An
upside breakout most often follows a partial decline.
If the formation is especially broad, buy at the lower
trendline and sell at the top. If the stock executes a partial
rise and begins falling, close out the position as it may
breakout downward. Alternatively, sell short at the top
trendline once prices are heading down and close the
position after it rebounds off the lower trendline.
For intraformation trading, place a stop on the other side
of the trendiines, just to catch an adverse breakout. Move
the stop as prices cross the formation. Pick areas showing
support or resistance.
Note: Unless trading the trendiines, always wait for the breakout then trade with the trend.

Cat* Corp. (Diversified Co, NYSE, CMT)
Feb95 Mar Aug Sep
Figure 6.6 Calculated minimum price move using the measure rule. Take the dif
ference between the high and low in the formation and add the result to the
breakout price. The target price is the minimum you can expect, and prices reach
the target over 80% of the time.
If you are considering buying this stock, it is not obvious at first that a
broadening wedge is forming. Over time, once enough minor highs and lows
appear, draw the two trendlines. Then, it becomes a matter of waiting for a
breakout. Since most descending broadening wedges break out upward, that is
the way to trade it. However, with a failure rate of37%, you really need to wait
for prices to close above the top trendline before buying.
An exception to this rule is the partial decline. As prices move down from
the top trendline, watch them closely. Ifprices reach a support zone and begin
moving up, buy because that is a partial decline, and it usually signals an
impending upside breakout. Place a stop '/8 below the curl low, just in case.
If the formation is especially wide, try an intraformation trade. Buy at the
lower trendline and sell at the top one, or sell short at the top and cover near
the bottom. With the trendlines sloping downward, a short sale will be more
profitable. Use a stop Vfc beyond the appropriate trendline hi case ofan adverse
breakout. Adjust your stop as prices move in your favor. Place them H below a
support zone (long trades) or above a resistance area (short trades).
Sample Trade 99
Sample Trade
"Do you feel lucky, punk?" Mary growls as she looks at her computer screen
(Figure 6.6). She just finished watching a Dirty Harry movie and is feeling
ornery.
She decides to buy the stock as an intraformation trade once it rebounds
off the lower trendline. When it is clear the stock is climbing again, she pulls
the trigger and receives a fill at 43. Immediately, she places a stoploss order Vg
below the lower trendline, at a price of 42. If things go wrong, she only will
lose 3%. Then she waits.
The stock cooperates by moving higher each day. Soon it is at the top
trendline, and she waits for it to ricochet offthe line and begin heading down.
It does not. Prices close above the top trendline, signaling an upside breakout.
She calculates the price target using the measure rule, 495
/s and that is where
she places her sell order. She raises her stoploss point to 44, '/s below the
minor low in midApril.
As the stock advances each day, she keeps wondering why it has not
paused. She shrugs her shoulders and does not worry about it. When the stock
makes a new high at 473
/8, she raises her stop to 45%, slightly below the two
minor highs in late April and midMay.
In a burst of energy, the stock zooms up over a 2day period and reaches
her sell point. The stock sells at the high for the day, 495
/8. She has cleared over
$6 a share on her trade. Even better, the stock moves lower for several days,
reinforcing her sell decision.
It turns out that she sold too soon, but she does not care. She spots
another promising formation in a stock she has been following for quite some
time. She leans back in her chair, smiles and mumbles something about luck,
then runs to the VCR and plugs in another Dirty Harry movie

7
BumpandRun
Reversal Bottoms
RESULTS SNAPSHOT
Appearance
Failure rate
for upside breakout
Average rise
Volume trend
Throwbacks
Percentage meeting
See also
Looks like a frying pan with the handle on the left
following a trendline down until a large decline
ensues.
Shortterm (up to 3 months) bullish reversal
19%
9%
37%, with most likelyrise being 20%
High volume at formation start, bump start, and
breakout
38%
92%
Cup with Handle; Rounding Bottoms
More than a year after I discovered the bumpandrun reversal (BARR) top, I
decided to look for its complement: the BARR bottom. The reasoning is sim
ple. Many formations, such as double tops, ascending triangles, and triple tops
100
Tour 101
all have bottom versions. Why not the bumpandrun reversal? It never
dawned on me to look for the formation before then. As I searched through the
2,500 years of stock data looking for candidates, I was skeptical that the for
mation added real value. Some looked like cupwithhandle formations with
the handle coming first, whereas others looked like rounding bottoms. Only
after I compiled the statistics did my thoughts change.
If you wait for a breakout, the failure rate drops from 19% to 9%. I con
sider anything less than 20% to indicate a reliable formation. The average gain
is 37%. This is a bit shy ofthe usual 40% for bullish formations, but the most
likely rise, at 20%, is quite good.
The percentage of BARR bottoms meeting the price target is exceedingly
high, at 92%. I consider anything over 80% to be reliable. Of course, the mea
sure rule sets the price target. We see how to do that in the Trading Tactics
section of this chapter.
Tour
What is a bumpandrun reversal, anyway? If I had to name this formation
independent of all others, I would probably call it the frying pan or spoon for
mation because that is what it looks like. However, the formation is just a
BARR top flipped upside down, so I call it a BARR bottom. I guess a more
accurate description is an inverted BARR. Even the word bottom is a mis
nomer since the best performing BARRs appear in the middle of the yearly
price range.
Why do BARR bottoms occur? Like the top version, the BARR bottom
is a study in momentum. Consider the chart shown in Figure 7.1 on a weekly
scale. Since late 1991, the stock was moving sideways—a trading range
between 6'/2 and 11. However, that changed during the last week of October
1993, when the stock moved up and closed higher than the prior week. At first,
this did not seem unusual since many weeks close higher than the prior week,
but this one was different. It initiated a long climb to the highs of earlyJanu
ary. On the highest volume that the stock had seen in years, the stock hit a new
high of 143
/8 during the week ofJanuary 14, 1994. Volume began tapering off,
although it was still high, and prices tagged a much smaller peak during the
week of March 25, at 14. The two minor highs, one in midJanuary and
another in late March, formed the basis of a downsloping trendline.
As the weekly volume trended lower, so did enthusiasm for the stock.
Eventually, bullish sentiment could not sustain the high price and the stock
collapsed. As it headed down, volume continued to taper off. The upward
momentum experienced during the rise to the highs inJanuary was now work
ing against the stock. Over the course of a year, the stock gave back all its gains
and, by midFebruary 1995, it started sinking to new lows.

102 BumpandRun Reversal Bottoms Identification Guidelines 103
CKE Restaurants, Inc. (Restaurant, NYSE, CKR)
ASO N D 94 F M A M J J A S O N D 95 F M AM
9 3 F M A M |
Figure 7.1 Bumpandrun reversal bottom. Upward momentum propels prices
higher during late 1993 then stalls at the start of the new year. Volume tapers off
and prices follow. A cupwithhandle formation or rounding bottom takes shape
and prices climb 350% from the 1995 lows. A channel appears in late 1993 and a
falling wedge in late 1995.
High volume a month later was a key as it signaled a turning point. A
week later, again on high volume, the stock closed higher by over 10%. The
upward move had begun but soon stalled out. The stock moved sideways for
another 2 months, gathering strength for the uphill run. Then it took off, not
jumping up, but slowly moving higher, almost week after week. When the
stock reached the trendline in midAugust, it was clear that it had executed a
massive rounding bottom—a turn in the trend that signaled higher prices.
The stock pushed through the trendline on relatively high volume, then
paused for a month, and formed a falling wedge or pennant. Following that, on
very high volume, prices jumped up to new highs, but this did not last very
long as the stock entered a consolidation phase just below 18. There it stayed
for many months before the stock jumped up and ran still higher. By late June,
the stock had touched 283
/4) a rise ofabout 140% from the breakout, and 350%
from the low.
Many would recognize this formation as a cupwithhandle, and indeed it
is. But it is also a BARR bottom, as a cup does not depend on a downsloping
trendline and a larger handle on the left such as that shown in Figure 7.1.
Whatever you call the formation, the result is still the same: Prices move
higher.
Table 7.1 shows a host of characteristics that correctly identify a BARR bot
tom. Figure 7.2 illustrates the various characteristics outlined in Table 7.1.
Overall, the formation appears as a frying pan. The handle, or leadin phase,
forms a trendline that slopes downward at an angle of about 30 degrees to 45
degrees, sometimes more and sometimes less. Draw the trendline along the
daily high prices as the line signals a buying opportunity once it is pierced.
Unlike BARR tops, sometimes horizontal trendlines in the leadin phase
contain valid BARR bottoms. Such situations are rare, though, and should
probably be avoided. The trendlines in this study are higher on the left and
slope downward—these give the best performance.
Table 7.1
Identification Characteristics of BumpandRun Reversal Bottoms
Frying pan shape The formation looks like a frying pan with the handle on
the left sloping downward to the pan. After a deepening
decline that takes prices into the pan base, prices level out
and eventually soar out the right side.
Downsloping top trendline, The handle forms a downsloping trendline that
leadin height approximates 3045 degrees (but this varies with
scaling). The handle portion of the formation is called the
leadin as it leads in to the bump phase. The leadin
height measures from the trendline drawn across the
highs to the low (not necessarily the lowest low) of the
formation. Select the widest distance from the trendline
to the low, measured vertically, in the first quarter of the
formation. The duration of the leadin should be at least a
month, but varies depending on the situation.
Bump phase The bump is analogous to the frying pan base. The down
sloping trendline deepens to 60 degrees or more. Prices
drop rapidly then level out and turn around, usually
forming a rounded turn. After the turn, prices move up
and sometimes pause at the 30 degree trendline before
moving higher. The bump height, as measured from the
trendline to the lowest low, should be at least twice the
leadin height. Strict adherence to this rule is not
required, but it serves as a good, general guideline.
Uphill run Once prices lift out of the bump phase, they begin an
uphill run that carries prices higher.
Volume Volume is typically high during the three critical parts of
the formation: formation start, bump start, and upside
breakout. However, high volume is not a prerequisite.
Note: The frying pan shape and descending trendline is a study in bearish momentum.

Bethlehem Steel Corp. (Steel (Integrated), NYSE, BS)
'12
Figure 7.2 Various components of a bumpandrun reversal bottom. A price
dropoff follows the leadin phase where prices move in a narrow range. The bump
forms, then rounds upward as prices leave the bowl and move higher on the uphill
run to new highs.
Calculate the leadin height once a trendline forms. Do this by finding
the widest distance from the trendline to the daily low, measured vertically, in
the first quarter of the formation. In Figure 7.2, the leadin height calculation
uses prices onJune 16 (point A). On that day, the low is 175
/s and the trendline
has a value of about 203
/s. Thus, the leadin height is the difference between
these two, or 2 /4. Both the minimum bump height and the target price use the
leadin height, so the calculation is important.
After the leadin phase comes the bump phase. Prices decline rapidly,
although usually not quite as rapidly as that shown in Figure 7.2, and form a
new trendline that slopes down at about 45 degrees to 60 degrees or more.
Volume is noticeably higher at the start of the bump, but selling pressure over
takes buying demand and the truth finally comes out: There are problems with
the company. The stock continues down as the smart money and the momen
tum players leave the stock in droves.
Eventually, downward pressure abates allowing the stock to recover. It
rounds over and touches the original 30 degree trendline. Here, it may move
lower for a while or it may sail right through the trendline. About a third ofthe
time, prices start moving higher, then throw back to the trendline before con
tinuing up.
Volume picks up as prices break out of the formation and move higher.
Rising prices characterize the uphill run phase. Within 3 months, about half
*, Focus on Failures 105
the formations will have hit their ultimate high and started moving down
again.
Focus on Failures
Figure 7.3 shows what a BARR bottom failure looks like. The stock starts its
ascent in June 1994 when it hits a low of 24'/2 (not shown). Although the rise
is not a straightline path, prices reach a new high a year later (the highest peak
on the left). Then it is downhill from there. The decline is quite orderly with
peaks that follow the trendline down. During early September, however, prices
drop rapidly on high volume as the bump forms. Prices quickly reach a low of
335
/8 before rebounding. Having sliced through the trendline and moving just
a bit higher, prices throw back and follow the trendline lower. Prices follow
ing a trendline lower are not unusual, but what is unusual is that prices do not
continue their climb. Instead, they drop offthe end ofthe trendline and plum
met. By late June, they slip to under $20 a share, less than halfwhat they were
at the high.
Why did the BARR fail? This formation is not a perfect example of a
BARR bottom, but few formations are. In this case, the bump height is less
Apple Computer Inc. (Computers & Peripherals, NASDAQ, AAPL)
)un95
Figure 7.3 A bumpandrun reversal bottom failure (less than 20% fail). The
bump height is less than twice the leadin height, a clue that the pattern probably
is not worth investing in. A trendline drawn from point A to B (not shown) satisfies
the bump to leadin height guideline. Using the new line, an investor waiting for
prices to move above the trendline would not buy this stock.

than twice the leadin height. However, this depends on how the trendline is
drawn. Drawing a trendline beginning from the peak at point A, the bump to
leadin height is about 2 to 1 . The new trendline also touches the peak at B. So,
if you wait for prices to move above the new trendline before investing, you
would not purchase this stock. Sometimes it is wise to draw alternative views
just to see how the chart pattern behaves (do not forget to draw the pattern on
logarithmbased scales).
Statistics _
To determine how the average BARR bottom performs, I searched the hun
dreds ofstocks in the database and tabulated their performance statistics. Table
7.2 shows the results. Compared with other formations, the BARR bottom is
plentiful, appearing 360 times in 2,500 years of daily price data. When the
BARR bottom appears, it acts like a reversal ofthe prevailing trend 55% ofthe
time and a consolidation the remainder of the time.
The BARR bottom performs well with 81 % of the formations moving
upward at least 5 % after the breakout. If an investor waits for an upside break
out, the performance is even better, with 91% of the formations moving
higher. This performance corresponds to a failure rate of 19% and 9%, respec
tively. Failure rates below 20% are acceptable.
Once an upside breakout occurs, how high does the price rise? On aver
age, prices climb by 37% as measured from the low of the breakout to the
Description
Table 7.2
General Statistics for BumpandRun Reversal Bottoms
Statistic
1991 to 1966
Failure rate
Failure rate if waited for upside breakout
Most likely rise
Number of rounded appearing bumps
Number of multiple bump formations
360
67 or 19%
32 or 9%
37%
20%
271 or 92%
5 months (157 days)
279 or 78%
52 or 14%
Note: This formation sees prices rise 37% above the breakout, on average.
Statistics 107
ultimate high. This is a very good number but is even higher if you happen
to catch the rise sooner. From the low during the bump phase, the rise is
an astounding 57%. Of course, trying to predict the bottom is exceedingly
difficult and I would not use the 57% figure as anything but a statistical
observation.
Undo influence by a number of large gains can affect the average, so I
created a frequency distribution of gains to check the result (Figure 7.4). The
graph shows that the most likely rise is 20%. The graph also shows the rela
tively high number oflarge gains (41 formations or 14% with gains over 90%).
The many outsized gains apparently pull up the overall average.
If you trade this formation, you might be lucky to have a large gain, but
you probably will not. Instead, expect a more reasonable rise and use the likely
gain of 20% as the benchmark. That is the tallest column in Figure 7.4—the
one with the highest frequency—and it represents the most likely gain.
The measure rule, which predicts the target price, works 92% ofthe time
when using the leadin height as the benchmark. Simply add the largest height
found during the first quarter of the formation to the breakout price. The
result is the target price to which prices rise, at a minimum. Later (see Trad
ing Tactics), I discuss details on how to do this.
Substituting the bump height (the widest distance between the trendline
and price low, measured vertically) for the leadin height, results in a perfor
mance drop to 67% of formations meeting the predicted price. This is still a
good number, but ifyou depend on it as part ofyour riskreward ratio, it might
disappoint when your gains fall short.
>90
Figure 7.4 Frequency distribution of gains for bumpandrun reversal bottoms.
The relatively high number of large gains tends to pull up the average.

It takes about 5 months for the formation to complete, but this varies
widely. At a minimum, each formation is a month long as that is what is
required of the leadin duration. When you add in the bump phase, the dura
tion grows.
Three out of four times (78%), the bump appears rounded. Only 52, or
14%, of the BARR bottoms have dual or multiple bumps. Dual or multiple
bumps are significant declines after the main bump touches the trendline. Most
of the time, the second or later bumps do not decline to the low seen during
formation of the main bump. The actual number of secondary bumps declin
ing below the main bump are few but not tabulated.
Figure 7.5 shows a good example of a multiple bump BARR. The first
bump completes in midAugust 1993 when prices touch the downsloping
trendline. If you purchased the stock at any time during the first bump, you
would have been toast. From the high of 223
/4 on August 19, the stock declined
to 173
/8 on October 1, nearly a 25% fall.
After that, it is all uphill. The stock moves up smartly and crests at 28^2
in midJanuary 1994. From the bump low, that is a 64% move and a 33% rise
from the breakout. Figure 7.5 imparts a valuable lesson: Consider waiting for
the upside breakout before buying into a situation. Not surprisingly, this les
son applies to many formations, not just die BARR bottom.
The dual bump is unusual in that the second bump is lower than the first.
As mentioned, dual bumps are a rarity, occurring only 14% of the time. Of
Inco Ltd (Metals & Mining (Div.), NYSE, N)
May 93 Dec Jan 94
Figure 7.5 A dual bumpandrun reversal bottom. Consider waiting to buy the
stock until after it breaks out upward. Had you bought into this situation during
July, you would have lost money in the short term.
Statistics 109
course, that is scant comfort ifyou already bought into a situation and it begins
declining again.
Table 7.3 shows breakout statistics. Of the 360 formations I examined,
only 9% have upside breakouts with prices that fail to move meaningfully
(more than 5%) upward. I discovered no formations that break out downward,
quickly turn around, and move meaningfully higher. That is not to suggest
such a scenario will never happen; it simply means it did not happen on my
shift. Upside breakouts occur most ofthe time (90%) with downside breakouts
responsible for the remainder.
Once a stock moves above the trendline, it throws back to the trendline
about a third (38%) of the time. To accomplish the maneuver, it takes slightly
less than 2 weeks. Some throwbacks finish within a few days and others take
almost a month before they touch the trendline and start rising again. Excluded
from the tally are throwbacks over 30 days as I consider them to be due to nor
mal price action and not part of a throwback.
Once prices begin moving up, it takes slightly over 5 months (160 days)
to reach the ultimate high. This sounds about right as a 37% average rise does
not happen overnight.
To gauge where in the yearly price range the breakout occurs, I exclude
any formation that occurs less than a year from the start. Then I sort the gains
into three bins each representing a third of the yearly price range. The result
ing frequency distribution shows that most ofthe formations have breakouts in
the center third of the range. Only 20% fall into the high category.
Due to the way the BARR bottom forms, with a downsloping trendline,
one can infer that BARRs start near die yearly high, then move down in price
Description
Table 7.3
Breakout Statistics for BumpandRun Reversal Bottoms
Statistic
Upside breakout
Downside breakout
Throwbacks
Percentage of breakouts occurring near the 12month
Start high to bump low
Start high to breakout low
32 or 9%
325 or 90%
35 or 10%
136 or 38%
13 days
5 months (160 days)
L38%, C43%, H20%
L39%, C42%, H28%
28% decline
17% decline
Note: Nearly all the BARR bottoms break out upward, with over a third throwing back to
the trendline.

to the center range before the breakout. To test this assumption, I compared
the price at which the formations start (the daily high) with the yearly price
range. Over 8 out of 10 formations (82%) have their start at the top ofthe price
range and decline into the center of the range. The implications of this statis
tic are significant: BARRs begin life near the yearly high. After reviewing the
charts, this clearly is the case. BARRs often start just after prices climb to a new
high. Prices drift lower and make another try at a new high but fail. The two
peaks, the first one higher than the second, set the stage for a downsloping
trendline. Prices then move down and a BARR forms when prices suddenly
drop during the bump phase.
Is there a difference in performance among the three price ranges? I
computed the gains for the formations that fall into each of the three bins.
Formations in the center price range have an average rise of 42 %, followed
closely by the lower third at 39%. In last place is the highest third, with a 28%
average gain.
I expected formations in the lowest grouping to place first and they
almost did. I reasoned that the most beatup stocks would rise furthest. On the
other hand, failed momentum stocks, those that are in the highest third of their
price range, would have the most difficult time recovering. That is exactly
what appears to be happening.
How far down from the formation start are the bump low and breakout
points? The statistics say that, 011 average, die low that occurs during the bump
phase is 28% below the high. Since the breakout is above the bump low, it is
still 17% below the start.
Many times, prices rise up to their old highs and stop near that level then
decline. Several formations base their performance on this assumption, includ
ing double and triple tops. I set out to determine if there is any significance to
the rise in the stock after a breakout when compared with the starting point for
BARRs. Figure 7.6 shows the results ofthe analysis. I compared the daily high
price at the start of the formation with the ultimate high after the breakout.
Pictured are the results of a frequency distribution of the percentage gains in
the stock price as compared with the starting point. Positive values mean the
stock climbs above the starting price (on its way to the ultimate high), whereas
negative values mean it falls short. There are a number of points with rises
above 25%, but each bin contains less than half the number contained in the
25% category and are not shown for clarity.
The graph suggests that 34% of the formations stop rising within plus or
minus 5% of the starting price. Almost twothirds (61%) are within 15% ofthe
start. Since these are averages, your results may vary, but it does suggest that a
measure rule based on the attainment of the old high would work well. If prices
rise to the formation starting point (or the nearest high), that may be as far
as they go.
Nine months after reaching a low, the stock in Figure 7.7 climbed to its
highest price but closed lower for the day. For the next 6 weeks or so, it moved
% Difference from Start to Ultimate High
Figure 7.6 Price start versus ultimate high. Comparison between the ultimate
high after a bumpandrun reversal bottom compared with the daily high at the
start of the formation. Does the stock rise up to the old high and stop? A value of
zero means the two points are at the same price level. You can see that many
bumpandrun reversal bottoms have prices that are within 5% of each other.
Feb93 Mar
Figure 7.7 Bumpandrun reversal that stopped rising within 15% of its old
highs. Sixtyone percent of bumpandrun reversal bottoms perform this way.
m

in a descending triangle shape, with lower highs and higher lows, until late
May 1993. Prices moved up marginally higher but were still below the peak in
late April. Then parts began falling off the semiconductor stock. It plummeted
over $4 to close at 233
/4 on June 7. On high but receding volume the price
moved lower until it reached a low of 203
/8 in midJune.
After spending a few days near the low, the stock began climbing: It was
leaving the bump phase. On August 5 it intersected the downsloping trendline
signaling higher prices ahead, and that is just what happened. The price con
tinued rising until it reached 325
/s, just '/4 point below the old high. Its sojourn
there lasted for about 2 weeks before it began a new journey downward. The
decline sent the stock lower and it did not stop until it hit 163
/4, about half its
prior level, in earlyJanuary 1994.
Ifyou used the measure rule (see the Trading Tactics section), you would
have sold at the top, the day after the stock reached 32%. The target price
exactly matched the leadin height added to the breakout price. This fact, cou
pled with matching the old high, reinforced the sell signal. Together, the two
signals would have taken you out of the stock before the 50% decline began,
potentially saving you a bundle.
The last statistics table, Table 7.4, shows the results of the volume study.
I examined the most important points in the pattern: formation start, bump
start, and breakout. Volume at the start of the formation is 20% above the day
before (or 120% ofthe prior figure), on average. This is only marginally higher
than normal and you might not even notice it, since volume tends to bounce
around a lot anyway. However, upon entering the bump phase, when prices
decline drastically, the volume tells a different tale. It rises 75% on the day of
the break and more than doubles (223%) the next day and remains high. In
many of the charts that accompany this chapter, you can see high volume sur
rounding the bump phase. The breakout day also exhibits high volume, but it
quickly tapers off. On the day ofthe upside breakout, volume is 39% above the
prior day, on average, and trends downward from there.
Figure 7.8 shows a typical volume pattern. At the start of the formation,
volume is quite high for a few days but quickly recedes. As prices move down,
Table 7.4
Volume Statistics for BumpandRun Reversal Bottoms
Volume at formation start versus day before
Volume at bump start and next 2 days
versus day before bump start
Breakout day (and succeeding days)
volume compared with day before breakout
120%
175%, 223%, 192%
139%, 130%, 115%, 109%, 99%, 96%
Trading Tactics 113
General Housewares Corp. (Household Products, NYSE, CHW)
14
Note: Volume is heavy during the formation start, bump start, and breakout.
Figure 7.8 Volume pattern. Volume is usually high at the start of the formation,
during the beginning of the bump phase, and during the breakout. Only during
the start of the bump is the volume in this chart muted.
volume follows. Even at the start of the bump, volume is quiet. Once prices
begin descending more rapidly, volume rises a bit overall, especially near the
minor lows (late October and December). During the breakout, volume spikes
upward and propels the stock higher. Enthusiasm is high enough that the stock
gaps upward, not once, but several times. From the bump low, the stock climbs
almost 80%, or 50% from the breakout.
Trading Tactics
Table 7.5 outlines trading tactics for BARR bottoms. After properly identify
ing a BARR bottom, you will want to compute how profitable the formation is
likely to be. You do that using the measure rule. Compute the leadin height
by measuring the widest distance from the trendline to the daily low, vertically,
in the first quarter of the formation. This is before the bump phase. Add the
difference to the breakout price to get the target price. Over 90% of the time
prices reach the target, and so the measure rule should serve as a minimum
move estimate.
The measure rule relies on knowing the breakout price. Even without the
breakout price, you can still see how profitable the formation is likely to be by

Table 7.5
Trading Tactics for BumpandRun Reversal Bottoms
Measure rule Compute the leadin height and add it to the breakout price (use
the daily low to be conservative). The result is the minimum price
to which the stock will rise.
Wait for confirmation Waiting for the breakout improves investment performance. The
close should be above the downsloping trendline before you buy
the stock.
Sell at old high When prices rise to the old high, consider selling if the stock
shows weakness.
Stops Place a stop VB below prior resistance. As prices rise, raise the stop.
Note: Profitabilitiy improves if you wait for breakout confirmation.
computing the leadin height as a percentage of the current price. If the num
ber is small, consider looking elsewhere for a more promising trade.
Ifthe stock looks as ifit might be profitable, then it is wise to wait for break
out confirmation. The confirmation point is when prices rise above the trendline
formed during the leadin phase. Should the price close above the trendline, buy
the stock.
A second corollary to the measure rule is to sell at the old high. I have dis
cussed how often a stock showing a BARR bottom stops near the old high
(which is the start of the formation). Place a sell order near the price level of
die old high. That will keep your profits intact should the stock then turn
down. If you are reluctant to sell your holdings, why not sell half when the
stock reaches the old high, then see what happens?
As always, place a stoploss order Vs below the nearest support zone.
Move the stop upward as the stock advances. That way, when prices turn
down, you will not lose too much. There is nothing worse dian riding a stock
up and following it all the way back down.
Sample Trade
Perhaps the most interesting way to illustrate trading tactics is by example.John
is new to investing and he did not take die time to learn thoroughly about BARR
bottoms. As he flipped through his stock charts one day, he noticed an intrigu
ing situation developing in the stock depicted in Figure 7.9. During August, the
stock peaked, declined a bit, dien formed a second minor high. As the stock
declined from die second high,John drew a tentative trendline down connecting
diem. Soon, he noticed that the stock was descending in a sort of channel. He
drew a second trendline, parallel to die first, that connected the lows.
Sample Trade 115
Aug 91 Sep Oct Nov Dec Jan 92 Feb Mar Apr May Jun
Figure 7.9 A bumpandrun reversal bottom failure in which John invested. He
finally sold the stock just 2 days before it reached its low.
However, the stock soon pierced through the second trendline and
moved lower. When it declined even further, John thought he recognized a
BARR bottom forming. He drew a third trendline, parallel to the other two
and leadin height apart.
As the stock dipped below the lowest trendline, he believed that the
decline was at an end. So, the following day he pulled the trigger and bought
100 shares at 18'/4. He was pleased to acquire the stock a bit below the closing
price for the day.
For the next week, die stock shot upward and pierced the second trend
line. John was brimming with enthusiasm and believed that picking stocks was
an easy game, as he put it. As the stock moved into a consolidation period, John
showed no concern. Flat periods of trading often follow quick rises.
When the stock neared the top trendline, John calculated the target price.
He computed the leadin height by subtracting the daily low from the trend
line at its widest part in the first quarter of the formation. He used the low of
August 20, at 24, and subtracted this from the trendline value of 26, measured
vertically. This left him with a leadin height of$2. John believed that the stock
would likely break out at about 211
A, so this gave him a target price of 23'/4,
which is the leadin height added to the breakout price.
John recalled diat this was a minimum price move achieved by the vast
majority of BARR formations, so he was confident that he could hold out for

larger gains. From his purchase point, he calculated that he would receive at
least a 25% return if everything worked out as planned.
For about a month, the stock moved sideways but this did not alarm him.
He even expected the stock to decline a bit and recapture some of its quick
gains. Secretly, he hoped that the stock would soon break out of its trading
range and head higher. He was confident that it would move up—it was only
a question of when.
He was wrong. Indeed, the stock did break out of its trading range, but
it headed lower, not higher. After it approached the top trendline, the stock
continued down and touched the middle trendline. John knew that a stock
often retraces 40% to 60% of its gains. He grabbed his calculator and com
puted the retrace value.
The stock reached a high of 21.43 in a straightline run from the low at
18, a rise of about 31
/? points. Now the stock was retracing the gains and had
moved down to 183
/4, a 78% retrace. Clearly, this was out ofthe realm of a sim
ple retrace. John suspected that a trend change had occurred, but hoped that
the pause he was seeing as it touched the middle trendline would give the stock
support and call an end to the decline. For a while, it did. The stock paused for
3 days at the trendline then started moving lower again. It quickly fell below
the purchase price and headed down.
Although John had purchased the stock as a shortterm play, he con
vinced himself that he really liked the company and would not mind holding it
for the long term. Now, at least, that is what it would take for him to recoup
his losses and get out at breakeven.
The stock quickly moved down through the third trendline, heading
lower. The easy game was now turning into a disaster. John first considered sell
ing on December 11, when the stock reached 123
/4, for a 30% loss. He delayed
the selling decision by saying that the holding was a longterm one and he
should expect to come across such declines in the short term.
The next day, the stock closed higher and it gave him renewed hope.
Indeed, it closed even higher the following day. But the 2day recovery was an
illusion and the stock declined again. As it plunged below 123
/4, John threw up
his hands and told his broker to dump the dog. He received a fill at 12'A, the
low for the day. Two days later, the stock bottomed out at about 103
/4. From
the buy point, John lost 35%.
As upset as this made John, the stock was not finished tormenting him.
He continued to follow the stock and watched it move higher. He extended the
BARR trendline downward (Figure 7.10) and noticed that a new, larger BARR
had formed. After suffering through the large bump, the stock moved higher
until it touched the BARR trendline. Then, the stock followed it lower, unable
to pierce the resistance line.
Sample Trade 117
Varfty Corp. (Machinery, NYSE, VAT)
Figure 7.10 A bumpandrun reversal bottom on a weekly scale. After the break
out, the stock climbed over 350%.
During the week ofMarch 27, 1992, the stock closed above the trendline
for the first time in months. The BARR was complete and a confirmed break
out was occurring. Did John buy the stock? No. For several months, he
watched its progress as it moved higher almost week after week. Disgusted, he
quit following the stock.
In April 1994, John took another look at the stock and was surprised to
see that it continued moving higher. It had just reached a high of 50'/8, a climb
of almost 370%. He grabbed his calculator and realized that his mistake cost
him gains of over $3,000.
What did he do wrong? Several things. He did not wait for the BARR to
pierce the trendline and move higher. If he had, he would have purchased
closer to the low, saving him precious capital. Next, he did not cut his losses
short. After he bought the stock, he should have determined his sell point. The
middle trendline would have been a good place for a stoploss order. In this
case, it would have taken him out of the stock at about 177
/8, a small decline
from the purchase price of 18'/4. Instead, he followed the stock down and
changed his investment philosophy from a shortterm trade to a longterm
investment.
When he finally gave up all hope of recovering from this trade, the stock
was near the low. Apparently other investors felt the same way as the high

volume peaks during late November and through most of December attest.
That is a common scenario: Novice investors buy near the top and sell near the
bottom, exactly the opposite ofwhat they should be doing.
But there is good news. John has learned from his mistakes. Since that
trade, he has learned to wait for a confirmed breakout before placing a trade
and now uses stoploss orders to limit his downside exposure. Does this mean
he is a model trader? No. Now he is making other types of mistakes.
8
BumpandRun
Reversal Tops
RESULTS S N A P S H O T
Appearance
Failure rate
Average decline
Fullbacks
Percentage meeting
See also
Prices rise steadily along a trendline, bump up, round
over, then decline through the trendline and continue
down.
19%
24%, with the most likely decline between 15% and
20%
39%
88%
Rounding Tops
Ifyou were thinking of buying stock in a company, wouldn't it be wonderful if
you knew the purchase price would be less tomorrow? Of course! But how do
you predict tomorrow's price? That is the question I was working on when I
discovered this formation. I was trying to figure out a reliable way to determine
if tomorrow's price would be higher or lower than today's and by how much.
119

120 BumpandRun Reversal Tops Identification Guidelines 121
I tried all sorts of mathematical games to boost the accuracy of the pre
diction with only limited success. Then I moved to the visual world. I drew a
trendline along a stock chart and wondered ifI could determine how far prices
would decline below the line. I looked at many stock charts and trendlines try
ing to see ifthere was a relationship between a trendline and the breakdown of
the trend. That is when I discovered it: the bumpandrun formation—BARF
for short. I toyed with the idea of leaving the name as is but decided that the
investment community would not believe the veracity of the new formation.
So, I changed the name to bumpandrun reversal (BARR), a slightly more
descriptive and palatable acronym. —
Tour
As I looked at the various trendlines, I discovered that pronounced breakdowns
share several characteristics. Look at Figure 8.1, a good BARR example. The
overall formation reminds me of a mountain range. The foothills at the start of
the formation are low and subdued, not venturing too far above the upsloping
plain. Volume at the start of the formation is high but quickly recedes. The
mountains themselves rise up well above the foothills on high volume. Investor
enthusiasm continues high as prices round over at the top, then diminishes on
the far side. When the mountains end, prices decline sharply and continue
moving down. That is a BARR. Prices bumpup, round over, and run back
down again. The formation is the visual representation of momentum. The
base of the formation follows a trendline that always slopes upward. It signals
investors' eagerness to acquire the stock. As each day goes by, investors bid
higher to reluctant sellers and the price rises.
Other momentum players eventually notice the rise in the stock price.
Many jump on the bandwagon the day after a surprisingly good earnings
announcement. That is when the bump begins. Volume spikes upward along
with the stock price. Quickly rising prices entice others to join the fray and
that, in turn, sends the stock even higher. As momentum increases, prices jump
up to form a new, highersloping trendline. Then things start going wrong.
Upward momentum continues until supply catches up with demand. As
that happens, the rise slows and the smart money turns cautious. Investor
enthusiasm wanes and the war between supply and demand turns. The stock
rounds over and starts heading down. When the smart money sees prices
falling, they sell and the decline picks up speed. Downward momentum
increases and returns prices to the trendline. At this point, buying enthusiasm
may increase and send prices back up for one last try at a new peak. Usually,
however, prices do not bounce off the trendline but continue moving down.
Sometimes there is a pause and sometimes prices just plunge straight through
the resistance line, as illustrated in Figure 8.1.
FieldcrastCannon Inc. (Textile, NYSE, FLD)
Nov93
Figure 8.1 Good example of a bumpandrun reversal. Prices move up along the
trendline in the leadin phase, jump up during the bump phase, then crash down
through the trendline during the downhill run. Volume at the start of the forma
tion and again at the start of the bump is usually high but tapers off as the bump
rounds over. About half the time volume picks up as prices pierce the trendline.
Once prices pierce the trendline, volume increases as investors dump the
stock. This selling alarms more investors and the downward trend feeds on
itself. Eventually, after several months of declining prices, the selling pressure
abates and buying enthusiasm halts the downward slide. Prices tentatively level
out and perhaps even rebound a bit. Once the cause of the reversal fades from
memory, prices start rising again and the process begins anew.
Table 8.1 outlines the various parts of the formation that are illustrated in Fig
ure 8.1. In the figure, a trendline drawn below the lows in the stock extends
until it intersects prices as they decline in May. Volume is high at the start, and
the trend is up. That is a key consideration: Prices must be rising. The trend
line should be approximately 30 degrees, but the degree ofslope depends on the
scaling used to view the chart. If the trendline isflat or nearly so, it is not a good
BARR candidate. A rising trendline shows investor enthusiasm for the stock.
However, the trendline should not be too steep either. Steep trendlines (over 60
degrees or so) do not allow enough room for the bump to complete properly.

Table 8.1
Identification Characteristics of BumpandRun Reversal Tops
Rising trendline
Leadin, leadin height
Rounded bump
Downhill run
A trendline connecting the lows rises steadily: no horizontal
or nearhorizontal trendlines. The trendline usually rises at
about 3045 degrees (although this varies with scaling).
Avoid trendlines that are too steep (over 60 degrees): There
is not enough room for the bump.
The leadin is the section just before prices move up sharply
in the bump phase. Leadin prices should have a range of
at least $1 (preferably $2 or more), as measured from the
highest high to the trendline, vertically, during the first
quarter of the overall formation length. Minimum leadin
length is 1 month with no maximum value.
Prices rise up (trendline slope is 4560 degrees or more) on
high volume usually after a favorable event (unexpectedly
good earnings, an analyst recommends or upgrades the
stock, higher store sales, that sort of thing). Prices
eventually round over and decline back to the 30 degree
trendline. The bump must be at least twice the leadin
height, measured from highest high to the trendline,
vertically.
After returning to the trendline, prices may bounce back up
and form a second bump or slide along the trendline.
Eventually prices drop through the trendline and continue
down.
The first part of the formation, called the leadin phase, leads to the bump
phase. The leadin phase should be at least 1 month long and usually falls in
the 2 to 3month range, but can be considerably longer. Prices oscillate up
and down in this phase and have a range of at least $1 as measured from the
highest high to the trendline. This range, called the leadin height, is calcu
lated using prices from the first quarter of the formation.
Figure 8.1, for example, shows that the highest high during the first quar
ter of the formation occurs on January 12, 1994, at 255
/s. The trendline directly
below this date has a value of about 24%, giving a leadin height of !3
/8. The
height is important because the minimum bump height and target price, cal
culated later, use this value. A more accurate approach is to use the largest dis
tance from the trendline to the high, which is not necessarily found between
the highest high and the trendline. Use whatever method makes you feel
comfortable.
During the leadin phase, subdued price action looks as if the stock is
gathering strength for the bump phase. Prices do not move very far away from
the trendline and usually appear rounded. If you visualize the formation as a
mountain range, the leadin phase represents the foothills.
Volume during the leadin phase is high at the start. Often this is due to
events that occur just before the formation begins. Volume drops off until the
start of the bump, when it suddenly rises. The higher share turnover and
expanding enthusiasm for the stock forces prices up. In Figure 8.1, this price
rise occurs on February 17 and is accompanied by volume that is the highest in
half a year.
Prices jump up at the bump start and quickly rise from a low of 26'/2 to a
high of 34% during late March. Volume remains high throughout this period
then quickly tapers off as prices round over at the top. Many times, the top
takes on the appearance of a headandshoulders formation or a double or
triple top. If you recognize any of these formations on your chart, ignore the
BARR top formation and obey die implications of the individual formations.
The bump height, as measured from the highest high to the trendline,
should be at least twice the leadin height. In this example, the bump height is
8 (that is, 343
/s 263
/s). This is more than twice the leadin height of !3
/s.
The reason for the minimum twotoone ratio is arbitrary. The idea is to
make sure that investor enthusiasm and, hence, momentum are getting carried
away. An upsloping trendline that turns into a bump with a higher sloping
trendline emphasizes the rising momentum. Sustaining such unbounded
enthusiasm for too long is difficult and the stock price eventually declines. In
Figure 8.1, that is exactly what happens. Prices round over and start heading
down. Sometimes the decline is orderly and sometimes it is choppy. In nearly
all cases, prices return to the trendline. Once there, the stock may do several
things. Fairly often prices bump up again, and that is called a BARR with a dual
bump or a dual BARR. Occasionally, a dual BARR consists of several bumps
but the result is still the same. Prices eventually fall below the trendline.
Sometimes prices slide up along the trendline for a month or so before
continuing down. At other times, prices drop straight through the trendline,
turn around and climb again, before ultimately dropping. In a few rare cases,
prices descend from the bump high and never make it back to the trendline
before moving higher. These cases commonly appear on weekly or monthly
price charts.
Focus on Failures
In Figure 8.2, a weekly chart, die first BARR on die left shows high volume
during the initial stages of the bump, as you would expect. The bump height
to leadin height ratio looks good (over 2:1) and clearly investor enthusiasm is
high. However, prices continue climbing instead of rounding over and head
ing down.
Contrast die failed BARR with die one in the center. The middle BARR
has a nicely rounded appearance. The volume pattern is what you would
expect: high at die start, at the start of the bump, and when prices cross die

124 BumpandRun Reversal Tops Statistics 125
Caterpillar (Machinery (const/mining), NYSE, CAT)
9 3 F M A M J | A S O N D
Figure 8.2 A bumpandrun reversal on a weekly chart. The formation on the left
fails as prices climb away instead of moving below the trendline. The rounded
appearing center bumpandrun reversal has good volume characteristics—high
volume at the formation start, bump start, and trendline crossing. However, prices
decline below the trendline just 4%. That is called a 5% failure. The right bump
andrun reversal is a dual bumpandrun reversal formation because prices
approach the trendline in March, form a second peak, then drop below the trend
line.
trendline. However, prices drop below the trendline by just 4%. Any formation
recovering after moving less than 5% below the breakout point is called a
5% failure.
The BARR on the right is a dual BARR. Prices near the trendline in late
March 1994, then just as quickly climb again forming a second peak before
dropping through the trendline. Often the peak of the second bump is below
the first.
On weekly and monthly price charts, you often see prices moving up
steadily over time. However, without the sharp bumpup of prices, the rising
trend should not be labeled a budding BARR. The slope of the price trend
line should rise from about 30% at the start to 60% or higher during the
bump phase.
As you look at Figure 8.2, you might think that many BARR formations
appear as the failure on the left. However, the statistics show that that is not the
case. Ten percent have upside breakouts like the one shown. The other 90%
make it to the trendline and begin moving down. Ofthose heading down, 10%
will move down less than 5% before recovering and moving back up substan
tially. They appear as the BARR in the center and are 5% failures. Together,
the two failure types cause the formation to fail 19% of the time. That is still
below the 20% maximum that I consider reliable formations to possess.
To reduce the failure rate to just 9%, wait for prices to close below the
trendline. Waiting boosts the success rate to 91% but reduces the profit that
you would make ifyou sold near the top. In the Trading Tactics section ofthis
chapter, I show you how to sell near the top before the decline really begins.
That way you can keep more of your profits or make even more by shorting.
Statistics
Table 8.2 shows general statistics for BARR tops. Most of the formations (531
or 82%) qualify as reversals of the upward trend. A substantial number (81%)
perform as expected, which is to say that prices decline below the trendline by
at least 5%. For those formations that succeed, the average decline is 24% as
measured from the breakout point to the ultimate low.
Since averages can be misleading (because a few high numbers can pull
the average up), I graphed the frequency distribution of declines from the bump
high to the ultimate low. Figure 8.3 shows the result. A frequency distribution
gives a good indication of what the most likely decline will be for a stock in
which you invest. The numbers are free of the distorting effects that large
returns can have on averages. Figure 8.3 shows that the most likely decline is
in the 30% to 40% range.
Figure 8.4 shows a frequency distribution of declines as measured from
the high price at the breakout to the ultimate low. I use the high price because it
Table 8.2
General Statistics for BumpandRun Reversal Tops
1991 to 1996
Failure rate
Most likely decline
Number of roundedappearing bumps
Number of multiple bump formations
Higher second (or later) bump
650
531 reversals, 119 consolidations
123 or 19%
24%
15% to 20%
462 or 88%
7 months (213 days)
493 or 76%
166 or 26%
55 or 8%
Note: More than 80% of the formations studied have downside breakouts with an average
decline of 24% below the breakout point.

40 50 60 70
Percentage Decline
Figure 8.3 Frequency distribution of declines measured from bump high to ulti
mate low. Most of the stocks in the study suffer declines between 30% and 40%,
as measured from the highest high during the bump phase to the ultimate low.
Percentage Decline
Figure 8.4 Frequency distribution of declines measured from breakout high to
ultimate low. Looking at the decline after a downside breakout shows additional
declines of 15% to 20% as measured from the high at the breakout (the point clos
est to the trendline) to the ultimate low.
126
Statistics 127
is closest to the trendline on the day of the breakout. Figure 8.4 shows that the
most frequent declines occur in the 15% to 20% range. Depending on when
you take your profits, you can do exceedingly well with this formation. In the
Trading Tactics section, I show you a technique to assist you in selling near the
top instead ofwaiting for prices to plunge to the trendline (the breakout point).
For this formation, I designed the measure rule so that more formations
meet the price target. I use the height of the leadin subtracted from the break
out point to predict the minimum price move. The Trading Tactics section of
this chapter explains this more thoroughly, but the approach means 88% ofthe
stocks meet or exceed their price targets, on average.
BARRs take time to form, about 7 months on average. When they form
and move into the bump stage, the bump appears rounded 76% of the time.
Compare Figure 8.5 with Figure 8.6. The stock in Figure 8.5 has a bump
with a rounded appearance, giving investors plenty of time to sell the stock
near the high. Figure 8.6, on the other hand, shows a chart pattern with a much
narrower peak. Investors had only a few days to catch the top before prices
moved down quickly. The chart in Figure 8.6 is also a dual BARR. There is a
second smaller bump just before prices head below the trendline. Dual or mul
tiple bump BARRs occur 26% of the time with only 8% of second or addi
tional bumps having peaks that rise above the first bump.
MagneTek, Inc. (Electrical Equipment, NYSE, MAC)
Oct 92 Nov Dec Aug Sep Oct
Figure 8.5 A bumpandrun reversal with a rounded bump occurs 76% of the
time, on average. Notice the premature downside breakouts in midApril, a week
or two before the actual breakout. The ultimate low reached in October is at a
price of 121
/4, a decline of over 50%.

Pi irvir^_ar>f4 Diii» Doi/arcral Trr*c
DUII lk/~Cll IU~'iui i " v. v v^i j«i i jfs*>
Figure 8.6 A bumpandrun reversal with a pointedlooking first bump, leaving
investors precious little time to get out of the stock. Many semiconductor stocks
showed similar price patterns in late 1995, setting the stage for an industrywide
downturn. The ultimate low reached in midJanuary 1996 comes after a decline of
nearly 70%.
Table 8.3 shows statistics related to breakouts. Only 9% ofthe formations
have premature breakouts. As you would expect, prices stay at or above the
trendline until the genuine downside breakout occurs. Only 9% ofthe forma
tions studied break out downward and, without declining meaningfully, move
back up. These are 5% failures.
Most of the BARR formations have downside breakouts: 584 or 90%
move lower with only one moving horizontally. Over a third (39%) of the for
mations show a pullback to the formation base. Fullbacks occurring over a
month after a breakout are removed. Such price movements are really changes
in die trend and not pullbacks at all. Fullbacks complete their return to die for
mation base within 2 weeks (12 days), on average.
What is the volume pattern like on the day of the downside breakout?
Comparing the volume the day of the breakout and for the next week with the
volume the day before the breakout, you can see that the highest volume
occurs the day after die breakout (64% above the benchmark, or 164% of the
total). Presumably, once investors notice the stock moving below the trendline,
they sell the next day, sending the volume figure soaring.
On average, it takes 3 months (94 days) for prices to reach the ultimate
low. The ultimate low is the lowest price before a significant trend change
occurs.
Statistic 129
Table 8.3
Breakout Statistics for BumpandRun Reversal Tops
Premature downside breakouts
Upside breakout
Downside breakout
Horizontal breakout
Pullback
Breakout day (and succeeding days)
volume compared with day before
breakout
For successful formations, days to
ultimate low
near the 12month low (L), center (C),
or high (H)
Percentage decline for each 12month
lookback period
58 or 9%
57 or 9%
65 or 10%
584 or 90%
1 or 0%
226 or 39%
12 days "
1 36%, 164%, 126%, 118%, 116%, 115%
3 months (94 days)
LI 2%, C43%, H46%
L46%, C37%, H30%
Note: The most significant declines occur near the yearly price low, but BARRs do not
often occur there.
At what point in the yearly price range do downside breakouts occur? I
removed breakouts that occurred less than 1 year after the start of the study
and compared the price on the breakout day with the range over the prior year.
I divided the yearly price range into thirds and compared the breakout price
with three categories: the lowest third, center, and highest third. The results
show that only 12% break out near the yearly low, 43 % are in die middle, and
the rest occur near the yearly high. These results make sense. The highest
enthusiasm for a stock is when it reaches new highs. Such bullish enthusiasm
feeds momentum and propels the stock even higher. Eventually, supply rises
along with the price and quenches demand. When that happens the stock
rounds over and heads down. Even though the stock declines, there are still
many investors who try buying on the dips or who believe the decline will be
short. They help slow the decline and may even turn it around.
Mapping the performance over the three categories results in some sur
prises. For those formations breaking out near the yearly low, prices decline by
an average of 46%. Breakouts in the highest third of the yearly price range
decline by only 30%. I expected the reverse. I assumed that you would get the
largest declines near the yearly high, not the yearly low.
Upon reflection, this makes sense. Since the BARR is a representation of
momentum, upside momentum carries prices upward. Even after a reversal,

130 BumpandRun Reversal Tops Trading Tactics 131
some investors continue to hope (by buying more shares as prices decline) that
the stock is only retracing its gains and will soon rebound.
When a BARR occurs near the yearly low, presumably the stock is already
in the doghouse. Although some investors are exuberant about the rising price,
once it begins to descend, they quickly run for cover. Bad news follows bad
news and sends the stock down even further.
On the basis of the results in Table 8.3, you could argue that you should
short stocks that appear on the new low list in the newspaper and not on the
new high list. In any case, the largest declines from BARR formations occur
with breakouts in the lowest third of the yearly price range. _____
Trading Tactics
Table 8.4 lists tools to help judge when to sell a stock that contains a BARR as
well as the minimum price decline to expect from such a formation. As you
view your stock charts periodically, some stocks will follow trendlines upward.
These are the ones to monitor closely. Occasionally, one will begin a rapid
climb on high volume and enter the bump phase.
By definition, a BARR is only valid when the bump height, as measured
from the highest high to the trendline, is at least twice the leadin height. Two
Table 8.4
Trading Tactics for BumpandRun Reversal Tops
Measure rule Compute the leadin height (see Table 8.1 for the definition) and
subtract the result from the value of the trendline where prices
cross the trendline moving down (end of the bump). The result is
the minimum price move to expect. Almost 9 out of 10 stocks
meet their price targets.
Warning line Drawn parallel to the trendline and leadin height above it. The
line warns that the stock is making a move and is entering the sell
zone, an area between the warning and sell lines.
Sellline A second trendline parallel to the warning line and leadin height
above it. Consider selling when prices touch the sell line,
especially if the bump is narrow. Delay selling if prices continue
moving up. Draw additional lines parallel to the original trendline
and leadin height above the prior line. When the stock rounds
over and touches the lower trendline, sell the stock.
Sell zone The zone alerts the investor to begin doing research to determine
if taking profits is wise. Since valid bumps always touch the sell
line (by definition, the minimum bump height is twice the leadin
height, and that is where the sell line appears), an investor should
be ready to make a sell decision.
Note: Consider selling when prices rise above the sell line.
lines parallel to the trendline assist in that determination. The first line, called
the warning line, is leadin height above the trendline. A second trendline, par
allel to the first two and leadin height above the warning line, is the sell line.
The warning line serves as a signal that a BARR may be forming. Once
prices move solidly above the line, consider doing any fundamental or techni
cal research on the stock to prepare yourself for a sale.
By the time prices touch the sell line, you should have a firm grasp of the
company, industry, and market outlook. The sell line is not an automatic sell
trigger, but it does confirm that a BARR is present. The sell line touch indi
cates that the momentum players have the upper hand. The game could con
tinue for several weeks or months before the downhill run phase sets in, so do
not be in too much of a rush to sell. Since most bumps appear rounded, there
is ample time to sell the stock. By waiting, you are giving the momentum play
ers additional time to push the stock even higher.
However, there are situations when you will want to pull the trigger
quickly. If the company, industry, or market look dicey, then perhaps it is time
to take profits. You might not be selling at the exact top, but you never go
broke taking a profit. Also, ifthe bump does not appear rounded, then consider
selling. A quick decline often follows a quick rise.
Figure 8.7 shows the BARR trendline and the two parallel warning and
sell lines, each line leadin height from the other. The chart is on a weekly scale
Comsat Corp. (Telecom. Services, NYSE, CQ)
9 2 F M A M ) ) A S O N D 9 3 F M A M | ) A S O N D 9 4 F M A M | J A S O N D 9 5 F M
Figure 8.7 Bumpandrun reversal trendline and two parallel warning and sell
lines. There is plenty of time to take profits in this bumpandrun reversal. The stock
reached a low of 17Yt in December, a 40% decline from the sell point in July. Also
shown in the July to September period is a double top.

and emphasizes the relaxed nature of some BARRs. If you owned the stock
depicted in Figure 8.7 and sold it when prices pierced the sell line moving
down, you would not have sold at the top. However, you would have avoided
the 40% decline that followed. The decline also points out that it can be easy
to make money, on paper, in the stock market but difficult to keep it.
Figure 8.7 also shows the measure rule in action. The measure rule is a
method used to predict the minimum price decline of the stock. For BARRs in
this study, almost 9 out of 10 stocks decline below the predicted price.
To compute the predicted minimum decline, calculate the leadin height
by splitting the formation along the trendline into four equal parts. In the first
quarter of the formation, compute the height from the highest high to the
trendline, measured vertically (or use the widest distance between the two).
Subtract the result from where the trendline is pierced, heading down. In Fig
ure 8.7, the leadin height is 3'/2 (that is, 2l
/i 18). The target price is thus
215
/8 (2S'/8 3!
/2), reached during the week of the breakout.
After the breakout, the stock rises back up to meet the trendline before
resuming its decline. Since a trendline denotes a resistance area when ap
proached from below, it is no surprise prices turn away. Prices form a double
top in the July to September period and plunge downward.
Sample Trade
Jenny is a librarian. Before she goes home at the end of each day, she logs onto
the Internet and checks her stock portfolio. She did not notice it at first, but by
midSeptember, Jenny spotted a BARR forming in a stock she owned (Figure
8.8). She spent an hour searching the Internet for anything she could find
about the company. She checked the fundamentals, analysts' recommenda
tions, insider buying and selling patterns, and anything else she could think of.
She reviewed the reasons she bought the stock. Using the Peter Lynch
style of investing—that of buying a stock one is familiar with—held a special
appeal to her. She liked shopping at the grocery store chain and the products
they sold were something she could really sink her teeth into. She felt com
fortable owning the stock.
Jenny printed out the price chart and examined the BARR in detail. She
drew the trendline along the bottom, divided the length of it into four equal
parts, and computed the leadin height. Then she drew the warning and sell
lines parallel to the trendline, each separated by the leadin height. She com
puted the minimum target price to which the stock was likely to decline. From
the current price of 30, the target price was 23, a decline of almost 25%. Even
though she still liked the stock, such a large decline made her nervous.
She looked back through the chart price history and searched for support
zones so she could better gauge the area where any decline might stop. The
first support area was in the 23 to 24 zone, where a prior advance had paused.
Sample Trade 133
Wlnn Dixie Stores Inc. (Grocery, NYSE, WIN)
|un 92 Jul Aug Sep Oct Nov Dec |an 93 Feb Mar Apr
Figure 8.8 Detailed bumpandrun reversal with sell lines, jenny raised her sell
point as the stock climbed. Eventually, she sold the stock the day after it pierced a
lower sell line.
Interestingly, that was also the predicted decline point for the stock. If the
stock fell below the support point, she noticed a second, more robust support
area between 20 and 22.
What of the possible reward? How high could she expect the stock to
rise? Longterm price charts were no help as the stock was making new highs
almost daily. Jenny shrugged her shoulders as there was no way to determine
where the rise would stop. Her only guess was that it might pause at 3 5, 40, or
45, price points where investors might decide to sell. Any one of those points
could turn the stock downward, she decided. Even the current 30 level might
be the highest price the stock sees.
After her analysis was complete, she was still confident that the stock held
promise of additional gains. As with any stock caught by upward momentum,
there was no telling how high the stock would climb before it stopped. She
decided to hang on to the stock. If the stock declined to the warning line, she
would sell it. She placed a stoploss order at 27'/2, the current value of the
warning line.
During late September and into the start of October, the stock followed
the sell line upward. On October 12, the stock jumped upward again. After a
week or so, Jenny was able to draw another sell line parallel to the original
BARR trendline that intersected stock prices. She decided that should the stock
fall to the lower sell line, she would dump the stock. She raised her stoploss
point to 31. But the stock did not return to the lower sell line.

The stock reached a minor high of 343
/8 on October 19, then retraced
some of its prior gains. It curled around and reached a low of 327
/8 before turn
ing around. Jenny printed out another price chart and drew a new trendline.
This line had a slope of about 60 degrees. She smiled as the BARR was per
forming exactly as predicted.
During the first part of December, prices pierced the 60 degree trendline
when the stock began moving sideways. Jenny suspected that the rise was
nearly over, but one could never tell for sure until it was too late. She decided
that should the stock decline below the latest sell line, she would close out
her position.
The stock moved up again. A few days after Christmas, the stock reached
a new high of 393
/4 and Jenny was able to draw another sell line. During the
next 2 weeks, the stock declined to the lower sell line, then rebounded to chal
lenge its recent high. OnJanuary 15, it peaked at 397
/s, a smidgen below the 40
resistance number she estimated earlier.
ToJenny, the day looked like a oneday reversal, but she could not be sure.
Taken together, die two highest points looked like a double top but the reces
sion between them was not deep enough to qualify and the two peaks were a bit
too close together. Still, it was a warning sign and it made her nervous.
Less than a week later, the stock declined below the lower sell line.
Should she sell or hold on for additional gains? She looked back at the profit
she had made so far and decided not to be greedy. She sold the stock at 363
/4 on
January 22. The next trading day, the stock closed up l'/4 at 38, and she was
crestfallen.
She continued to monitor the stock and watched it hesitantly move
higher over the next 2 weeks. She tried to take solace in the large profit she
achieved, but it was little comfort in the face of missed gains. Did she sell too
soon? Should she have held on? On February 23, her questions were answered
when the stock dropped below her sell point, heading down.
Jenny watched the stock drop to 35 and find support at that level. Then,
it continued moving down. In early April, the stock declined below the origi
nal trendline and she calculated the minimum target price of 31. This was
reached within the week and the stock continued falling.
She turned her attention to other interesting situations and forgot her
trade untilJuly 1994. By chance, she pulled up a chart ofthe company and was
horrified to see that the stock had declined to a low of about 21, almost a 50%
decline from the high.
9
Cup with Handle
RESULTS SNAPSHOT
Appearance
Failure rate
Failure rate if
waited for breakout
Average rise
Throwbacks
Percentage meeting
Surprising finding
See also
Looks like a cup profile with the handle on die right
Shortterm (up to 3 months) bullish consolidation
26%
10%
38%, with most likely rise between 10% and 20%
74%
49% using full formation height, 73% using half
formation height
Cups with a higher right lip perform better, 40%
versus 35% average gain.
BumpandRun Reversal Bottom; Rounded Bottom
Shown above are the important statistics for the cupwithhandle formation.
The failure rate is 26%, above the 20% that I consider acceptable. However, if
you wait for an upside breakout, the failure rate drops to 10%. The average gain
is 3 8%, which is good, but below the 40% garnered for most bullish formations.
The most likely rise ranges between 10% and 20% and it is evenly distributed.
A closer examination ofthe results shows that 39% ofthe formations have gains
135

136 Cup with Handle Identification Guidelines 137
less than 20%, whereas the outliers, those with gains over 50%, represent 27%
of the formations.
Throwbacks after an upside breakout occur in nearly three out of four
chart patterns. This suggests an interesting way to play the cupwithhandle
pattern: Wait for the throwback before buying the stock. This technique also
boosts the average gain by a small amount. I discuss this further in the Trad
ing Tactics section of this chapter.
The percentage of formations meeting the predicted price target is a
pathetic 49%. Most measure rules, which predict the target price, involve com
puting the formation height and adding the result to the breakout price. With
this pattern, the cup height can be considerable so it should come as no sur
prise that only half the formations meet their price targets. I explore changes
to the measure rule in the Trading Tactics section that improve the price
prediction.
A surprising finding that is discussed in the Statistics section is a ten
dency for cups with a higher right lip to outperform those with a higher left lip.
The difference, an average gain of 40% versus 35%, is statistically significant.
Tour
The cupwithhandle formation was popularized by William J. O'Neil in his
book, How to Make Money in Stocks (McGrawHill, 1988). He gives a couple of
examples such as that shown in Figure 9.1. The stock climbed 345% in less
than 2 months (computed from the right cup lip to the ultimate high). This was
the best performing stock in this study. Unfortunately, it does not meet
O'Neil's criteria for a cupwithhandle formation. I discuss my interpretation
of his criteria in a moment, but let us first take a closer look at the chart pat
tern. The stock began rising in early August at a price of about 5'/2 and climbed
steadily until it bumped up in early December. Volume, incidentally, was very
high for the stock at this stage. The stock climbed robustly then rounded over
and plunged back through an earlier trendline. It completed a bumpandrun
reversal (BARR). During its climb, the stock reached a high of 267
/s during late
December and a low of 123
/s after the BARR top—a loss of 54%. The rise and
decline formed the left side of the cup. Over the next 2 months, prices mean
dered upward and pierced the old high during late March. The rise to the old
high completed the right side of the cup.
Profittaking stunted the climb and prices moved horizontally for almost
2 weeks before resuming their rise. This formed the cup handle (incidentally,
the handle in this formation is a high, tight flag formation). Volume during
formation of the handle was down sloping—higher at the start and trending
lower. When prices rose above the cup lip, a breakout occurred. This accom
panied a surge in volume that propelled prices higher. However, a week after
the breakout, prices threw back to the handle top before continuing upward.
Figure 9.1 Bumpandrun reversal that leads to a cupwithhandle formation.
Note the price scale as the breakout occurs at about 30 and the stock climbs to
120 in less than 2 months. The cup handle is a high, tight flag formation.
This throwback allowed nimble investors the opportunity to enter long posi
tions or add to existing ones. By late May, just 44 days after the breakout, the
stock reached the ultimate high of 120.
In the study of chart formations, when I search a database for various patterns,
I ignore most conventional selection criteria. I let the formations determine
their own characteristics. That is the approach I used in selecting the cupwith
handle formation. After making my selections, I sorted the database according
to my interpretation of O'Neil's selection guidelines and compared the per
formance. Table 9.1 shows the O'Neil selection criteria, the guidelines I used
to select formations, and my interpretation of his criteria.
O'Neil outlines many selection guidelines in his quest to find suitable
cupwithhandle patterns. He found that the performance of a stock, relative to
the performance of other stocks, is important. A stock with improving relative
strength helps weed out underperforming situations.
I do not know which stocks he used to compute his relative strength char
acteristics, so I did not use relative strength as a selection rule. If I had, the
number of stocks meeting O'Neil's guidelines may have decreased further (just

Table 9.1
Two Different Approaches to O'Neil's CupwithHandle Pattern
O'Neil Criteria
Unfiltered
Selection
Guidelines
Filtered Selection
Guidelines: The O'Neil
Interpretation
Improving relative strength
Substantial increase in volume
during prior uptrend
Rise before cup is at least 30%
Ushaped cup
Cups without handles allowed
Cup duration: 7 to 65 weeks
Cup depth: 12% or 15% to
33%. Some decline 40% to
50%
Handle duration: usually at
least 1 to 2 weeks
Handle downward price trend
Handle downward volume
trend
Handles form in upper half of
cup
Handle forms above 200 day
price moving average
Handle price drop should be
10% to 15% from high unless
stock forms a very large cup
High breakout volume, at least
50% above normal
Saucer with handle price
pattern has more shallow cups
None suggested
None
None
Same
Same
Cups must have
handles
Same
None
1 week minimum
None
None
Selected if handle
toofo like it formed
in upper half; 16%
failed but were
close and used
anyway
None
None
None
None
Cup edges should
be at about the
same price level
None, information unavailable
Very high volume during rise
to cup
Same
Same
Cups must have handles
Same
12% to 3 3%
1 week minimum
Same
Same
Same
Handle low above 200 day
moving average
Handle decline from right cup
high to handle low must be
15% or less
Breakout volume at least 50%
above 25day moving average
No distinction made
Cup edges should be at about
the same price level
Note: The best performance comes from the unfiltered selections in the center column.
The word same refers to the guideline shown in the left column.
138
9% meet his guidelines as it is) and the impact on performance is unknown.
For those stocks that meet his selection criteria, I looked at each cupwith
handle formation and verified that there is very high volume somewhere on the
rise leading to creation of the cup. Of the formations obeying the O'Neil cri
teria, only one successful formation was excluded as a result.
As I was selecting cupwithhandle formations, it became apparent that
locating cups during an uptrend is important. So, I adopted O'Neil's criteria of
a minimum 30% rise leading up to the cup.
All the cups are Ushaped (Vshaped ones being removed). Also removed
from the study were cups without handles. To me, a cup without a handle is a
rounding bottom. I discuss rounding bottoms in Chapter 34.
I use a strict interpretation of O'Neil's cup depth. A maximum depth of
50%, although increasing the number of cups meeting his guidelines, raises the
failure rate along with the average gain. However, neither the failure rate nor
the average gain changes significantly, so I used a range of 12% to 33%.
O'Neil specifies a number of guidelines for the handle. He says the han
dle should be a minimum of 1 to 2 weeks long, but does not set a maximum
duration. In my observation of the formation, prices can, and often do, move
horizontally for several months before staging a breakout.
As I examined each chart for the pattern, I eliminated those with handles
that form well below the midpoint. However, I was not concerned ifmy casual
observations included a few cups with handles that fell below the center.
Removing all handles that are lower than the cup midpoint boosts the average
gain just 1% to 39%.
As you can see from Table 9.1,1 ignore many ofthe O'Neil criteria when
selecting cupwithhandle formations for further analysis. Once I collected the
chart patterns, I filtered the cups through my interpretation of the O'Neil cri
teria. This grouping of methods, unfiltered and filtered, yielded two sets of
performance statistics. But first, let us take a look at a few examples of cup
withhandle patterns.
As mentioned before, the cup pattern shown in Figure 9.1 does not meet
the O'Neil criteria. Why? The cup depth, at 54%, is too deep to qualify. Addi
tionally, the handle price trend is upward, not downward. Price and volume
trends were evaluated using linear regression from the day after the right cup
lip to the day before the breakout. Excluding those 2 days helps remove possi
bly large price moves. I used closing prices in the calculation on the remaining
data (for the handle price trend).
Figure 9.2 shows another good example of the cupwithhandle pattern.
The cup gently rounds over and climbs just beyond the old high then pauses.
Prices drift down in the handle, along with a downtrending volume pattern
before the breakout. Then volume surges and prices move smartly upward.
Two days after the breakout, prices move marginally lower again and enter the
region of the right cup lip. It is a brief throwback and prices are soon on their
way again. Less than 2 months later, die stock tops out at 15l
/2 for a rise of22%.

140 Cup with Handle Genetech, Inc. (Drug, NYSE, CNE)
Mentor Graphics (Computers and Peripherals, NASDAQ, MENT) _ 1 -/
I Next Cup
Apr 93 May Jun |ul Aug Sep Oct Nov Dec |an 94 Feb
Figure 9.2 A cupwithhandle pattern. The cup and handle are shaped nicely
with the right cup lip slightly higher than the left.
91 A S O N D 9 2 F M A M AS OND
Figure 9.3 Cupwithhandie pattern on a weekly scale. The failure at 10% to 15%
above the breakout is quite typical for this formation. However, this stock recov
ered and continued upward.
If you look on either side of the cup in Figure 9.2 you will find two addi
tional cups (portions of which are shown) that fail. The one on the left breaks
out downward and the one on the right fails to continue rising by more than
5%, so it too is a failure. Only die center cup works as expected but even it
shows muted gains. We see in the Statistics section that a significant number
of cupwithhandle formations fail to rise very far.
Figure 9.3 also shows a cupwithhandle formation but on the weekly
time scale. When I was searching for the various formations, I found that
weekly scales provide an easy way to identify many of the formations. Of
course, I also looked at daily price data to refine the weekly patterns and iden
tify new formations that I may have missed.
The chart in Figure 9.3 shows an example of a cupwithhandle formation
in which the rise falters after rising just 11%. Fortunately, after declining back
to the handle base, the stock recovers and goes on to form new highs. Ulti
mately, the stock gains 52%.
Figure 9.3 also highlights an incorrectly selected cup: an inner cup. There
is no 30% rise leading up to the formation (since prices are trending downward)
and the handle lasts just 2 days. However, inner cups offer wonderful trading
opportunities as they allow you to get in on the ground floor of an impending
rise. Even if prices only rise to the height of the outer left cup lip, the move can
be significant. A discussion of trading tactics occurs later in this chapter.
Figure 9.4 shows another example of an errant cup selection. The rise
from point A to point B is less than 30%. Had you invested in this pattern after
Adobe Systems (Computer Software & Svcs., NASDAQ, ADBE)
Invalid Cup Selection
Apr 93
Figure 9.4 An invalid cupwithhandle pattern. The rise from point A to point B
is less than 30%. The two outer peaks (in June and March) do not create a cup
either because the handle drops down too far (point C)—well below the cup mid
point.
141

142 Cup with Handle
prices rose above the cup lip, you would have seen the stock climb to 341
/2, an
increase of just 11%. After it reached the high, the stock plummeted. In less
than a month, prices declined to 2ll
/i, a loss of 38%.
Focus on Failures
The cupwithhandle formation, like most formations, suffers from two types
of failures. The first type of failure is shown in Figure 9.5. The cup formed
after an extended rise that began in midDecember 1993 at I5l
/i and rose to the
left cup lip at 243
/s. Prices quickly reversed course and moved lower, then
became choppy as they traversed the cup bottom. Once an upward trend was
underway, the choppiness smoothed out and prices soared to the right edge
of the cup.
The day before prices reached a new high, high volume soaked up the
demand for the stock. The stock coasted to a new high the next day then
moved lower. A handle formed about a week later in the 222 3 range. The vol
ume trend during this time was sloping downward, as you would expect. How
ever, prices dropped through the handle low, pulled back into the handle, then
dropped away again. From that point on, it was all downhill. A low reached in
late January at H1
/: shows a decline of 42% from the cup lip.
|an 94 Feb Mar Apr May |un Jul Aug Sep Oct Nov Dec
Figure 9.5 A cupwithhandle formation that breaks out downward. It should serve
as a reminder to always wait for the breakout to move above the cup lip before buy
ing the stock. This cupwithhandle formation turned into a double top.
As you look at the cupwithhandle formation in Figure 9.5, you see lit
tle that is out of the ordinary. The right edge ofthe cup is somewhat above the
left edge. Minor differences between the two cup edges are normal. Sometimes
the left edge is higher and sometimes the right one is higher.
Volume during formation of the cup is about average for this stock.
Instead ofa cupwithhandle formation, what you really are looking at is a dou
ble top. The two widely spaced peaks, the first in March and the second in
August, predict a decline in the stock.
The vast majority of cupwithhandle formation failures break out down
ward. Of the formation failures identified in the database, 74 have downside
breakouts and only two ofthose turn around and finish higher by more than 5%.
The second type of failure is the inability of the stock to rise by at least
5% before declining. Figure 9.6 shows this situation. The nicely shaped cup
forms after an extended price rise from 33 to 45. The two cup edges are at
about the same price level. The handle seems to form a small cup of its own.
Prices move up sharply in late September and break above the right cup lip and
continue higher, but only briefly. The stock tops at 477
/8, moves horizontally
for about 3 weeks, then starts down. Two months later, the stock hits a low of
375
/g. The rise after the breakout is slightly less than 5%. I classify as a failure
a stock that does not continue moving more than 5% in the direction of the
breakout. Of the formations with upside breakouts, only 10% or 30 failed
because they did not continue climbing by more than 5%. It seems that once
an upside breakout occurs, prices generally continue climbing—at least 5%
anyway.
Figure 9.6 A cupwithhandle formation 5% failure. Although prices break out up
ward, they move less than 5% away from the cup lip before plunging downward.

144 Cup with Handle V
Statistics
As mentioned earlier, I first selected the cupwithhandle formations then fil
tered out those chart patterns that did not obey my interpretation of O'Neil's
criteria. Listed in Table 9.2 are the results.Just 9% ofthe selected patterns met
his selection criteria (outlined in Table 9.1). Of those meeting the criteria,
only 62% perform as expected. That is to say, 38% either break out downward
or fail to rise by more than 5% before reaching the ultimate high. The average
rise of successful formations is 34%, but the most likely rise is just 15%. Since
only 23 formations were successful (a small sample count), the frequency dis
tribution used to determine die mosu likely rise is suspect but it does agree with
the unfiltered (nonO'Neil) formations.
O'Neil suggests that there should be a substantial increase in volume
somewhere during the rise to the cup. I excluded only four formations because
Table 9.2
Statistics and Results for CupwithHandle Formations Filtered by Many
of O'Neil's Rules
Description Statistic or Result
Number of formations
Failure rate
Is substantial increase in volume during
prior uptrend important?
Is cup length from 7 to 65 weeks
important?
Is cup length related to ultimate gain?
Does cup depth (12% to 33%)
improve performance?
Is a handle duration (1 week
minimum) important?
Do downsloping handle price trends
improve performance?
Does a downsloping handle volume
trend improve performance?
Is it important that a handle be above
cup midpoint?
Is handle low above 200day moving
average important to performance?
Is it important that handle low is 15%
or less from cup lip?
Is high breakout volume important to
performance?
Is cup depth related to ultimate gain?
Statistics 145
37 out of 391
14 or 38%
34%, but most likely rise is 15%
No. Only 4 formations were excluded
(if included, performance deteriorates).
Unknown. All selected cups fall in this
range.
No. Relationship is random.
No. It is detrimental and limits
performance.
Unknown. Cups with shorter handles
were eliminated.
No. They are detrimental.
No. It is detrimental.
Yes, but most formations were
selected with this in mind.
No. Has no bearing on performance.
No. Has no bearing on performance.
Yes, but only minimally.
No. Relationship is random.
of this item (three formation failures and one 22% gain). However, including
all four formations would increase the failure rate. I did not include this
requirement in the unfiltered cup selections.
Is the cup length important to performance? All the cupwithhandle pat
terns I selected fall within the range of 7 to 65 weeks. However, I did a scatter
plot to determine if shallow cups perform better than deep ones. The plot sug
gests die relationship is random.
Does cup depth improve performance? Placing specified limits on the cup
depth is detrimental to performance. Changing the minimum depth to 30%
from 12% and the maximum depth to 55% from 33% improves both the fail
ure rate 'to 33% from 38%) and the average gain (to 36% from 34%). I chose
not to specify any cup depth in the unfiltered cup selections.
All the cupwithhandle formations I chose had handles that were at least
1 week long. Snorter handle cups were removed from the statistics so it is not
clear if this rule is important.
Rules regarding downsloping price and volume trends were found to
limit performance. Removing the two rules decreases the failure rate and
improves the average rise. These two factors are primarily responsible for the
relatively poor performance of the filtered cup selections.
Is it important how low a handle goes? Yes, and no. I found it important
that the handle remain above the midpoint (although most cups were selected
with this in mind, so a fair analysis cannot be performed) but it is irrelevant that
it remain above the 200day price moving average or within 15% from the
right cup lip.
Is breakout volume important? Removing this selection rule hurts per
formance but the change is slight.
O'Neil made a distinction between cupwithhandle patterns and saucer
withhandle patterns. I believe that the difference between the two is only a
matter ofcup depth, already addressed by the cup depth rule. However, I won
dered if cup depth played an important part in determining the average gain.
A scatter plot of the cup depth versus percentage gain for unfiltered cups sug
gests the relationship is random. The scatter plot for the filtered cups also sug
gests the relationship is random, but samples are too few to be meaningful.
Table 9.3 shows statistics for the unfiltered cupwithhandle formations.
Only those rules outlined in Table 9.1 for unfiltered selections apply. There are
391 formations identified, with 302 ofthem acting as consolidations ofthe pre
vailing trend. Eightynine formations act as reversals, with the vast majority of
them being failures to perform as expected. The failure rate is 26%, well below
the 38% found for the filtered variety. The failure rate improves dramatically
(to 10%) if you wait for the upside breakout before investing.
The average rise at 38% handily beats the filteredcup average rise of
34%. However, the most likely rise is between 10% and 20%. Figure 9.7 shows
that most likely gain splits evenly between the first three columns (10%
through 20%). The tallest column suggests that large gains skew the average

Table
General Statistics for Unfiltered
Description
1991 to 1996
Failure rate
Most likely rise
exceeding price target (measure rule
using full cup height)
exceeding price target (measure rule
using half cup height)
Do short handles show larger gains?
Does a higher right cup lip mean larger gains?
9.3 N
CupwithHandle Formations
Statistic
391
1 02 or 26%
30 or 1 0%
38%
1 0% to 20%
151 or 49%
223 or 73%
7 months (208 days)
Yes, but relationship may be weak
Yes, with gains of 40% versus 35%
Note: Both the failure rate and average gain improve over the filtered variety.
rise upward. A quarter of the formations have a rise of less than 15%, whereas
half the formations top out with less than 25% gains. The message from the
frequency distribution graph is clear: If you think investing in a cupwith
handle formation will yield outsized gains, you should think again. The statis
tics suggest that you have only a 27% chance of selecting a cup formation that
yields a gain over 50%. The chances of doubling your money are just 7%.
A discussion of the measure rule occurs in the Trading Tactics section,
but it involves adding the formation height to the breakout price (the right cup
lip is the breakout point) to predict the target price. The target price serves as
a minimum price move. Only 49% of the formations climb to the target price.
If you use half the cup height in the calculation, then 73% of the formations
reach their price targets.
The average formation length is 7 months, as measured from the left cup
lip to the breakout.
O'Neil suggests that the handle should be at least 1 to 2 weeks long but he
did not specify a maximum length. I looked at a scatter plot of handle length
versus percentage gain for successful formations (see Figure 9.8). There are a
few formations with short handles (about 3 weeks) and large gains (over 200%)
and there are longer handles with shorter gains.
Statistics 147
Figure 9.7 Frequency distribution of gains for cupwithhandle pattern. The most
likely rise is between 10% and 20%, but the outsized gains over 50% distort the
average.
Figure 9.8 Handle length versus percentage gain. Cups with shorter handles
seem to perform better but the relationship may be weak. The line emphasizes the
relationship but has no statistical significance.
146

148 Cup with Handle
Do cups with higher right sides rise further? Yes. The average gain is
40% for cups with a higher right lip versus 35% for those with higher left lips.
The differences are statistically significant meaning that the results are proba
bly not due to chance.
Table 9.4 shows breakout statistics for unfiltered cups. Most (79%) ofthe
cupwithhandle formations have upside breakouts. Only 10% of the forma
tions break out upward but fail to rise by more than 5%. This suggests that
once a cupwithhandle formation breaks out upward, it generally continues
moving upward (but not necessarily very far; the most likely gain is only 10%
to 20%).
Almost three out of every four formations (74%) experience a throwback
to the price level ofthe right cup lip. This is a very high number and it suggests
a way to improve performance. A discussion ofthis follows in the Trading Tac
tics section.
The average time to complete a throwback is 12 days, which seems to be
about the number that many ofthe formation types in this book achieve. None
of the throwbacks takes longer than 30 days. Although several formations do
return to the cup top after 30 days, this behavior is normal price action, not a
throwback. Figure 9.3 shows this extended behavior during the February to
April 1993 period (remember, the scale is weekly).
It takes 6.5 months for the average formation to reach the ultimate high.
However, a frequency distribution of the duration shows that almost half
(47%) of the formations reach the ultimate high quickly (less than 3 months).
Over a third (35%) take longer than 6 months. This pattern parallels the most
likely rise and average rise for the formations. The most likely rise lands within
the 10% to 20% range while the average gain is 38%. A frequency distribution
Table 9.4
Breakout Statistics for Unfiltered CupwithHandle Formations
Upside breakout
Downside breakout
Horizontal breakout
Throwbacks
ultimate high
Percentage of breakouts occurring near
the 12month price low (L), center (C),
or high (H)
307 or 79%
74 or 1 9%
10 or 3%
30 or 10%
226 or 74%
12 days
L0%, C0%, HI 00%
180%, 151%, 127%, 112%, 108%, 108%
Trading Tactics 149
of the gains over the short, intermediate and longterm periods shows that
the longer it takes to reach the ultimate high, the larger the average gain.
I compared the breakout price level with the yearly price range. Every
formation has a breakout in the highest third of the range. You might consider
this surprising except that one selection criterion says that the cup occurs after
a minimum 30% price rise. This uptrend forces the breakout (which is itself at
the top ofthe formation) into the upper price range.
Breakout volume on the day of the breakout is quite high—registering
80% above the prior day (or 180% of the prior total). Volume remains high
throughout the following week.
Trading Tactics
Table 9.5 lists trading tactics. The measure rule predicts the price to which the
stock will rise, at a minimum. The traditional method involves determining the
height of the formation from lowest low in the cup to the high at the right cup
lip. Adding die difference to the high at the right cup lip results in the target
Table 9.5
Trading Tactics for CupwithHandle Formations
Measure rule
Use unfiltered criteria
Buy after throwback
Buy inner cup
Watch for 15% failure
Stop loss
Compute the formation height by subtracting the lowest
low reached in the cup from the high at the right cup lip.
Add the difference to the high at the right cup lip and the
result is the target price to which prices will climb, at a
minimum. Only 49% of the formations rise this far. Use half
the cup height to get a more realistic price target (met 73%
of the time).
To achieve the best performance, use the unfiltered criteria
when selecting formations.
Do not buy the stock until after the throwback occurs. Once
prices slip below the right cup lip after an upside breakout,
which occurs 74% of the time, wait for prices to close
above the lip before buying. This technique reduces failures
and improves performance.
If you discover a cup within a cup, buy on the breakout of
the inner cup (when prices rise above the inner cup lip). Be
prepared to sell at the price of the old high.
Many cups fail after rising only 10% to 15%. Be sure to use
stoploss orders to limit losses or to maximize gains.
Place a stoploss order below the handle to limit losses.
Raise the stop to breakeven or just below the nearest
support zone when prices rise.

150 Cup with Handle
price. However, this method only has a 49% success rate (less than halfthe for
mations reach their price targets). For a better target, compute the cup height
and take half ofit. Then continue as before. The stock reaches the new, lower
priced target 73% of the time. This is still shy of the 80% number I consider
reliable, but it gives a more accurate indication of the likely price rise.
Figure 9.9 is an example of the two measure rules in practice. Compute
the cup height by taking the difference between the right cup high (point A at
19) and the cup low (point B at 10). Add the difference (9) to the right cup lip
to get the price target (28). MidMay sees prices hit the target but plummet the
following week.
A more conservative price target uses half the formation height. This
gives a target of just 231
/z, reached during early July. The stock climbs to the
nearer target quickly and without the severe declines experienced on the way
to the more risky price target.
When selecting cupwithhandle formations, use the unfiltered selection
guidelines outlined in Table 9.1. When compared with the filtered selection
guidelines, they improve performance.
Usually, I recommend buying a stock once it breaks out, but with 74% of
the formations throwing back to the right cup lip, you might as well wait for
the throwback. Buy after it throws back to the right cup lip and once it rises above
the lip again. This technique improves the percentage gain from 38% to 39%.
Telxon Corp. (Electronics, NASDAQ, TLXN)
93 M | ) A S O N D 9 4 F M A M | | A S O N D 9 5 F M A M I | A S O N D 9 6 F M A M J |
Figure 9.9 Example of the two measure rules in practice. Compute the formation
height, divide by 2, and add the value to the right cup lip to get a conservative
price target. Trade the inner cupwithhandle formation for a better entry price. A
rightangled ascending broadening top appears during June and July 1995.
Sample Trade 151
Consider applying this technique to the stock shown in Figure 9.4.
Assume for a moment that the chart shows a valid cupwithhandle pattern. If
you bought the stock after the breakout, you would have received a fill between
30'/4 and 32 (which was the price range the day after the stock rose above the
right cup lip). The stock reached a high of 34:
/2 before throwing back.
Perhaps you would have held onto the stock and watched your gains
evaporate. However, had you waited for the throwback, you would not have
purchased the stock at all. Why? First you would have waited for the throw
back that eventually occurred on March 16 (the day prices gapped down).
Then you would have waited for prices to close above the right cup lip. This
never happened, so you would not have purchased the stock.
To be fair, this technique means you will miss some opportunities. Figure
9.9, for example, shows a stock without a throwback (remember, throwbacks
occur within a month of a breakout and the chart uses a weekly scale). This
means you would not have bought the stock.
Figure 9.9 also shows an inner cup. Ifyou are going to trade this forma
tion and can identify an inner cup, buy it. An inner cup appears as two widely
spaced minor highs that are at about the same price level. You score as the
stock advances to the old high (the outer, left cup lip) and further if the outer
cupwithhandle formation succeeds. Playing the inner cup shown in Figure
9.9 would have boosted profits about $2 a share or 12%.
Once you initiate a trade, place a stoploss order l
/s below the handle low.
The handle is a place of support and sometimes declines will stop at that point.
Placing a stop just below the low point will get you out ofthose situations when
the stock continues tumbling.
When the stock rises, move your stop to l
/s below the support zone near
est your breakeven point. That way, if the stock declines, you will be pro
tected. Continue raising the stop as prices climb. This technique forces you to
eventually take profits but saves you from watching them fritter away during a
reversal.
Sample Trade
Cody is in high school. He is not sure what he wants to do for a living, but he
still has a few years to figure it out before he graduates. When he is not chas
ing after cheerleaders, he either has his nose buried in the financial pages or is
reviewing charts on the computer screen. His interest in stocks follows in his
father's footsteps: The man works for a brokerage firm and taught Cody the
ropes.
Although Cody does not belong to the investment club at school, he pals
around with the players. One day, he overheard them talking about the stock
pictured in Figure 9.9. At first he did not think much about it until he looked
deeper. That is when he saw it: a cupwithhandle pattern.

152 Cup with Handle
He was not convinced the stock was a good trade, but did not have the
money to buy it anyway. He decided to paper trade it to see what he could
learn. On the daily time scale, he saw an inner cup forming at point C, so that
is the one he decided to trade.
Week after week, he waited for the buy signal but it did not come. Even
tually, the stock climbed above the right cup lip but he missed it. When he
pulled up the stock chart on the computer, a throwback had already occurred.
So, he waited for prices to climb above the cup lip again.
That happened on May 9, his girlfriend's birthday. Sensing a positive
omen, he made a notation to buy the stock, on paper, at the closing price the
following day (filled at 15'A). When he met his girlfriend the next day, she was
not impressed with the birthday present he gave her, and the stock closed
lower as well.
Two weeks later, the stock was moving up. Cody placed his stop H below
the handle low, at 143
/s (point D, which also marks the purchase point). When
the stock climbed above the outer cup, he raised the stop to Vs below the han
dle low or 17'/2. Then, he noticed a problem forming: a rightangled broaden
ing top formation. To him that was a bearish signal, so he moved his stop up
to just below the base at 20J
/4. Then he waited.
He got word that the stock was in trouble from his pals. They were not
too happy with the company for some reason. When he pulled the stock up on
his computer screen, he noticed that it had hit his stop in late August when
prices momentarily dipped. Cody whipped out his calculator and tallied up his
gains. He made $5 a share for a gain of over 30%. He chuckled to himself that
next time he would use his paper profits to buy his girl something other than
cubic zirconium.
10
DeadCat Bounce
RESULTS SNAPSHOT
Appearance
Failure rate
Event decline
Postevent decline
Surprising findings
An upward bounce and a declining price trend follow
a dramatic decline.
Longterm (over 6 months) bearish reversal
10%
The larger the event decline, the larger the bounce.
The larger the event loss, the quicker die bounce
reaches its high.
If you trade stocks long enough, you will probably run across this puppy: the
deadcat bounce. (I could not resist the pun). It is not so much a chart forma
tion as it is a warning to exit the stock quickly after a dramatic decline.
The event decline, which is the decline that spawns the deadcat bounce,
averages 25%. After the event decline, prices bounce up, round over, and con
tinue down another 15%, on average (measured from the event low to the ulti
mate low). Together, the two declines see prices move lower by an average of
37%.
The Results Snapshot outlines several unusual findings. The first one
reminds me of a bouncing ball: The larger the event decline, the larger the
bounce. When an event sends prices tumbling rapidly and severely, the
153

154 DeadCat Bounce Identification Guidelines 155
bounce, while correspondingly higher, occurs quicker. Short duration bounces
(to the recovery high, anyway) follow large event losses; smaller event losses
are more shallow and take longer to reach their bounce high.
Tour
What is a deadcat bounce? The name comes from the behavior of a stock after
an unexpected negative event. Figure 10.1 shows a typical example of a dead
cat bounce. In late September, the smart money started selling their holdings,
driving down the price and pushing up the volume trend. Prices declined from
a high of 427
/i6 to 3513
/i6 in just over a week. On October 9, a major brokerage
house lowered its intermediateterm rating on the stock. Down it went. In
2 days the stock dropped from a high of 37 /i6 to a low of 26, a decline of
over 30%.
For the next week and a half, the stock recovered somewhat, rising to
3213
/i6 and enticing novice investors to buy the stock. The stock moved lower,
then climbed again to form a double top. This was the end of the good news.
From the second peak, it was all downhill until midJanuary, when the stock
bottomed at 1813
/is. From the high before the event began to the ultimate low,
the stock plunged 50%! Welcome to the dead cat bounce.
Andrew Corporation (Telecom. Equipment, NASDAQ, ANDW)
Figure 10.1 Typical example of a deadcat bounce. A major brokerage firm low
ered its rating on the stock, sending it tumbling 50% in about 3% months. The
deadcat bounce allowed astute investors to sell their holdings and minimize their
losses before the decline resumed. The twin peaks in midOctober and early
November are a double top signaling further declines.
Table 10.1
Identification Characteristics of a DeadCat Bounce
Price gap The daily high is below the prior day's low, leaving a price gap
(breakaway) on the chart.
plunge On the negative announcement, prices gap down and plunge,
usually between 20% and 30% but can be as much as 70%
covering 2 or 3 days.
Bounce Prices recover somewhat and move upward. Do not be fooled; the
decline is not over.
Decline After the bounce finishes, another decline begins. This one is more
sedate but prices typically decline another 5% to 25%.
Are there characteristics common to the deadcat bounce? Yes, and Table 10.1
lists them. Consider Figure 10.2, a 47%, 1day decline. The stock peaked in
early February at 281
/2. It moved lower following a downsloping trendline until
late April. Then, it curled around at a low of 153
/4 and moved to reach a new
minor high at 217
/8. Then the Food and Drug Administration's advisory panel
rejected Cephalon's Myotrophin drug application. When the news hit the
Street, the stock gapped down and traded at almost half its value. Volume was
a massive 8.4 million shares, more than 15 times normal. During the next 3
Cephalon Inc. (Drug, NASDAQ, CEPH)
Event High
Apr 97 Jun jul Aug
Figure 10.2 A negative announcement triggered the deadcat bounce,
began when prices gapped down, bounced upward, then trended lower.
which

156 DeadCat Bounce
days, the stock recovered a portion of its decline by gaining $2 a share (low to
high). Then the remainder ofthe decline set in. As ifrubbing salt in the wound,
the stock moved down again in an almost straightline fashion. From the recov
ery high of 13'/2 to the new low of 9l
/2, the stock declined another 30%.
Figure 10.3 shows an even more alarming decline. Just 3 days before the
massive decline, a brokerage firm reported that it believed the company would
continue seeing strong sales and earnings trends. Perhaps this boosted expec
tations, but when the company reported a quarterly loss—instead of the profit
the Street was expecting—the stock dropped almost 43 points in 1 day. That is
a decline of 62 %.
The stock gapped downward, a characteristic that most deadcat bounces
share. Anegative news announcement is so surprising that sell orders overwhelm
buying demand. The stock declines and opens at a much lower price. Volume
shoots upward, typically several times the normal rate. Figure 10.3 shows that 49
million shares exchanged hands on the news, about 20 times normal.
Usually the 1day decline establishes a new low and prices begin recover
ing almost immediately. Figure 10.3 shows that the stock made a new low the
following day but then closed up a day later.
After a massive decline, the bounce phase begins. Most of the time, a
stock will rise up and retrace some ofits losses. However, the bounce phase for
Oxford Health Plans was brief—only 1 day. The stock closed higher, but the
downward trend resumed the next day. In less than 2 months, the stock
dropped by half, from a high of 28% to a low of 133
/4.
Oxford Health Plans (Medical Services, NASDAQ, OXHP)
Oct97
Figure 10.3 Negative news announcement triggered the massive 1day decline,
which saw prices drop by 43 points or over 60%, but the decline was not over as
the stock fell an additional 43%.
What types ofevents cause these massive declines? Almost all the events are
company specific: negative earnings surprises, bad samestore sales numbers,
failed mergers, accounting sleight of hand, outright fraud—that sort of thing.
Sometimes the news affects more than one company. Figure 10.2 shows what
happened to Cephalon, but Chiron stock was not immune. Chiron has a joint
development and marketing agreement with Cephalon for the Myotrophin drug,
so its stock also took a hit, but not nearly as large (less than 5%) as Cephalon.
Most ofthe time investors cannot predict the event. Ifyou own the stock,
you will lose your shirt. The question then becomes, how much of your
remaining wardrobe do you want to lose? We see in the Trading Tactics sec
tion that it pays to sell quickly.
Focus on Failures
Not all massive declines end in a deadcat bounce. Consider the event shown
in Figure 10.4. On April 3, 1997, die company released earnings that fell short
of expectations and announced that its merger with another company was ter
minated. Several brokerages downgraded the stock. It tumbled from a high of
17'A to a low of 95
/s, a decline of 44%. Like all deadcat bounces, the stock
recovered. However, instead of bouncing up then turning down and moving
lower, this stock continued trending up. In less than 3 months, the stock recov
ered its entire loss.
Checkpoint Systems (Precision Instrument, NYSE, CKP)
Figure 10.4 A deadcat bounce formation failure. After the decline, the stock
moved higher and kept rising instead of moving back down.

158 DeadCat Bounce
Why did the stock fail to bounce and head lower? Events that take place just
after the negative news announcement explain the stock's behavior. Several insid
ers bought the stock. Even the company got into the act and announced it was
purchasing 10% of the stock. Together, the news sent the stock moving higher.
Subsequent events kept the momentum building and the stock continued rising.
A survey of all 24 failures in the database reveals that 83% have stock
trends that continue moving lower without any significant upward bounce.
The remainder (4 formations) are similar to that shown in Figure 10.4, that is,
prices rebound and move up.
Statistics
Table 10.2 contains the general statistics for the deadcat bounce. I located 244
formations in 500 stocks over 5 years. Of these formations, 54% act as rever
sals of the prevailing trend, and the remainder act as consolidations. Almost all
the formations (90%) perform as expected. That is to say, after a major decline,
the stock bounces and heads lower. Only 20 move lower without a significant
bounce and 4 trend upward; they are the failures.
The average decline, as measured from the high the day before the major
decline to the recovery low, is 37%. The recovery low is usually the same as the
Table 10.2
General Statistics for the DeadCat Bounce
from1991 to 1996
Failure rate
(event high to recovery low)
Average event decline
Average event duration
Number of formations with price gap
Number of gaps closed during recovery
Number of gaps closed in 3 months
Number of gaps closed in 6 months
Average recovery bounce height
Additional postevent decline
Average postevent duration
Number of formations declining below
event low
244
24 or 10%
37%, with most likely loss between
30% and 40%
25%, with most likely loss 20%
2 days; 92% last 4 days or less
194 or 80%
48 or 22%
80 or 36%
119 or 54%
19%
15%, with most likely loss between
5% and 25%
3 months (90 days)
198 or 81%
Statistics 159
ultimate low. After the stock completes its major decline and bounces upward, the
recovery low is the lowest price before any significant rise signals a trend change.
The most likely decline is in the 30% to 40% range. This is unusual as
most formations have a few large declines that skew the overall average
upward. With the most likely decline near the average, the declines are well
spread as seen in Figure 10.5. Notice the bellshaped graph. The graph shows
the results of a frequency distribution of losses from the high price the day
before the decline to the recovery low.
The average event loss is 25%. The event loss is the difference between
the high the day before the large decline to the lowest low before the bounce
begins, expressed as a percentage. A frequency distribution of the event loss
shows that the most likely event loss is 20%.
The event loss happens quickly, usually in just 2 days. This is the massive
decline that gets the ball rolling. Most times the stock suffers a large 1day loss
then makes a lower low the next day before entering the bounce phase (a 2day
decline occurs 71% of the time). Sometimes the stock will continue moving
down for a few days, but most (92%) begin rebounding in 4 days or less.
Most market participants find the negative news surprising. As such, prices
gap lower (that is, the current day's high is below the prior day's low, leaving a
gap on the price chart). Only 22% of the formations showing a gap close them
(prices rising far enough to fill the gap) during the bounce or recovery phase.
Even in 3 months, only 36% close the gap and slightly over half (54%) close
in less than 6 months. Since 80% of the formations show gaps and only half
50 t
Percentage Loss
Figure 10.5 Frequency distribution of total loss. Note the bellshaped graph of
total losses from the high the day before a large decline to the recovery low. The
most likely decline is 30% to 40%.

160 DeadCat Bounce
close within 6 months, the longterm bearish implication of this formation
becomes obvious.
During the bounce phase of a deadcat bounce, the stock rises 19%, on
average (as measured from the event low to the recovery high). Does the
bounce height relate to the severity of event loss? Yes. In simple terms, the
larger the event loss, the larger the bounce. Figure 10.6 shows the relationship.
An additional 15% loss (as measured from the event low to the recovery
low) occurs after the recovery bounce completes and prices start declining
again. Prices usually fall below the event low and continue moving down. A
frequency distribution of the postevent loss ranges from 5% to 25% and is
evenly distributed. This wide range suggests that even though the event
decline may be massive, sometimes substantial additional losses follow. The
postevent loss occurs much more slowly, averaging about 3 months (90 days)
in duration, as measured from the event low to the recovery low.
Figure 10.7 shows the relationship between event loss and the number of
days to the recovery high (the highest high reached during the bounce).
Although it may be difficult to see the relationship, one can generally say that
as the size of the event loss grows, the days to the recovery high lessen. In other
words, the bounce becomes steeper (occurring quicker). This seems to be con
trary to what you would expect.
Taken together, Figures 10.6 and 10.7 suggest that short duration, high
recoveries (to the recovery high, anyway) follow large event losses. Smaller
losses are more shallow and take longer to reach their bounce high.
Figure 10.6 Relationship between event loss and bounce height. Like a ball, the
larger the loss the larger the bounce.
Statistics 161
Figure 10.7 Relationship between the event loss and days to the bounce high.
The line helps describe the relationship but has no statistical significance.
Over 8 out of 10 (81%) formations decline below the event low. The
event low is the low price reached after the massive decline but before the
bounce. This statistic emphasizes that even though you have suffered a painful
loss (or had your profits trimmed), you should still sell your position because
further losses are coming.
Table 10.3 shows additional statistics related to the deadcat bounce.
Every formation showed extraordinarily high volume on the day the event
occurred. After prices decline, they recover and reach a bounce high in less
than 3 weeks (19 days).
Where in the yearly price range does the event occur? A frequency dis
tribution of the day before the event indicates that most stocks (40%) are
within a third of their yearly high. The center third of the yearly price range
follows closely with 39%. When you substitute the percentage loss into the
yearly price range, you find returns behave about the same. The average for
mation declines between 35% and 37% regardless of where it begins in the
yearly price range (as measured from the day before the event begins).
Breakout volume is massive (596% of the prior day's volume) and con
tinues to be high for the next week.
I looked at gaps and tried to discover ifthey hold any special significance. Do
formations with gaps have larger event losses? No, since the average decline is
24% for those formations with gaps and 27% without. Do formations with gaps
have a better recovery (that is, do they bounce higher)? Again, the answer is no as
the recovery rise is 18% for those formations with gaps versus 20% without.

Table 10.3
Additional Statistics Related to the DeadCat Bounce
High volume event (at least 50% above prior day) 238 or 1 00%
For successful formations, days to recovery high
(from event end) 1 9 days
Percentage of deadcat bounces occurring near the
1 2month low (L), center (C), or high (H) L20%, C39%, H40%
Percentage loss for each 1 2month lookback period L37%, C35%, H37%
with day before breakout 596%, 304%, 1 77%,
154%, 140%, 125%
Percentage of formations with gaps having larger
event declines than formations without gaps 24% versus 27%
Percentage of formations with gaps having better
recovery (a higher bounce) than formations
without gaps 1 8% versus 20%
Note: The negative event occurs on high volume, the bounce occurs quickly, and the
recovery is slow.
Trading Tactics
There is not much that can be said for trading the deadcat bounce unless you
are shorting the stock (see Table 10.4). For long positions, wait for the bounce
then sell. About a third of the time (36%), prices reach the bounce high in the
first week. Over half the formations (56%) take 2 weeks or less. When the
stock peaks and rounds over, dump it. If you choose to hang onto your posi
tion, you will likely incur further losses and it may take well over 6 months
before you come close to recovering them. Why not invest your remaining
capital in a more promising situation?
If you are considering selling the stock short to profit from the impend
ing decline, look at Figure 10.8, which shows the percentage decline from the
bounce high to the recovery low. On average, the most likely decline is
between 15% and 25%, large enough to risk a trade.
The last trading tactic is to be aware of the deadcat bounce and its effect
on prices. When a severe decline takes hold of a stock, the cause is not trivial.
It takes time for the company to fix the problem and recover. Ignore any bull
ish chart formation occurring in a stock in less than 6 months (even up to a
year). The chart pattern will likely fail or, if it works, the rise may be short
lived when the company announces more bad news (such as poor quarterly
earnings).
162 DeadCat Bounce Sample Trade 163
I auie IU.H
Trading Tactics for the DeadCat Bounce
Wait for rise, then sell The stock will make a new low then begin to bounce. Sell
after the bounce rounds over, usually in 1 to 2 weeks. The
worse the event decline, the quicker the bounce high
appears, and the higher prices bounce.
Short sales Sell short after the bounce rounds over. Expect a decline to
at least the event low. Most of the time (81%), the stock
continues lower by another 15%, on average.
Avoid formations Avoid all bullish chart formations for at least 6 months (or
even up to a year) in a stock showing a deadcat bounce. If
they work at all, the gains are below average.
Note: Wait for the bounce, then sell or sell short.
Sample Trade
Once satisfied that you know the implications of the bad news and the reasons
for the stock's massive decline, short the stock. Consider Figure 10.9, a dead
cat bounce in Cerner Corporation. The stock dropped five points (25%) after
the company said earnings would fall short of expectations and the outlook for
Figure 10.8 Frequency distribution of declines from the bounce high to the
recovery low. If you time your short sale correctly, the profits can be rewarding.

164 DeadCat Bounce
]un96 Aug Sep
Figure 10.9 A negative earnings announcement triggers a deadcat bounce. Jill
sold the stock short just after the bounce high then covered when prices closed
above the trendline. The trade resulted in a 20% gain in 1 month.
the remainder of the year was grim. The stock closed higher on each of the
next 4 days then closed lower. Jill, after seeing the stock climb the hill, sold the
stock short and received a fill at the closing price of 15. She then waited, watch
ing the stock closely. It continued moving down—as predicted.
The earnings announcement forced the stock down another 20% in 2
days. Expecting another deadcat bounce, Jill held on to her position. The
stock rose in an uneven fashion over then next week or so, then rounded over
and headed lower.
Jill connected the tops from the preannouncement day onward in a down
sloping trendline. When prices eventually closed above the trendline, she knew
it was time to close out the position. The next day she bought the stock back
and received a fill at 12, 1
A below the daily close. She sat back and totaled up
her profits and realized she made almost $3 a share, or about 20% in just
1 month.
As good as the trade was, had she waited until November to close out the
position, she would have made an additional $1.50 a share (the stock reached a
low of 10'/2). However, between the time of covering the short and the ultimate
low, the stock climbed back to 171
A. The moral is, you never go broke taking
a profit.
11
Diamond Tops
and Bottoms
RESULTS SNAPSHOT
Diamond Tops
Appearance
Failure rate
Average decline
Volume trend
Fullbacks
Percentage meeting
See also
Diamond Bottoms
Appearance
Failure rate
Averagerise
Average volume trend
Diamond pattern forms after an upward price trend.
25%
Downward until breakout
59%
79%
HeadandShoulders Tops
Diamond pattern forms after a downward price trend.
Short to intermediateterm (up to 6 months) bullish
reversal
13%
165

Throwbacks
Percentage meeting
See also
43%
95%
HeadandShoulders Bottoms
The Results Snapshot shows the important results of diamond tops and bot
toms. In appearance, the only difference between the two diamond patterns is
the price trend leading to the formation. For diamond tops, the prior price
trend is upward, whereas diamond bottoms have price trends that lead down to
the formation.
The performance of the two types is similar. Both act as reversals of die
prevailing price trend with a volume trend that diminishes over time. Volume
on the day of the breakout is also high.
The failure rate for tops, at 25%, is more than double the rate for bot
toms. I consider failure rates above 20% to be alarming, so you might consider
tops unreliable.
The average decline (21%) and rise (35%) is about what you would expect
for reversals, with bottoms a bit shy of die usual 40% rise for bullish forma
tions. However, the most likely decline for tops is near the average, suggesting
that there are few large declines to distort the average. Diamond bottoms, with
a likely rise of just 15%, are well away from the 35% average gain. This sug
gests there are a high number of formations with smaller gains that balance a
few larger ones.
Baker Hughes (Oilfield Svcs./Equipment, NYSE, BHI)
25
24
23
17
Feb 95 Mar
Figure 11.1 A good example of a diamond top. Notice that prices quickly return
to the $20 level.
Tour 167
Tour
What does a diamond top or bottom look like? Figure 11.1 shows a good
example of a diamond top. After rising steadily since midFebruary, the stock
jumps 11
/2 points on April 20. Volume on that day is well above average. Then,
the stock forms higher highs and lower lows, as ifit is tracing a broadening top
pattern. Volume diminishes throughout the early pattern development. Then,
things reverse. The third minor high calls the top and prices decline. The
minor lows begin moving higher even as the tops are descending. This phase
of die pattern looks like a symmetrical triangle. However, volume continues
receding, albeit at an irregular rate.
Prices break down out of the pattern on June 8 accompanied by volume
that is about average. Prices meander sideways for about 2 weeks before plung
ing and retracing all the gains since the midApril, 1day rise.
The pattern is a diamond top; it signals a reversal of the prevailing price
trend. The chart in Figure 11.1 shows the typical behavior of a top: Prices
return to the level before the diamond begins. In this regard, the reversal
stands out like a sore thumb. Of course, not all tops act this way. Some signal
a reversal of the primary trend and prices not only retrace their recent gains
but continue moving down.
The diamond bottom, shown in Figure 11.2, is similar to the top version
with the exception of the prevailing price trend. In Figure 11.2, prices are
trending down toward the diamond bottom. In Figure 11.1, prices are trend
ing upward before the start of the formation.
Coors, Adolph Co. (Beverage (Alcoholic), NASDAQ, ACCOB)
24
]u!91
Figure 11.2 A diamond bottom reversal. Volume typically recedes through the
formation until the breakout day.

The diamond bottom begins by widening out and tracing higher highs
and lower lows, then the process reverses. The price range narrows until the
breakout occurs.
Volume throughout the formation is diminishing. The breakout usually
sports a significant rise in volume. Figure 11.2 shows high volume on the
breakout when prices gap through the diamond boundary. In less than 3
months, the stock climbs over 20% to a high of 22'/4.
Table 11.1 lists the identification guidelines for diamond tops and bottoms.
Consider the diamond top pictured in Figure 11.3. The shortterm price trend
is up just before the formation, leading to the minor high on the left. Then
prices decline and form a minor low before moving higher again. In late Sep
tember, prices reach a new high before cascading downward to finish below the
prior minor low. Again, prices rise up and form another minor high before
breaking down through the upward trendline on the right. The fluctuations of
minor highs and lows form a diamond shape when the peaks and valleys con
nect such as that shown in Figure 11.3. Notice that the diamond is not sym
metrical; irregular diamond shapes are common for diamonds.
Table 11.1
Identification Characteristics of Diamond Tops and Bottoms
Prior price trend
Diamond shape
Volume trend
Breakout volume
Support and resistance (SAR)
For diamond tops, prices usually trend up to the
formation, whereas bottoms usually form at the end of
a downward price trend. With this definition, diamond
tops (or bottoms) need not form at the top (or
bottom) of a price chart—they can form anywhere.
Prices form higher highs and lower lows (widening
appearance), then lower highs and higher lows
(narrowing appearance). Trendlines surrounding the
minor highs and lows resemble a diamond. The
diamond need not appear symmetrical.
Diminishing over the length of the formation
Usually high and it can continue high for several days
The formation creates a location for support or
resistance. Diamond tops usually show SAR near the
top of the formation, whereas diamond bottoms show
SAR near the formation bottom. SAR duration can last
up to a year or more.
Asarco Inc. (Copper, NYSE, AR)
Oct
Figure 11.3 A diamond top masking a headandshoulders top. In either case,
the bearish outlook is certain.
The volume trend is receding, especially in the latter half of the formation
when the price range is narrowing (and the chart pattern resembles a symmet
rical triangle). The breakout volume is usually high but is not a prerequisite to
a properly behaved diamond. In Figure 11.3, the volume on the breakout day
and succeeding days is tepid at best but trend upward as prices fall.
The pattern is a headandshoulders top, with the left shoulder, head, and
right shoulder marked on Figure 11.3. The volume pattern is typical for a
headandshoulders top, with the right shoulder volume vastly diminished
when compared to the left shoulder or head volume.
Should you locate a diamond pattern and discover that it may be a head
andshoulders top, do not worry. In both cases, the formation is bearish. When
such a collision occurs, choose the formation that gives you the more conser
vative performance results (see the measure rule).
Support and resistance for diamond tops commonly appear at the top of
the formation, as seen in Figure 11.4. The diamond reversal forms a resistance
level, repelling prices during the rise in March and April 1993, and is not
pierced until a year later.
Acongestion zone forms in October 1993 and lasts through March ofthe
following year before prices climb convincingly above the resistance area. Even
then, during April and May 1994, prices are buoyed by the support zone at 31
created a year and a half earlier.
Figure 11.5 shows a diamond bottom. The price trend is downward for
nearly 2 months, leading to the formation. Prices rebound slightly and the

Gillette Co. (Toiletries/Cosmetics, NYSE, C)
Figure 1 1 .4 Support and resistance for diamond top appears at top of formation.
A support and resistance zone at 31 created by the diamond top lasts for a year
and a half. Note the weekly time scale.
range widens as higher highs and lower lows appear. Then the tide turns and
the range narrows; higher lows follow lower highs. The diamond pattern takes
shape after connecting the boundaries of the price movements.
Trading volume throughout the formation is receding. This is typical but
not a prerequisite for a wellformed diamond bottom. As in diamond tops,
there are wide variations in the volume pattern. Overall, however, the volume
trend diminishes over time until the breakout, then volume usually jumps
upward. Figure 11.5 shows that breakout volume is four times the prior day but
is just slightly above average for the stock.
Figure 11.6 illustrates the support area often promoted by diamond bot
toms. The figure shows support at the $10 level on a weekly scale. Although
support varies from diamond to diamond, when it appears after a diamond bot
tom, it is usually near the base of the formation. Another area of support com
monly appears when the stock throws back to the level of the breakout. Figure
11.6 shows an example of this. After climbing away from the formation after
the breakout, a stock sometimes pauses, reverses course, and heads lower. Sup
port meets prices that decline into the formation area, usually stopping briefly
near the breakout price, then prices turn around again and head back up. This
throwback to the formation happens more than a third of the time (43 %) and
represents another opportunity to initiate a trade or add to a position.
«tw^ enterprises (Manuf. Houslng/Rec. Veh., NYSE, FLE)
|,t'1
''
Figure 11.5 A diamond bottom with receding volume trend. Prices quickly
recover and reach new highs.
Teradyne Inc. (Semiconductor Cap Equip., NYSE, TER)
93 M A M I A S
Figure 11.6 Support areas for diamond bottoms are near the base of the forma
tion. Shown here is support at 10 on a weekly scale.
171

Focus on Failures
Not all diamond tops and bottoms work out as expected. Figure 11.7 shows a
diamond top that fails to breakout downward. As in Figure 11.1,1 would expect
the stock to return to its jumpoffpoint, that is, the 108 level. Instead, the stock
breaks out upward and continues moving higher until it reaches a high of
1273
/4. The move, which occurs less than 2 weeks after the breakout, signals a
peak for the stock. From that point, it is all downhill. In early April 1994, the
stock touches bottom at 92'/2, well below the breakout price of 1183
/8. If you
sold the stock before the breakout, you may be upset that it continues higher,
but eventually your tactic pays off. The stock declines 22% from the breakout
low to die ultimate low in 1 year.
Is there a reason the stock continues higher? Certainly the breakout vol
ume is not supportive of an upward move. Although volume on the day of the
breakout is 69% above the prior day, there is no significant volume spike
shown on the chart (translation: breakout volume is not above average). This
weak volume is unusual but not unheard of. The weak volume breakout and
decreasing volume trend over the next week or so, even as prices climb, is a
warning that the upward momentum is running out of steam.
Although it is not clear from looking at Figure 11.7, the volume trend,
measured by the slope of a linear regression line ofvolume, is downward. This
is typical for most diamond tops and bottoms but offers no clue to the eventual
failure of this situation. I can see no reason why this diamond fails to breakout
Atlantic Richfield Co. (Petroleum (Integrated), NYSE, ARC)
Nov 92 Dec |an 93 Feb Mar Apr May
Figure 11.7 A failure of a diamond top to reverse direction.
downward. The lesson is that you should wait for the breakout before trading
your position.
Figure 11.8 is an example of a failed diamond bottom. The stock reaches
a high in early February at a price of 32'/4, then heads lower. The diamond
forms at about the level where the prior downtrend stops descending (in early
April). One could expect the stock to stop falling when it meets support. For
almost 2 months (April and May), that is exactly what happens. The stock
moves horizontally, forms the diamond bottom, but then breaks out down
ward. Briefly, the stock pulls back to the breakout price then heads lower. It
reaches a low of lO'/z in midJune then moves sideways for 5 months (not
shown). By the end of this study, the stock is trading at less than $1.
Breakout volume is above average and remains high for several days. The
pullback to the breakout level gives investors the opportunity to sell their posi
tions before the downhill run resumes.
The diamond low, at 16'/4, approximates the low in early April. These two
lows, when taken together, initially appear to be a double bottom. Volume is
higher on the left side of the double bottom than the right. The rise between
the two troughs is sufficient to validate a double bottom. However, prices fail
to rise above the confirmation point (211
A, or the highest high between the two
bottoms) so the formation is not a double bottom. This failure suggests prices
will move lower.
Double bottoms and diamond bottoms are both bullish formations, so
why did the stock fall? I could find no technical evidence to suggest why the
50Off Stores Inc. (Retail (Special Lines), NASDAQ, FOFF)
Feb 92 Mar Apr Jim jul
Figure 11.8 A diamond bottom failure. The diamond bottom fails to reverse as
prices break out downward and continue moving down. The double bottom is
unconfirmed; it is not a true double bottom.

two formations rail to perform as expected. However, since the stock continues
down to less than $1, this decline suggests that the fundamental situation is
decidedly weak. This alone may explain the stock's behavior.
Statistics
Table 11.2 contains general statistics for diamond tops and bottoms. Due to
the dearth of formations located in 2,500 years of daily price data, I reviewed a
more recent database to augment the statistics. The numbers of formations
uncovered indicate diamond tops and bottoms arc a rare breed. Most of the
time, they act as reversals of the prevailing price trend (80% for bottoms and
78% for tops).
The failure rate is higher for diamond tops than bottoms (25% versus
13%). Failure rates above 20% suggest an unreliable formation, so, if you
decide to trade a diamond top, be extra careful. Only diamond tops suffer from
5% failures. That is when prices move less than 5% in the breakout direction
before reversing. With just 45 diamond bottoms, the sample size is not large
enough to really determine the 5% failure rate.
The average rise after a bottom is 35%, although the most likely rise is
just 15%. Figure 11.9 shows the most likely rise and it is computed using a fre
Figure 11.9 Frequency distribution of gains for diamond bottoms. The chart sug
gests the most likely rise after a breakout from a diamond bottom is 20%, but the
actual value is 15% (using a finer scale than the one shown).
quency distribution of gains. The column with the highest frequency becomes
the most likely rise. Only 37 formations qualify for the chart. Still, the 20%
column stands out as the one with the highest frequency. Scaling the columns
in 5% increments (instead of 10%) shows that the 15% column has the high
est frequency. Even though the 20% column is highest in the chart, the most
likely rise is really 15%.
The average decline from a diamond top is 21 %, although the most likely
decline is 20%. Figure 11.10 shows a frequency distribution of the losses.
There are a larger number of samples and the bellshaped curve is smoother.
Diamond tops are one of the few cases where the average decline is near the
most likely decline. This suggests the declines are evenly distributed about the
average (there are few large declines that pull the average upward).
The 35% rise for diamond bottoms is below the average 40% return for
other types of bullish formations. Diamond tops, on the other hand, show
declines (21 %) similar to other bearish formation types.
The measure rule, discussed in the Trading Tactics section of this chap
ter, involves calculating the formation height and either adding or subtracting
the difference from the breakout price. The result is the minimum target price.
For bottoms, nearly all (95%) of the formations reach the predicted target
prices. However, tops have a success rate of 79%. I consider values above 80%
to be reliable.
The formation length from start to breakout is similar for both diamond
types, at about 7 weeks (49 and 52 days). That is comparatively short.
Statistics 175
Table 11.2
General Statistics for Diamond Tops and Bottoms
Description
Number of formations in 500
stocks from 1991 to 1996
stocks from 1 996 to 1 998
Failure rate
Volume for breakout day and
next 5 days compared with day
before breakout
Bottoms
34
11
6 or 1 3%
35% rise
15% rise
35 or 95%
1 .5 months (49 days)
151%, 1 74%, 151%, 123%,
1 37%, 1 39%
Tops
111
27
31 consolidations,
1 07 reversals
35 or 25%
21% decline
20% decline
86 or 79%
1 .5 months (52 days)
152%, 200%, 177%,
1 60%, 160%, 1 79%

176 Diamond Tops and Bottoms Statistics 177
that fail to rise by more than 5%. As explained earlier, this is probably due to
the small sample size.
The significance of this result is that once prices break out of the forma
tion, they continue moving in the same direction (by at least 5% anyway), sug
gesting that a trader should wait for the breakout, then trade with the trend.
Breakout statistics listed in Table 11.3 show bottoms having upside break
outs 82% of the time and tops having downside breakouts 79% of the time.
Bottoms have throwbacks to the price level of the breakout (where prices
. pierce the diamond boundary) 43% of the time, whereas pullbacks after a dia
mond top reversal occur just over half the time (59%). Throwbacks allow
" investors another opportunity to either place a trade or add to their position.
Pullbacks are an opportunity to exit the position before the decline resumes or
to initiate a short sale. If you are on the wrong side of a trade and are given
another opportunity to leave the trade, take it. If you do not, things only
get worse.
For both throwbacks and pullbacks, the average time for prices to return
to the breakout point is 11 days. In all cases, the throwback or pullback occurs
in less than 30 days. Anything taking longer than a month classifies as normal
price action and is not a throwback or pullback.
For successful formations, the average time to reach the ultimate low or
high is almost twice as long for bottoms (122 days) than it is for tops (65 days).
This makes sense when coupled with the average price rise and fall. Bottoms
rise almost twice as far (35%) as tops fall (21 %), so it is not unusual for bottoms
to take almost twice as long to cover almost twice the distance.
Where do breakouts occur in the yearly price range? For bottoms, 56%
of breakouts occur in the center third of the yearly price range. For tops, 69%
of breakouts happen in the top third of the price range. Considering these
numbers, you may wonder how a diamond bottom can occur in the top of the
price range or a diamond top can occur in the bottom of the price range.
I base my definition of a bottom or top on the price trend leading to the
formation, not the price itself. Diamond bottoms have price trends that decline
to the start ofthe formation; diamond tops have price trends that lead up to the
formation at its start. If this is still confusing, review Figures 11.1 and 11.2.
The formations are grouped according to the price trend because they
share the same characteristics. When prices rise up to a formation (as in the
case of a diamond top), expect a reversal to send prices declining once the for
mation completes. If you classify a formation based on price, then tops will
form at the top of the yearly price range and bottoms will form at the bottom.
How do you classify a formation that appears in the middle of the range? It can
be either a top or a bottom. See the dilemma?
Substituting the percentage gains or losses in the yearly price range, we
find most successful bottom reversals rise by 37% (and they have breakouts that
occur in the center third of the yearly price range). Since all three of the gains
(30%, 37%, and 35%) are relatively close, and since the frequency distribution
Figure 11.10 Frequency distribution of losses for diamond tops. The most likely
decline after a diamond top is 20%.
Table 11.3 shows breakout statistics for diamond tops and bottoms. Both
tops and bottoms have breakout volume that is half again has much as the
prior day, on average. It remains high throughout the next 5 trading days. Ten
diamond top formations have downside breakouts that fail to continue declin
ing by more than 5%. Since this represents only 9% of the formations, the
result is quite good. There are no diamond bottoms that have upside breakouts
Table 11.3
Breakout Statistics for Diamond Tops and Bottoms
Description Bottoms Tops
Upside/downside breakout but failure 0 10 or 9%
Upside breakout 37 or 82% 29 or 21 %
Downside breakout 7 or 16% 109 or 79%
Horizontal breakout 1 or 2% 0
Throwbacks/pullbacks 16 or 43% 64 or 59%
Average time to throwback/pullback
completion 11 days 11 days
ultimate high/low 4 months (122 days) 2 months (65 days)
12month low (L), center (C), or high (H) L16%, C56%, H28% L7%, C24%, H69%
Percentage gain/loss for each 12month
lookback period L30%, C37%, H35% L26%, C15%, H20%

used to gather these statistics rests on a few samples (5, 18, and 9, respectively),
do not read too much into the numbers.
Tops tell a similar story. The statistics suggest tops occurring in the lower
third of the price range decline further. This may or may not be true since it is
based on only five samples (the other two categories have 17 and 50 samples,
respectively).
Trading Tactics
Table 11.4 shows trading tactics for diamond tops and bottoms. Use the mea
sure rule to predict the minimum price move. Consider Figure 11.11, a chart
of a diamond top and diamond bottom. Compute the measure rule by first
finding the formation height. Locate the lowest low in die formation (shown
as point B) and subtract it from the highest high (point A). Subtract the differ
ence (in the case of diamond tops) or add it (for bottoms) to point C—the
breakout price. The result is the minimum target price.
For the diamond top, the formation height is 75
/s (that is, 79'A 715
/s).
Subtract the difference from point C to get the target price of 657
/s (or 731
/?
75
/s). Point C, incidentally, is where prices pierce the diamond trendline. Prices
meet the target just 1 week after the breakout.
Calculation of the measure rule for diamond bottoms proceeds in a sim
ilar manner. The height turns out to be 33
/4 (that is, 713
/4 68). Add the differ
ence to point C, the location where prices pierce the diamond boundary. The
target price is 74'/2 (or 703
/4 + 33
/4). Prices meet the target in just 3 days.
As noted in the Statistics section ofthis chapter, prices fulfill the measure
rule 95% of the time for diamond bottoms and 79% for tops. Both these num
Table11.4
Trading Tactics for Diamond Tops and Bottoms
Measure rule Compute the formation height by subtracting the lowest low
from the highest high in the formation. For tops, subtract the
difference from the location where prices pierce the diamond
boundary. For bottoms, add the difference to the breakout
price. The result is the minimum price move to expect.
Alternatively, formations often return to price levels from which
they begin. The base serves as a minimum price move.
Wait for breakout For best results, wait for prices to close outside the diamond
trendline before placing a trade.
Risk/reward Look for support (risk) and resistance (reward) zones before
placing a trade. These zones are where the trend is likely to stop.
From the current closing price (before the breakout), compute
the difference between the zones and the current price. The
ratio of the two must be compelling enough to risk a trade.
Trading Tactics 179
Dow Chemical Co. (Chemical (Basic), NYSE, DOW)
Diamond Triangle 66
Bottom
Aug 94 Sep Oct Nov Dec |an 95 Feb Mar Apr May )un Jul Aug
Figure 11.11 Adiamond top and a diamond bottom. Compute the measure rule
using the formation height by subtracting point B from point A. For diamond tops,
subtract the difference from point C and for bottoms, add the difference. The
result is the expected minimum price move. Diamonds often return to their base.
The Expected Decline and Expected Rise lines are another way to gauge the mini
mum price move. A symmetrical triangle appears in late May.
bers are high enough that you should consider the measure rule a reliable price
prediction mechanism.
However, there is an alternative method that sometimes yields more
accurate results. The method involves looking at the price chart and seeing if
there is something to reverse. By this I mean diamonds sometimes form after
a quick runup in prices. The reversal will usually erase these gains and return
prices to where they were before the runup.
Figure 11.1 is a good example of this. Prices make a 1day jump from the
20 area. After the reversal completes, prices quickly return to the same level.
Figure 11.11 also shows the two jumpoff points. The diamond top starts
climbing from the 68 level. Prices quickly return to this level after the reversal
(see the Expected Decline line shown in Figure 11.11).
The diamond bottom has a start from the 74 level (see the Expected Rise
line in the figure) to which prices quickly return after the reversal. After reach
ing that level, prices do not exceed it for almost 2 months.
When trading technical formations like diamond tops and bottoms, it
is always safest to wait for the breakout. Although failures from bottoms
occur only 13 % of the time, tops are more error prone (with a failure rate of
25%).

If you do not wait for the breakout, you may face a situation similar to
that discussed in the Focus on Failures section of this chapter. Instead of
reversing, prices resume their original trend and the investor misses out on
additional gains or suffers larger losses.
Before placing a trade, consider the risk/reward ratio. In essence, you
first identify the support and resistance levels and calculate the difference
between those levels and the current price. Trades that result in risk/reward
ratios of one to four or higher are worth making. When the ratio drops below
one to four, the risk may be too high to warrant a trade. An example makes the
calculation clearer. For the diamond bottom shown in Figure 1 1.6, assume the
figure is all that is known about the stock. On the left side of the figure, the
stock descends to 6'/2 before rebounding. The stock bounces from this level,
suggesting that the price level is a support zone.
The measure rule suggests prices will rise to 145
/s, about where prices
topped out recently (the resistance level shown in Figure 1 1 .6). The close the
day before the breakout, is 1 1 /i6. Calculating the differences between the sup
port and resistance zones to the closing price gives a ratio of (1 113
/i6 6l
/i) to
(145
/g H13
/i6) or 5.31 to 2.82. The ratio is slightly less than two to one and it
warns that the risk of failure is higher than the potential reward. The 2:1 ratio
is well outside the 1 :4 minimum. You would be taking on twice as much risk as
potential reward. You could always tighten up the risk by placing a stoploss
order at die base of the diamond or even closer to the purchase point. The
closer the stop is to the buy point, the more likely normal price fluctuations
will take you out.
Sample Trade
Scott recently graduated from engineering college and took his first profes
sional job at a growing software company. The job pays well, but he has many
school loans and a mountain of debt. He thought ofusing his paycheck to keep
ahead of the bills while depending on the bull market to furnish the luxuries.
He had his eye on a new stereo system and wanted it for a party he was
hosting during the Fourth of July festivities. That did not leave him much
time, so he searched for a chart pattern he could trade profitably. He chose the
diamond bottom shown in Figure 11.11. Scott first noticed the diamond in
May, a few days before the breakout. He believed that the price would not
decline below 697
/s, l
/z below the round number of 70 and at the same level as
a couple of price peaks in January.
Risking just $0.75 with a possible reward of $3.75 gave him a risk to
reward ratio of 1:5, he calculated. If everything worked as planned, he would
make a tidy sum, enough to buy the stereo.
The day after the stock broke out upward, he bought and received a fill at
71/4 (near point C in Figure 11.11). That was higher than he liked, but with
Sample Trade 181
the strength shown, he was sure the trade would work out. Scott dutifully
placed his stoploss order at 697
/s with his broker. Three days later the stock
closed at 75, above the target price. He dropped by the music store just to fon
dle the knobs and flip the switches of his dream machine.
Then things began going wrong. The stock closed down nearly $3 to
72'/s. It dropped to 713
/8 the next day and made a lower low a day later. Sud
denly, Scott was losing money and his stereo pipe dream was in danger of
plugging. Should he sell the stock and put off the party for another time?
Luck was on his side and prices began climbing again. Soon, they were at
74, but the honeymoon did not last long. Prices completed a symmetrical tri
angle but Scott did not see it. They broke out downward through the support
trendline (extend the lower right diamond diagonal toward the triangle). The
stock even gave him another chance to get out at a profit when it attempted a
pullback to the triangle boundary. Scott was busy making party plans and
missed the signal. When he received a call from his broker in midJune report
ing that the stop took him out at 697
/8, Scott scratched his head and wondered
what went wrong. Do you know the answer?

12
Double Bottoms
RESULTS SNAPSHOT
Appearance
Failure rate
Failure rate ifwaited
for breakout
Averagerise
Volume trend
Throwbacks
Percentage meeting
Surprising finding
See also
A downward price trend bottoms out, rises, then
bottoms again before climbing.
64%
3%
40%, with most likely rise between 20 and 30%
68%
68%
Bottoms closer together show larger gains
HeadandShoulders Bottoms, Complex; Horn
Bottoms
Perhaps the biggest surprise with double bottoms is the high failure rate at
64%. Only a third of the formations classify as true double bottoms. They are
the ones that have prices rising above the confirmation point, which is the
highest high between the two lows. The failure rate tumbles to just 3 % if one
waits for confirmation. Only those formations with confirmed breakouts are evalu
ated in this study.
182
' Tour 183
The average rise is 40% but is tempered by a third ofthe formations hav
ing gains less than 15%. These small rises are balanced by a third of the for
mations showing gains over 45%. The most likely rise is between 20% and
30%, relatively high for bullish formations.
Throwbacks occur 68% of the time, suggesting it is wise to wait for a
throwback and invest once prices turn upward. In some cases, waiting for a
throwback can save you from making an unprofitable trade.
A surprising finding is that bottoms closer together outperform those
spaced farther apart. The Statistics section of this chapter examines this in
more detail.
Tour
What does a double bottom look like? Figure 12.1 shows a good example of a
double bottom. Prices reach a high in midMarch then head lower. For the
next 3 months, prices continue down in a steady decline to die low inJune.
Volume picks up as prices near the low then peg the meter at over 1.1
million shares onJune 18, the day prices reach a low of 12.69. From the March
high, the stock declines 47% in 3 months. The high volume marks the turning
point and the stock moves upward. However, a retest ofthe low is in store and
prices round over and head down again. In late August, prices make another
low when the stock drops to 13.06, also on high volume.
Fleetwood Enterprises (Manuf. Housing/Rec. Veil., NYSE, FLE)
Mar 92 Apr May Jun |ul Aug Sep Oct Nov Dec
Figure 12.1 A double bottom occurs after a downward price trend. High volume
commonly occurs on the first bottom.

184 Double Bottoms W
The day after the low, on a burst of buying enthusiasm, the stock jumps
up and reaches the confirmation point in just 2 days. Instead of continuing
upward, however, the stock throws back to the breakout point and moves hor
izontally for just over a week before resuming its move upward. By late Janu
ary, the stock reaches a high of 267
/8, again of75% from the breakout price.
Figure 12.1 shows a double bottom and the gains achieved by such a for
mation. Are there key elements that make up a double bottom? Yes, and a dis
cussion of the key elements follows in the next section.
Not any two bottoms at the same price level will suffice for a double bottom.
Listed in Table 12.1 are a number of guidelines that make correct selection eas
ier. While considering the guidelines, look at Figure 12.2. The stock begins
declining in midOctober 1993 from a price of about 56l
/2. It bottoms out at
Table 12.1
Identification Characteristics of Double Bottoms
General Mills Inc. (Food Processing, NYSE, CIS)
Downward price trend
Rise between bottoms
Dual bottoms
Bottom distance
Prices rise after right bottom
Bottom volume
Breakout volume
Confirmation point
Prices trend down (short term) and should not drift
below the left bottom.
There should be a 10% to 20% rise (or more) between
the two bottoms, measured from low to high. Peaks close
together tend to be at the lower end of the range. The
rise usually looks rounded but can be irregular.
Bottom to bottom price variation is 4% or less. This is not
crucial except that the two bottoms should appear near
the same price level.
Bottoms should be at least a few weeks apart (many
consider a month to be the minimum), formed by two
separate minor lows (not part of the same consolidation
area). Minimum bottom separation is not critical as the
best gains come from formations with bottoms about 3
months apart, on average.
After the second bottom, prices must rise above the
confirmation point without first falling below the right
bottom low.
Usually higher on the left bottom than the right.
Volume usually rises substantially.
The confirmation point is the highest high between the
two bottoms. It confirms that a twin bottom formation is
a true double bottom. A breakout occurs when prices rise
above the confirmation point.
Apr 94 May |un ]ul Aug Sep Oct Nov Dec |an 95 Feb Mar
Figure 12.2 Invalid double bottom. Points A and B do not depict a double bot
tom because there are lower lows to the immediate left of point A.
about 411
/? in midMay. Prices never drop below the left low on the way to the
bottom. The reason for this guideline is that you should use the two lowest
minor lows on die price chart. Do not try to select one low then a nearby low
just to satisfy the guidelines. The two points marked A and B in Figure 12.2 rep
resent an incorrectly selected double bottom because point A has lower lows to
die left of it.
The rise between the two bottoms should climb at least 10%, as measured
from the low at the bottom to the rise high. The confirmation point is die
highest high between the two bottoms, and it is used to calculate the measure
rule and to gauge the breakout price (more about diat later). Figure 12.2 shows
a rise from the right bottom, at 415
/32, to a high of Ml
h. That is a rise of 15%,
well above the 10% threshold.
The bottom to bottom price variation should be 4% or less. The basic
rule is that the two bottoms should appear to be near one another on the price
scale. Figure 12.2 shows a price variation of about 1%.
The two bottoms should be at least a few weeks apart but are often sepa
rated by many months, as shown in Figure 12.2.1 set a 10day minimum as die
standard for selections in this study (15 days between turns out to be the mea
sured minimum for all double bottoms hi this study). A month is the minimum
separation that many professionals view as leading to powerful rallies. I set a
lower standard to help verify diat this is true. It turns out that peaks close

186 Double Bottoms
together perform better than those spaced farther apart. I limited the maxi
mum separation to about a year (the widest had a separation of 374 days).
Many of the identification guidelines are arbitrary and the classic defini
tion of a double bottom has different ones. The classic definition says that the
two bottoms should be atleast 1 month apart, separated byless than a 3% price
variation, and have a confirmation point that rises 20% above the low (bottoms
closer together have somewhat lower confirmation points). The rise between
the two troughs should look rounded.
I examined the performance difference between my definition and the
classic one and optimized the parameters to achieve the best performance.
What I discovered is that there is no meaningful performance difference
between the various settings, so I used the less stringent guidelines in the sta
tistical evaluation (10day minimum separation, 4% price variation, 10% min
imum rise to the confirmation point).
A double bottom is not a true double bottom until prices rise above the
confirmation point. In tabulating the statistics, / only count those double bottoms
in which prices rise above the confirmation point. Why? Because of the high failure
rate: 64%. There were 980 formations that looked like double bottoms, but
their price trends eventually moved below the second bottom. An additional
525 formations performed as expected by rising to the confirmation point and
continuing higher. If you buy a stock just after it touches the second bottom,
your chances ofhaving a successful trade are one in three. In other words, wait
for prices to rise above the confirmation point.
The volume chart for double bottoms usually shows the highest volume
occurring on the left bottom. Diminished volume appears on the right bottom,
and the volume trend of the overall formation is downward. None of these are
absolute rules. Sometimes volume is highest on the right bottom instead ofdie
left. However, on average, most of the formations obey the guidelines.
The breakout volume is high, usually well above the prior day's volume
and above the average volume as well. Again, this is not an inviolate rule so
expect exceptions.
Why do double bottoms form? To answer that question, consider the
double bottom shown in Figure 12.3. Prices reach a high in midApril 1993
and move horizontally until nervousness sets in during September. Then prices
start moving down, sliding from a high of 337
/g to 20'/4 by late June 1994, a
40% decline in 9 months.
After reaching a multiyear low inJune, prices recover some oftheir losses
by moving upward. After reaching a new low, a rebound is quite common with
a retest of the low typically following. A retest is just like it sounds; prices
return to the low and test to see ifthe stock can support itselfat that price level.
Ifit cannot, prices continue moving down. Otherwise, the low usually becomes
the end of the decline and rising prices result.
Such is the case depicted in Figure 12.3. It seems clear from the volume
pattern that many investors believe the low, shown as point B, is a retest of
Central and South West (Electric Utility (Central), NYSE, CSR)
May 94 |un
Figure 12.3 Prices confirm the breakout once they close above the confirmation
point, shown here as the horizontal line. The confirmation point is the highest high
reached between the two bottoms. Prices often throw back to this level after the
breakout.
point A. Volume surges on two occasions in the vain hope that the decline has
ended. Investors are wrong.
Prices hold at 21 for about a week before continuing down. As prices head
toward the level of the June low, volume surges again. This essentially marks
the end of the downward plunge. Prices hesitate at that level for slightly less
than 2 weeks before turning around and heading upward.
A double bottom is nothing more than a retest of the low. Investors buy
the stock in the hope that the decline has finally ended. Sometimes they are
right and sometimes they are not, which leads us into the next section: failures.
Focus on Failures
It is obvious that the formation pictured in Figure 12.4 is a double bottom. The
first bottom occurs after a downward price trend, as you would expect. The
two bottoms are far enough apart, the rise between them is sufficient to delin
eate two minor lows, and the price variation between the two bottoms is small.
The volume pattern is unusual in that the second bottom has a higher, denser
volume pattern than the first. However, this is not significant. After the second
bottom, prices rise at a steady rate until the confirmation point. Then prices
jump up and pierce the prior minor high at about 40s
/s. When prices close above
the confirmation line, it signals a valid breakout and confirms the double bot
tom formation.

188 Double Bottoms
In this case, as is common for most double bottoms, prices throw back to
the breakout point. However, prices continue moving down. Scrolling Figure
12.4 to the left, you would see prices making a new low in September 1993 at
31/8, below the February low of 34.
Had you purchased this stock on the breakout and held on, you would
have lost money. I call this type of failure a 5% failure. Prices do not rise by
more than 5% above the breakout before heading lower. Fortunately, 5% fail
ures are also rare; they occur only 17 times in this study. To put that statistic
in perspective, it means that on 525 separate occasions prices continue upward
by more than 5%.
Figure 12.5 shows an example ofa second type offailure that perhaps you,
too, have seen. Ted is a novice investor with an attitude. He looks at the stock
chart, checks the identification guidelines, and believes that the stock is mak
ing a double bottom. When prices rise after the second bottom, Ted decides to
pull the trigger early and buys the stock, receiving a fill at 425
/s. He reasons that
all the indications suggest the stock has completed a valid double bottom. That
being the case, why not get in now while the price is still low instead of wait
ing for prices to rise above the confirmation point (461
/*)? Ted makes a good
point. He is pleased with the stock's performance until it begins to round over.
Does he sell out now at a small profit or should he hold on and risk a downturn
while waiting for additional gains? This is a recurring investor dilemma.
He decides to hang on to his position. During May, the stock surges
upward again before beginning a downhill run. Ted watches in horror as his
profit vanishes and losses mount. Eventually, when prices spike downward, he
sells at the opening the next day and closes out his position.
Flightsafety Intl. Inc. (Aerospace/Defense, NYSE, FSI)
Jan 93 Feb Mar Apr May Jun jul Aug Sep
Figure 12.4 Example of a 5% failure. This rare occurrence happens when prices
plummet after rising less than 5%.
Statistics 189
Air Products and Chemicals Inc. (Chemical (Diversified), NYSE, APD)
|an93 Feb
Figure 12.5 Example of second type of failure—failing to wait for breakout con
firmation. Ted decided to get an early start on the double bottom but ended up
losing money.
What did he do wrong? He failed to wait for breakout confirmation.
Prices must close above the confirmation point before a trade is placed. Oth
erwise your chances of success are only one in three. We discuss how to trade
this formation properly in the Trading Tactics section of this chapter.
Statistics
The statistics shown in the following tables only refer to formations that qual
ify as being true double bottoms. This means prices must close above the con
firmation point (the highest price between the two bottoms). Table 12.2 shows
the general statistics for double bottoms. There are 542 formations in 2,500
years of daily price data. Of these formations, 372 or 69% act as reversals of the
prevailing trend.
Nearly all the formations (97%), continue moving above the confirma
tion point after an upside breakout. The remainder fail to continue upward by
more than 5% before heading back down. This statistic suggests that you
should buy the stock after an upside breakout. In all likelihood, prices will
continue rising.
The average rise for formations with successful upside breakouts is 40%.
However, the most likely rise is between 20% and 30%. Figure 12.6 shows a
graph of the gains. The chart is built by sorting the gains into various bins and
counting the number of entries in each bin. The resulting frequency distribution

190 Double Bottoms
Table 12.2
General Statistics for Double Bottoms
1991 to 1996
Failure rate
Most likely rise
Average price difference between bottoms
Average rise between bottoms
Percentage of times left bottom price is lower
than right
Cains for lower left versus lower right bottom
542
17 or 3%
40%
20% to 30%
370 or 68%
2 months (70 days)
2%
19%
52%
39% versus 39%
shows the influence large gains have on the overall average. You can see in the
chart that there are a significant number oflarge gains. Ifyou consider anything
above 50% a large gain, then almost a third of the formations fit into this cate
gory. The large gains tend to pull the overall average upward.
40 50 60
Percentage Gain
Figure 12.6 Frequency distribution of the gains for double bottoms. The most
likely rise is between 20% and 30%.
Statistics 191
The measure rule, discussed at length in the Trading Tactics section of
mis chapter, is a method used to predict the minimum price rise for a stock.
One simply computes the formation height and adds the value to the confir
mation point. The result is the price level to which it is hoped the stock will
rise at a minimum. Using this method, twothirds (68%) of the formations
meet or exceed their predicted price targets. This 68% value is low, as I con
sider values above 80% to be reliable.
The average formation length is just over 2 months (70 days) as measured
from bottom to bottom. The minimum length is arbitrarily limited to 10 days
as a selection guideline. Although I formally established no maximum length,
there were a few formations well over a year long. I removed them from con
sideration. As a result, the formation length ranges from 15 to 374 days.
The average price variation measured between the two bottoms is 2%,
half the 4% guideline maximum. The intent of this statistic is to show that the
wo bottoms appear to be on or near the same price level.
The rise between the two bottoms averages 19%, with a 10% minimum
set as a guideline. The confirmation point is the highest price reached between
the two bottoms. It is used to compute the formation height and to confirm an
upside breakout.
The two bottoms are at nearly the same price level, with slightly more left
bottoms at a lower price (52%) than right bottoms (48%). I checked the gains
for formations with either a lower left or lower right bottom and both types
perform equally well (in other words, formations with a lower left bottom have
gains of 39%, the same as those with a lower right bottom).
Do bottoms spaced closer together show larger gains than those spaced
farther apart? Yes. Consider Figure 12.7, a chart of the gains mapped onto a
frequency distribution of the distance between the two bottoms. The chart says
that if you have a bottom separation of 3 weeks, the average gain is 42%. A
larger separation, say about 4 months, gives an average gain of just 23% for
those formations in this study.
Table 12.3 shows breakout statistics for double bottoms. There is a time
delay between the second bottom and the actual breakout. The reason for the
delay is that it takes time for prices to rise from the second bottom to the con
firmation point. For the formations in this study, the delay is about a month
and a half (43 days).
Since this study only considers true double bottoms, those with prices ris
ing above the confirmation point, all 542 formations have upside breakouts.
Once an upside breakout occurs, 17 formations initially move higher but throw
back to the formation and continue down. In these cases, prices did not rise by
more than 5% before returning to the formation. This 5% failure, like that
shown in Figure 12.4, is quite rare and should not be of major concern.
There are a significant number of throwbacks to the breakout price
(68%). Such a high throwback rate suggests an investment strategy: Wait for

192 Double Bottoms
35 49 63 77
Trough Separation (days)
91
Figure 12.7 Price rise versus trough separation. Do bottoms closer together result
in the larger gains? The graph suggests the answer is yes.
the throwback, then invest after prices recover. The Trading Tactics section of
this chapter discusses this strategy further.
The average time for prices to return to the breakout price is 11 days,
which is about the same value for many formation types. No throwbacks occur
over 30 days. Throwbacks occurring beyond a month are not throwbacks at all,
just normal price fluctuations.
Table 12.3
Breakout Statistics for Double Bottoms
Average days after right bottom to breakout
Upside breakout
Throwbacks
Percentage of breakouts occurring near 12month low (L),
43 days
542 or 100%
17 or 3%
366 or 68%
11 days
7 months (204 days)
128%, C46%, H25%
L42%, C38%, H41 %
Statistics 193
Once an upside breakout occurs, how long does it take to reach the ulti
mate high? Although the average time is about 7 months (204 days), one
should temper this value by realizing it is an average. Should your particular
formation rise by only 10% or 15%, the time to reach the ultimate high likely
will be shorter—probably about 2 months.
Where in the yearly price range do double bottoms occur? Surprisingly,
most (46%) occur in the center third of the range. Mapping the percentage
gain into the yearly price range shows the average gain is about equal for all
three ranges. The result suggests that most double bottoms occur after a run
up in prices. Once prices back offtheir yearly high and trend down for a while,
they form a double bottom and begin climbing again.
You might expect that formations falling furthest (those in the lowest
third of the yearly price range) rebound higher. That appears to be the case,
with a 42% gain. You might also expect that those double bottoms occurring
near the yearly high would get caught in upward momentum and soar even
higher. Those formations with breakouts in the upper third oftheir price range
show gains of 41%.
Table 12.4 contains volume statistics for double bottoms. The first bot
tom has higher volume 58% of the time, whereas the right bottom has higher
volume 42% of the time. About twothirds of the time (65%), the volume
trend throughout the formation is downward. I computed this by finding the
slope of the linear regression line of volume over the formation from the left
to right bottom. Excluded from the calculation is the runup from the right
bottom to the breakout point because it often results in above average volume.
After a breakout, volume usually surges for a few days but quickly returns
to normal. Table 12.4 shows the average volume for the day of the breakout
until a week later. You can see how quickly the volume drops, from 165% of
the prior day's value to 18% below a week later.
Are low volume breakouts more likely to throw back to the breakout price?
No. Both high and low volume breakouts throw back to the breakout price 65%
ofthe time.
Table 12.4
Volume Statistics for Double Bottoms
Percentage of left bottoms having higher
volume
Number showing downward volume trend
Percentage of high and low volume breakouts
subject to throwback
58%
351 or 65%
165%, 133%, 105%, 92%, 89%, 82%
65%

194 Double Bottoms Trading Tactics 195
Trading Tactics
Table 12.5 shows trading tactics for double bottoms. The measure rule pre
dicts die minimum price move expected once a double bottom experiences an
upside breakout. Consider the chart pictured in Figure 12.8. To calculate the
predicted price, first determine the formation height by subtracting the lowest
low from the highest high in the formation. Tn Figure 12.8, the lowest low
occurs at the right bottom, with a price of27.57. The highest high, marked on
the figure by point A, is 31.09. Add the difference, 3.52, to the confirmation
point, or the highest high between the two bottoms (that is, 31.09 + 3.52).
Again, the highest high is point A. The result, 34.61, is the expected minimum
target. You can see in the chart that prices meet the target in late December.
A few days after meeting the target, prices momentarily descend before resum
ing their climb. During midApril, the stock reaches its ultimate high price of
40.26 before declining.
After you locate a potential double bottom, review the selection guide
lines before placing a trade. Figure 12.8 shows a declining price trend leading
to the first bottom. The rise between the two bottoms is about 13%, just above
the 10% threshold. The two bottoms are at nearly the same price level and sev
eral months apart.
Very high volume appears on the left bottom with substantially reduced
volume on the right bottom, which is typical. The overall volume trend slopes
downward from the left bottom to the right, as expected.
Once the second bottom ofa double bottom occurs, you can use the mea
sure rule to estimate the minimum price move. If the potential profit is large
enough, then wait for the breakout. This cannot be overemphasized. With a
dismal failure rate of 64%, you must wait for an upside breakout. Figure 12.5
is an example of what happens if you do not.
Bane One Corp. (Bank, NYSE, ONE)
Trading Tactic
Table 12.5
Trading Tactics for Double Bottoms
Explanation
Measure rule
Wait for breakout
Wait for throw/back
Compute the formation height by subtracting the lowest low
from the highest high in the formation. Add the difference to
the highest high between the two bottoms (the confirmation
point). The result is the expected minimum price move.
Since only 36% of the formations break out upward, you
must wait for an upside breakout before placing a trade.
Twothirds of the formations throw back to the breakout
price. Therefore, consider waiting for the throwback and for
prices to head upward again.
Apr 92 May
Figure 12.8 Double bottom trading dilemma. How do you trade this double
bottom? Do you buy just after the second bottom or wait for prices to rise above
the confirmation point? You wait for prices to recover after the throwback, then
buy. A rounding bottom appears from point A to the breakout.
A close above the confirmation point signals a breakout. The confirmation
point is simply a fancy way of saying the highest high reached between the two
bottoms. Shown is the confirmation point, marked point A in Figure 12.8, and
a line extending to the breakout point.
Once prices close above the breakout point, should you buy the stock?
Probably not. Since twothirds of the formations rise up, then quickly return
to the breakout price, it is wise to wait. Once prices complete the throwback to
the breakout point, they may continue moving down. Usually, however, they
turn around and start heading higher. In either case, wait for prices to stop
descending and begin climbing again. When that happens, buy the stock.
However, following this guideline means that you will miss some potentially
profitable opportunities.
For shortterm gains, sell as prices near the target price. Only 68% of the
formations meet their price targets, so be ready to take profits as the target
nears or if weakness intervenes.
For intermediate and longterm investors, you can hold on to the stock
and hope for an extended upward move. Of course, a review of the fundamen
tal factors supporting the price rise is often a key to large gains. Use the dou
ble bottom formation to time your entry and the fundamentals to justify a
continued presence in the stock.

196 Double Bottoms
SampleTrade
Lauren is a school teacher. Although she loves teaching kids, she would much
prefer raking in the dough by day trading stocks over die Internet. Until that
time, she shoehorns her investment activities into the few hours of free time
she has each week.
When she spotted the double bottom shown in Figure 12.8, she knew it
was love at first sight. The rounding bottom pattern (from point A to the
breakout) suggested higher highs were in store. However, she resisted the
temptation to get in early because she could not guarantee prices would con
tinue moving up. She justified her action by pretending that she was teaching
her students how to trade. If she could not do it properly, how could they?
When prices reached the high between the two bottoms, Lauren decided
to buy. Just before she placed her order, the broker read off the current quota
tion. It was well above the confirmation point. So she decided to wait and pray
for a throwback.
About 4 weeks after the breakout, prices dipped to the buy point, but would
they continue down? She had to wait until she felt confident that prices would
rise. To her, this occurred a day later, on November 27. That day prices made a
higher low and she felt comfortable buying the stock. It was a gamble, because 2
days ofrising prices hardly make a trend. Still, she was getting antsy and did not
want to wait too long and watch prices rise above the level that she could have
bought a month before. So, she bought the stock and received a fill at 313
/s.
The following day, volume spiked to over three million shares and prices
jumped over 3
/4 of a point. The spike made her nervous as it reminded her of a
oneday reversal, but the stock closed at the high for the day, which is odd for
the formation. That is when she remembered to place a stop. She chose a price
of 307
/s, about % below the recent minor lows, an area of prior support.
The following day prices moved down but succeeding days saw them
rebound. In midDecember, the stock went ballistic and fulfilled the measure
rule. She could not make up her mind if it was worth selling at that point. By
the time she decided to sell, the stock had returned to the upsloping trendline
(drawn connecting the lows in September throughJanuary), so she held on.
The stock moved up. In late March, the stock jumped sharply, climbing
almost 11
/2 in 1 day. That was a big move for the stock and she wondered what
was going on. She followed the stock closely and it became obvious the stock
had entered the bump phase of a bumpandrun reversal. Periodically, as the
stock climbed, she penciled in the sell lines parallel to the original bumpand
run reversal trendline. As she looked at the chart, she saw the narrow peak
appear and knew the end was near. When the stock dropped below the nearest
sell line, she placed an order to sell her holdings and received a fill at 385
/s.
She cleared 22% on the trade, but on an annualized basis, she made 60%.
She smiled, knowing that annualized numbers were something her math class
needed to learn. Now she had the perfect example.
13
Double Tops
Appearance
Failure rate
forbreakout
Average decline
Fullbacks
Percentage meeting
Surprising findings
See also
Two welldefined peaks, separated in time but at
nearly the same price level
65%
17%
69%
39%
Tops closer together, deep troughs, and high volume
breakouts all show larger losses.
HeadandShoulders Tops, Complex; Horn Tops
The study ofdouble tops results in several surprises. The failure rate at 65% is
worse than just tossing a coin. If you blindly sell your holdings before the
breakout, you will likely miss out on some handsome upside gains. If, instead,
you wait for the breakout, then you will be correct in dumping your shares
83% ofthe time. However, almost halfthe formations decline less than 15%,
and nearly two out of three formations fall shy of their predicted price targets.
197

198 Double Tops
Such poor performance statistics might make you think seriously about hang
ing onto your shares.
Fullbacks to the breakout point are high at 69% . Once a breakout occurs,
you can probably safely wait for the pullback to complete and prices to head
down again before selling.
Double tops spaced closer together have larger losses than those with
peaks spaced farther apart. With peaks close together, investors are more likely
to recognize a double top and try to take advantage of it.
There is also a relationship between the size of the decline between the
two peaks and the resulting percentage loss. Formations with large peakto
trough declines also have large losses when compared to small trough declines.
High volume breakouts result in statistically significant performance dif
ferences than those with low volume breakouts. In other words, formations
with high volume breakouts decline further. The Statistics section ofthis chap
ter explores these findings.
Tour
Along with the headandshoulders formation, double tops are perhaps the
most popular. Many novice investors see a dual peak on the stock chart and
proclaim it to be a double top. It probably is not. There are a number of char
acteristics that compose a true double top and I discuss them in a moment, but,
first, what does a double top look like?
Consider Figure 13.1, a double top in Pacific Scientific. The first thing
one notices are the twin peaks. They are near the same price level and widely
spaced. The price trend leading to the first peak is upward and prices fall away
after die second peak.
The intervening valley is just that: a valley that sees prices decline by 10%
or 20%, sometimes more. The valley floor forms the confirmation level or point.
The confirmation point is the lowest price between the two peaks and signals a
downside breakout once prices close below it. A twin peak formation can only
classify as a true double top once prices close below the confirmation point.
There is usually a pullback such as that shown in Figure 13.1. A pullback
allows investors another opportunity to exit their position before the decline
resumes. For more adventurous traders, the pullback is a chance to make a
short sale in the hope that prices will continue falling.
Table 13.1 contains a host of guidelines that assist in correctly identifying dou
ble tops. The general price trend is the first guideline. Prices should be trend
ing upward on their way to the first summit. The price trend should not be a
retrace in an extended decline but generally has the stairstep appearance such
V jic Scientific (Precision Instrument, NYSE, PSX)
Second Top
Sep 94 Oct Nov Dec }an 95 Feb Mar Apr May Jun
Figure 13.1 A double top has twin peaks that are usually several months apart
but quite near in price. Only when prices decline below the valley floor is a double
top confirmed as a valid formation.
Table 13.1
Identification Characteristics of Double Tops
Upward price trend
Decline between tops
Dual tops
Top distance
Prices decline after right top
Volume
Breakout volume
Confirmationpoint
Leading to the formation, prices trend upward over the
short to intermediate term (3 to 6 months) and usually
do not rise above the left top.
There should be at least a 10% decline between the two
tops, measured from high to low. Some analysts require a
20% decline with peaks spaced a month or more apart
(the deeper the trough the better the performance). The
valley is usually rounded looking but can be irregular.
Two distinct tops with a price variation between peaks of
3% or less. This is not crucial except that the two tops
should appear near the same price level and not be part
of the same consolidation pattern.
Tops should be a few weeks to a year apart. For widely
spaced peaks, use weekly charts.
After the second top, prices must dose below the con
firmation point without rising above the right top high.
Usually higher on the left top than the right. Overall
volume trend is downward.
Volume is usually high but need not be. High volume
breakouts decline further.
The confirmation point is the lowest low between the two
tops. Prices closing below the confirmation point confirm
a double top and the breakout.
199

200 Double Tops
as that shown in Figure 13.2. Ofcourse, rapid rises occur. The guideline is that
the rise should happen over several months and culminate in a top.
Figure 13.2 shows a rise that begins in October 1993 and peaks during the
following March. The climb takes prices from a low of 25l
/2 to a high of 40.
Then prices descend for a few months before climbing to the second peak. The
second peak tops out at 39, just 2.5% below the first peak.
The time between the two peaks is about 5 months, far enough apart to
form two distinct peaks with an intervening rounded, valley. The valley floor
bottoms out at 32'/z, far below the highest peak of 40. The decline, at 19%, is
well above the 10% minimum.
After the second peak tops out, prices decline away from it at a steady
rate. Soon prices close below the confirmation point, which is the lowest low
in the valley, and continue moving down. When that happens, it confirms both
the downside breakout and the formation as a double top. Prices must close below
the confirmation point before theformation becomes a true double top. This qualifi
cation cannot be overemphasized.
Prices often will decline below the second peak then turn around and
continue rising. Sometimes a third peak forms and sometimes prices just sail
away. In two out of three instances, prices will not descend to the confirmation
point at all—they just rise above the twin peaks and continue moving higher.
The Focus on Failures section discusses this behavior in more detail.
The volume pattern is what you would expect. It is usually higher on the
left peak, diminished on the second peak, and above average on the breakout.
Trinova Corp. (Machinery, NYSE, TNV)
First Top
O N 95
Figure 13.2 Double top on the weekly scale. Prices in this double top rise from a
floor of 25^ to 40 in about 5 months. Prices closing below the confirmation level
confirm a downside breakout and the formation itself.
v
Higher volume on the left peak and lower volume on the right peak help sup
port the overall receding volume trend of the formation. Figure 13.2 shows
higher volume on the left top and very low volume on the right one. Breakout
volume is at about the same rate as the first peak—high but not as enthusiastic
as it could be. Breakout volume is not a crucial factor in the validity of the for
mation, but formations with high volume breakouts tend to decline further.
Why do double tops form? Consider Figure 13.3, a wellshaped double
top that satisfies all the identification guidelines. The stock essentially begins
rising in October 1992 at a price of 97
/s. At the start, volume is unremarkable
but does have its moments. On spurts, like that shown during March and again
in April, volume spikes upward and helps propel the stock higher.
Many unfortunate investors bought near the left top hoping prices would
continue higher—a momentum play. However, astute technical investors rec
ognized the price pattern for what it really was: a measured move up. The first
upleg occurs in just 3 days. It is followed by horizontal movement for several
weeks and another swift rise to the first top.
Once the measured move completes, volume dries up and the upward
movement stalls. Prices move down and form a base in early May that sees a
low of 153
/4. The consolidation lasts almost 2 months on light turnover.
The price decline from peak to trough is not much in dollars, but it rep
resents a 20% decline. Comparatively few investors take advantage ofthe price
lull to add to their position or place new trades. Those investors that buy in at
the top swear they will sell just as soon as they get their money back. When
USF & G Corp. (Insurance (Prop./Casualty), NYSE, FG)
First Top SecondTop
Feb 93 Mar
Figure 13.3 Wellshaped double top. Prices do not push above this double top
for over 3 years. A measured move up formation forms the rise to the left top.

202 Double Tops
prices start to rise again, many of them pull the trigger and sell their shares.
The volume pattern, which up to this point has been flat, bumps up and takes
on a more rugged appearance (during lateJune and intoJuly). Other investors,
believing that the consolidation is over, buy for the first time.
As prices round over and form the second peak duringJuly some investors
correctly assume that a double top is forming. They sell their shares near the
top, content with the profits they have locked in. Other intrepid traders sell
short and hope prices fall. Prices do fall but stop at the top ofthe consolidation
area formed between the peaks a few months earlier.
After a prolonged attempt at creating a third peak in late August and into
September, prices gap below the confirmation point at 15%. A downside
breakout begins. The smart money sells their shares immediately and licks
their wounds. Others hope the selling is overdone while still others sell short.
The stock attempts a pullback in midOctober but gives up. For the next
3 years, until the end of this study, prices fail to rise above the high established
by the double top.
Focus on Failures
What does a double top failure look like and can anything be learned from it?
Consider Figure 13.4, a common failure of a double top. The twin peaks sat
isfy all the identification guidelines outlined in Table 13.1 with two exceptions.
Sep 95 Oct Apr May |un
Figure 13.4 A common double top failure. Prices decline after the second peak
then rise before reaching the confirmation point.
First, the volume pattern is suspect. Volume on formation ofthe left top is high
but lasts only 1 day. The right top volume is dense, high, and remains high for
about a week as the top forms. However, in defense of the formation, the vol
ume pattern often varies from the norm and offers little clue to the eventual
outcome.
The second guideline violated is the more important of the two. Prices
fail to close below the confirmation point. When considering all twin peak
chart patterns in this study, two out of three (65%) perform as the one shown
in Figure 13.4. In other words, they move higher. Why?
Expect top reversals (such as the double top) to perform poorly in a bull
market, whereas bottom reversals should excel. That appears to be the case
with many ofthe formations covered in this book. The key point to remember
about Figure 13.4 is that you must wait for prices to drop below the confirma
tion point before placing a trade. Otherwise, you stand a good chance of cash
ing out too soon or getting taken to the cleaners ifyou sell short.
If you do wait for confirmation, then the probability rises to 83% that
prices will continue moving down. Of course, that is small comfort if you hap
pen to run into a 5% failure. Consider Figure 13.5, a double top that obeys the
identification guidelines including closing below die confirmation level.
The uphill run starts in May 1992 and culminates in the top during
March 1993, representing a rise ofover 60%. Prices retreat for a month before
gathering steam and trying for a new high. They succeed at the beginning of
June, when prices crest the old high by 5
/8.
Figure 13.5 A double top formation that suffers a 5% failure. Prices fail to con
tinue moving down by more than 5% before rebounding.

204 Double Tops
However, the celebration is short and prices tumble. They drop over
20% before meeting support at 26. The new low is below the valley low, the
socalled confirmation point, but prices quickly turn around. Prices move up at
a smart pace and do not stop until they touch 39. That is a 50% move from the
low. If you sold your shares once prices closed below the confirmation point,
you would walk away from a chunk of money.
This type of failure is called a 5% failure. Prices break out downward but
fail to descend by more than 5% before turning around. Fortunately, 5% fail
ures are relatively rare for double tops but still represent 17% ofall formations
with confirmed breakouts.
Statistics
Table 13.2 shows general statistics for double tops, but it does not tell the com
plete story. There are 1,280 formations identified in 500 stocks over 5 years.
Out of these formations, 65% or 826 continue moving higher without first
descending below the confirmation point. Since a double top is only valid after a
confirmed breakout, excluded from the statistics are the 826 formations. In short,
they are not double tops.
The remaining 454 formations separate into reversals (75%) and consol
idations (25%) of the prevailing trend. The failure rate of these formations is
17%, meaning that 75 reverse course before moving down more than 5%. I
view failure rates below 20% as belonging to reliable formations. However, I
must emphasize that you have to wait for the breakout before investing.
Table 13.2
General Statistics for Double Tops
1991 to 1996
Failure rate
Most likely decline
Average price difference between tops
Average decline between tops
Percentage of left tops higher than right
454
11 3 consolidations, 341 reversals
75 or 1 7%
20%
10% to 15%
1 77 or 39%
2 months (57 days)
1%
15%
56% versus 44%
Statistics 205
The average decline for successful formations is 20%, but the most likely
decline is between 10% and 15%. Figure 13.6 shows this relationship. I created
the graph by sorting the percentage losses into ten bins and counting the
entries in each bin. The resulting frequency distribution shows an alarming
trend. Almost halfthe formations (47%) descend less than 15% below the con
firmation point before reaching the ultimate low.
The question then becomes, is it worth taking profits on a confirmed
double top? If prices continue down an additional 15 % and then turn around,
why not just wait for prices to recover? Those are good questions. If you sell
when the double top is confirmed, you may be selling near the ultimate low.
The average decline from the highest peak to the ultimate low is 32%.
Why not sell at the top? Why not take profits sooner, before the formation is
confirmed? If you do that, then you will most likely be giving up additional
profits as prices rise after you sell (remember, two out of three twin peak for
mations never fall below the confirmation point). Of course, you can always
sell and, if the stock does turn around, you can buy back in.
The measure rule estimates the price to which the stock will fall, at a min
imum. In the Trading Tactics section ofthis chapter I discuss the measure rule
further, but suffice it to say you simply compute the formation height and sub
tract the value from die confirmation point. The result is the minimum price
move expected.
Figure 13.6 Frequency distribution of losses for double tops. The figure shows
that almost half the confirmed double tops have declines less than 15%.

206 Double Tops
For double tops, only 39% of the formations decline far enough to fulfill
the prediction. Reliable values are above 80%. The poor showing of this for
mation further emphasizes that many double top patterns do not decline far
and this formation may not be worth trading at all.
Table 13.2 presents a few statistics related to the appearance of the dou
ble top. The average formation, as measured from peak to peak, is about 2
months (57 days) long. The average price difference between the two tops is
just $0.30 and the valley is 15% below the highest peak, on average. The left
top is usually higher than the right one but it is almost a wash (at 56% of for
mations versus 44%).
In an earlier study on a much smaller scale, I noticed a tendency of dou
ble tops to perform better with peaks closer together than with those spaced
farther apart. This study supports those findings.
I measured the time difference between peaks and created a frequency
distribution ofthe results. Then I mapped the corresponding losses for the for
mations and graphed the numbers in Figure 13.7. The graph suggests that
peaks closer together have higher losses (they perform better) than those
spaced more widely apart.
As with the earlier study, the sample size is a problem. The number of
entries in the bins is quite good until the interval gets above 77 days between
peaks. At 91 days, there are only 16 formations in that bin (followed by 15, 8,
and 8 for the next three larger intervals, respectively). Even though the 4 bins
Figure 13.7 Peak separation versus percentage loss. Peaks closer together per
form better than those spaced farther apart.
* Statistics 207
are below the minimum standard 30 entries, there are enough entries in the
odier bins to show the downward trend.
The reason for the improved performance of tops that are close together
is probably one of recognition. It is simply easier for investors to recognize
twin peaks that are 2 or 3 months apart than those that are separated by a year.
Once investors recognize a double top, they act on it as a group, sending
prices lower.
In another statistical oddity, I measured the depth of the trough between
the two peaks and compared it with the ultimate loss. I discovered that double
tops with large trough declines have larger losses than those with shallower
troughs. Figure 13.8 shows the relationship. For example, formations with
trough declines of 11% show an average loss of 15%, whereas formations with
trough declines of 18% have losses averaging 27%. The 18% trough value is
quite near the classic selection guideline of a 20% decline between tops.
Table 13.3 shows breakout statistics for double tops. After two peaks
occur, it may be many days before prices decline to the confirmation point.
The formations in this study take an average of about 5 weeks to make the
journey from the right top to the confirmation or breakout point.
As mentioned earlier, included in this study are confirmed double tops
only, which are formations in which prices close below the lowest price mea
sured between the two peaks. As such, all 454 formations have downside
Figure 13.8 Trough decline versus loss. Double tops with deep troughs perform
better than those with shallow ones.

208 Double Tops
Table 13.3
Breakout Statistics for Double Tops
Average days after right top to breakout
Downside breakout
Pullbacks
Percentage of breakouts occurring near 12month price
39 days
454 or 100%
75 or 1 7%
314 or 69%
10 days
3 months (83 days)
L24%, C50%, H26%
LI 7%, C20%, H20%
breakouts. There are, however, 75 formations that break out downward but
do not continue moving down by more than 5% before turning around.
These 5% failures are somewhat rare, occurring only 17% of the time.
About twothirds (69%) ofthe formations have pullbacks to the breakout
price. This is a high number and it suggests some reluctance ofprices to con
tinue moving below the breakout point. This belief is strengthened further by
the most likely decline being just 10% to 15%. Many times prices will drop by
10% or 15%, then pull back to the breakout point and continue higher. For
this reason, it is probably wise to place a trade after a pullback and after prices
begin heading down again. The average time for prices to pull back to the
breakout point is 10 days, which is about average for many formations in
this book.
Once a breakout occurs, it takes about 3 months (83 days) to reach the
ultimate low. The vast majority (70%) of formations reach the low in 3 months
or less. An additional 18% complete their descent in under 6 months. Thus, I
classify double tops as having shortterm investment implications.
Where in the yearly price range does the breakout occur? Most ofthe for
mations have breakouts in the center third of the price range. Substituting
performance figures in the frequency distribution, we find there is really no
clearcut winner. The percentage loss is about the same regardless ofwhere in
the yearly price range the breakout occurs.
Table 13.4 shows statistics related to double top volume. One ofthe iden
tification guidelines is that the left top has higher volume than the right top.
This occurs 57% of the time, whereas just over half (56%) show a declining
volume trend over the life of the formation (from peak to peak).
Table 13.4 also shows breakout volume when compared with the day
before the breakout. It starts out high (191% of the prior day's value) and
gradually diminishes.
Trading Tactics 209
Table 13.4
Volume Statistics for Double Tops
57%
256 or 56%
191 %, 158%, 124%,
11 3%, 116%, 112%
65%
67%
21% versus 15%.
Percentage of left tops having higher volume
Volume for breakout day and next 5 days compared with
day before breakout
Percentage of low volume breakouts subject to pullback
Percentage of high volume breakouts subject to pullback
Percentage of high volume downside breakouts resulting
in larger price declines compared to low volume breakouts
Are low volume breakouts more likely to pull back than high volume
ones? No. Only 65% oflow volume breakouts have pullbacks, whereas 67% of
high volume breakouts have pullbacks.
I define high volume as 150% ofthe day before the breakout, whereas low
volume is 75% of the prior day. If the low volume threshold drops from 75%
to 50% of the prior day's volume, then 62% of the low volume breakouts have
pullbacks. As such, you can probably argue that high volume breakouts are
more likely to pull back than lowvolume ones. This makes sense. When every
one sells their shares soon after a breakout, what is left is an unbalance of buy
ing demand (since the sellers have all sold), so the price rises and pulls back to
the confirmation point.
Do high volume breakouts send stocks lower? Yes, and the results are sta
tistically significant. Stocks with high volume breakouts suffer larger losses
(21%, on average) than do low volume breakouts (with an average 15% loss).
Trading Tactics
Table 13.5 contains suggested trading tactics. The first tactic is the measure
rule. The rule helps predict the price to which the stock will decline, at a min
imum. An example of the measure rule as it applies to double tops is shown in
Figure 13.9. The double top forms after a climb from 9!
/4 to 15'/s before
descending and climbing again to the top area. In this regard, the double top
is unusual as prices are higher 2 months before the formation tops out. Still, an
investor willing to trade this formation would first consider whether it is prof
itable to do so. That is where the measure rule comes into play.
First, compute the formation height by subtracting the highest high from
the lowest low in the formation. The highest high is the left top, at 143
/s, and

210 Double Tops
Trading Tactic
Measure rule
Do not trade
Tops close together
Deep troughs
Wait for breakout
Wait for pullback
Table 13.5
Trading Tactics for Double Tops
Explanation
Compute the formation height by subtracting the lowest low
from the highest high in the formation. Subtract the difference
from the lowest low between the two tops (the confirmation
point). The minimum price move results. Better performance
(70% versus 39% reach the target) occurs by dividing the
height in two before subtracting from the confirmation point.
With a likely decline of just 10% to 15%, is it really worth
selling your shares? If the answer is yes, then sell near the
second peak and buy back should prices close above the higher
peak (or begin a sustained uptrend).
For better performance, select tops that are closer together,
say, 60 days apart or less.
For better performance, the valley between the two tops
should be deep, 15% or more.
Since 65% of the formations break out upward, you must wait
for a downside breakout before placing a trade. On a confirmed
downside breakout, prices continue down 83% of the time.
Twothirds of the formations pull back to the breakout price.
Therefore, consider waiting for the pullback and for prices to
head downward again.
Trading Tactics 211
With such poor performance of double tops, one has to ask, why risk the
trade? Should you decide not to trade, prices will probably continue moving
higher after the second peak—especially in a bull market. About a third of the
rime, prices continue down. Sometimes stocks suffer agonizing declines, but
Figure 13.6 shows that only 4% (17 out of 379) of the formations with con
firmed downside breakouts have declines over 50%. Of course, a 30% decline
is nothing to sneeze at either.
A way to improve the performance of double tops is to sell just after the
second top forms. Most of the time, prices will rebound and move higher. Ifso,
you can always repurchase your shares and ride the stock up. Should prices
move down then you got out at the best time.
As mentioned earlier, the better performing double tops are those that
have peaks closer together with deep troughs. Figure 13.9 is an example. The
peaks are just 35 days apart and the valley descends 16% from the highest
peak. For this situation, the measure rule implies a minimum decline of 2'/4
points or 19% below the confirmation point, large enough to risk a trade.
When should you place the trade? Prices must close below the confirma
tion point before the double top is confirmed. In Figure 13.9, the little down
side spike to the left of the breakout point is not a breakout because prices do
not close below the confirmation line. Prices close below the line on February
2, a day ofvery high volume.
the lowest low is 12^8 (in the valley between the two peaks). Subtract the dif
ference, 2'/4, from the confirmation point (12 Vs) to arrive at the target price of
97
/s. Prices reach the target in late March.
Since the measure rule has a success rate of just 39%, meaning that only
about a third of the formations decline to the predicted price, it is worth con
sidering using half the formation height in the calculation. Doing so boosts the
success rate to 70%. For Figure 13.9, a new target price using half the forma
tion height is 11. Prices meet the new target the same day as the breakout.
It is always wise to check the fundamentals before placing a trade. Most
formations making a double top on weak fundamentals are ripe to fall. Unfor
tunately, the news from a company and from brokerage firms following the
company may be glowing just as the situation is about to change. You may find
brokerage firms upgrading the company or boosting earnings estimates near
the top. That is not their fault; it is just that earnings are notoriously difficult
to predict and brokerage firms get caught up in die enthusiasm.
For any trade, it is critical that you understand why the stock is perform
ing as it is. This understanding is even more important for double tops because
of the poor most likely decline (just 10% to 15%). If you discover in your
research that the fundamental factors are changing for the worse, then it gives
you added confidence to risk a trade.
Figure 13.9 Measure rule as it applies to double tops. Sell short after the pullback
once prices begin declining again.

212 Double Tops
With pullbacks occurring 69% of the time, it may pay to wait for a pull
back before investing. You can see hi Figure 13.9 that the stock pulls back to
the confirmation point and continues moving upward for several days. Once
prices begin dropping again, place the trade.
Sample Trade
Rachel is an executive secretary for a mutual fund company, one that is pros
pering. She has the intelligence and good fortune to date a few ofthe managers
running the funds. Along the way, she has picked up several investment
tips and speaks the lingo. Best of all, her buddies are still friends and willing to
help her.
When she spotted the situation shown in Figure 13.9, she asked her
friends if shorting the stock was a good idea. With their encouragement and
further research on her own, she decided to sell the stock short on March 8 and
received a fill at 12. In short order the stock drifted downward, easily fulfilling
the target price. Since prices were moving down, she felt no rush to cover her
position.
That changed on March 31 when prices closed higher. Since she usually
reviews her stocks at the end of the day, there was nothing to do but wait until
the stock opened in the morning. The following day, prices moved back down
again and she decided to wait another day.
When prices again moved up, she covered her short and received a fill at
9. She made almpst $3 of profit for each share (after commissions) or 25% in
3 weeks. Had she held on, her profits would have been even better. The stock
moved sideways for about 3 months before reaching an ultimate low of 6'/s.
14
Flags and Pennants
RESULTS SNAPSHOT
Flags
Appearance
Failure rate in uptrend
Failure rate in downtrend
Average rise in uptrend
Average decline in
downtrend
Volume trend
Fullbacks
Throwbacks
Percentage meeting
in an uptrend
Percentage meeting
predicted price
target in a downtrend
See also
A short sloping rectangle bounded by two parallel
trendlines
Shortterm (up to 3 months) consolidation
13%
12%
Downward
20%
10%
63%
61%
Rectangle Bottoms; RectangleTops
213

Pennants
Appearance
Average rise in uptrend
Average decline in
downtrend
Volume trend
Fullbacks
Throwbacks
Percentage meeting
in an uptrend
Percentage meeting
in a downtrend
See also
A short sloping triangle bounded by two converging
trendlines
19%
34%
Downward
17%
16%
58%
52%
Triangles, Symmetrical Bottoms; Triangles,
Symmetrical Tops; Wedges, Falling; Wedges,
Rising
Flags and pennants look alike and in many ways their performance is similar,
too. The formations are usually very short in duration, from a few days to 3
weeks, and mark the halfway point in a quick price move. These formations can
be profitable shortterm investments, but you must be nimble and attentive to
take full advantage of them. Pennants, with a failure rate of 34% in down
trends, are above the 20% rate I consider acceptable. Flags at 12% to 13 % and
pennants in a uptrend (19%) perform better.
The percentage of formations that meet or exceed their predicted price
targets is disappointing for both flags and pennants. I view values above 80%
to be reliable, but the results show values that range from 52% to 63%, sug
gesting that you should trade these formations with caution as your profits may
not be as large as you expect.
The most likely rise or decline is deceptive for these formations. When
the likely rise or decline value is above the average, it simply means that a fre
quency distribution shows more hits at a particular value, but the bin totals of
the prior columns are high, pulling the average downward.
Tour 215
Tour
Figure 14.1 shows a good example of a flag. It is bounded by two parallel
trendlines and usually is less than 3 weeks long (sometimes as short as a few
days). You see these formations appearing in strong uptrends or downtrends
(such as that shown in Figure 14.1), usually near the halfway point in the move.
This particular flag goes against the grain in the sense that prices rise in a
downtrend. This is the most common behavior—a retrace in a downtrend—
but it is not unusual for flags to appear horizontal (as short rectangles) or slope
downward (following the trend). Since flags can also appear in an uptrend, they
usually slope downward, but can be horizontal or slope upward too.
The volume trend is downward. As we see in the Statistics section of this
chapter, a receding volume trend usually accompanies flag formations.
Figure 14.2 shows what a pennant looks like. The only visual difference
between a flag and a pennant is the shape of the formation. Two sloping trend
lines that eventually meet outline a pennant formation, resembling a small
wedge. Sometimes the trendlines slope upward, as in Figure 14.2, and some
times they do not. Usually, they slope upward in a downtrend and downward
in an uptrend.
Like the flag formation, the volume pattern recedes. For pennants, the
receding volume trend is more prevalent, occurring in nearly all the formations
in this study.
Advanced Micro Devices, Inc. (Semiconductor, NYSE, AMD)
Sen Oct Nov
|un93
Figure 14.1 Aflag bounded by two, parallel trendlines usually has a receding vol
ume pattern.

Circuit City Stores (Retail (Special Lines), NYSE, CC)
Nov 93 |an 94
Figure 14.2 A pennant bounded by two converging trendlines looks like a short
rising wedge.
Table 14.1 outlines the identification characteristics for flags and pennants.
Two parallel trendlines bound die price action for flags as shown in Figure
14.3. Two converging trendlines outline the boundaries for pennants as shown
in Figure 14.4. In both figures die formations are short compared with many
Table 14.1
Identification Characteristics of Flags and Pennants
Prices bounded by
two trendlines
Threeweek maximum
Steep, quick price trend
Downward volume trend
Flags: price action bounded by two parallel trendlines.
Pennants: the two trendlines converge. For both patterns,
prices usually go against the prevailing trend: They rise in
a downtrend and fall in an uptrend, but exceptions are
common.
Flags and pennants are short, from a few days to 3 weeks.
Formations longer than 3 weeks may fail more often or are
better classified as symmetrical triangles, rectangles, or
wedges (rising or falling).
These formations usually form near the midpoint of a
steep, quick price trend. If you do not have a strong
advance or decline leading to the chart pattern, ignore the
formation.
Volume usually trends downward throughout the formation.
Georgia Gulf (Chemical (Basic), NYSE, GGC)
Figure 14.3 This flag appears about midway in a downtrend.
other chart patterns in this book. In die case ofFigure 14.3, the formation is 12
trading days long, whereas the pennant in Figure 14.4 is just 8 trading days
long. Many times when a formation is very short, such as 3 or 4 days, it appears
as a horizontal rectangle—a dark blob in die middle of a fast price trend. The
Figure 14.4 A short pennant forms after a quick price rise. The pennant slopes
downward and prices move upward after leaving the formation.

formations usually are shorter than 3 weeks but this is an arbitrary limit. Six
teen formations in this study (6%) have durations greater than 3 weeks (the
longest flag is 32 days and the longest pennant is 28 days).
Reliable flags and pennants appear during steep, quick price trends. The
trends might be up or down, but prices rise or fall quickly, moving several
points in just a few days to a fewweeks. In Figure 14.3, for example, the down
trend begins on January 18 and the flag begins on February 1. In that short
time, prices tumble from a high of 403
/4 to a low of 301
/g.
Although one might argue the uptrend in Figure 14.4 begins in early
April, I suggest the rise leading to the pennant begins later, on April 26, from
a low price of 223
/8. Six trading days later, the price climbs to the top of the
pennant at 313
/4. The later starting point is after the two minor highs and it
serves as a reference point for the measure rule.
Figure 14.3, the price trend in die flag slopes upward, whereas in Figure
14.4 the pennant slopes downward. This behavior is typical for the prevailing
price trend (that is, flags or pennants typically move against the trend). The
chart patterns usually appear near the midpoint of the move. As such, they are
often termed halfmast formations.
The volume trend nearly always recedes over the course ofthe formation.
However, this is not an inviolate rule but usually is the case. I should point out
that rising volume is no cause for alarm. Of the 45 formation failures, only 4
have rising volume trends.
When selecting a flag or pennant to trade, the most important guideline
is the rapid, steep price trend. If prices are meandering up or down and form a
flag or pennant, then look elsewhere. The flag or pennant must be a place
where the stock can take a breather from its rapid pace. Prices move against the
shortterm trend for several days before continuing on.
Focus on Failures
Like all formations, flags and pennants are not immune to failure. Figure 14.5
shows a flag failure. The flag, while obeying the confines of the two down
sloping trendlines, has a good volume trend. Prices should continue higher
after the flag completes but do not. Why? One explanation is that the form
ation is just too long at 26 days. Sometimes an excessively long formation
suggests an impending failure or a weak price move (after the breakout).
Trade flags or pennants more than 3 weeks long carefully or pass them up
entirely.
Figure 14.6, another flag formation, is also a failure. Prices should con
tinue rising after the flag completes. The duration is good, at 10 days (about
average for flags), and the volume trend is downward. However, the formation
has an inadequate price rise leading to it. The difference between the takeoff
..alog Device! Inc. (Semiconductor, NYSE, ADI)
Jan 96 May
Figure 14.5 A flag failure. The failure of prices to continue rising is probably due
to two factors: The price rise leading to the formation is short and the flag is longer
than normal.
Figure 14.6 Another flag formation failure. Prices rise for just 1 day before this
flag develops, much too short a rise to support a good formation.
719

point and the formation high is just over a dollar, well short of the 19% aver
age rise. There is probably little danger that you would select this formation to
trade. Since flags and pennants signal the halfway point, the predicted rise in
this example is just too small to take advantage of. An investor viewing this for
mation for trading would likely pass it by.
Figure 14.7 shows a failure of a pennant in a downtrend. The formation
probably reminds you of a short symmetrical triangle—one that acts as a rever
sal (which is unusual for a symmetrical triangle). The volume trend is receding,
as you would expect. The formation price trend, bounded by the two sloping
trendlines, looks good too. The price trend leading down to the formation rep
resents an 18% decline, exactly the average for a pennant in a downtrend.
Prices should continue moving lower after this formation completes but they
do not. Why?
You can see in Figure 14.7 that prices loop around the formation end
then head lower (a throwback). Ifyou held onto your short position, you would
eventually make money. However, I still classify this formation as a failure.
Prices should continue down immediately after piercing the trendline bound
ary. The reason prices ascend immediately is not clear. A scan of the database
reveals 66% of the formation failures (30 out of 45) fail in this manner. That is
to say, they move briefly in the wrong direction (a breakout failure) but soon
turn around (by throwback or pullback) and complete properly.
Alza (Drug, NYSE, AZA)
|an92 Feb
Figure 14.7 This pennant looks like a small symmetrical triangle. Prices
upward, throw back to the formation, and head lower.
break out
Statistics 221
Statistics
Table 14.2 shows the general statistics for both flags and pennants. I uncovered
144 flags and 106 pennants over 5 years in 500 stocks. This is fewer than I
expected.
All formations except 23 act as consolidations of the prevailing trend.
Those acting as reversals are also formation failures. Most of the failure rates
are quite reasonable. For flags, they are 12% and 13% for down and up trends.
Pennants have a wider spread, 34% and 19%, respectively. Pennants in a
downtrend fail more often than the 20% benchmark for reliable formations, so
you might consider avoiding trading them.
Nearly all the flags (87%) and many of the pennants (75%) behave as
expected, that is, prices continue in the same direction after the formation
completes as they were moving before the formation began.
Table 14.2 lists throwbacks and pullbacks for flags and pennants. Ignore
the possibility of a throwback or pullback when pondering a trade because
both happen so infrequently.
The average length, at 11 and 10 days for flags and pennants, respectively,
is quite short. As described in the Identification Guidelines, these formations
usually range from just a few days to about 3 weeks.
Seventyeight percent of the flags and 90% of the pennants have volume
trends that recede over the course of the formation. I confirmed this by exam
ining the slope of the line formed using linear regression on the volume data
from the formation start to its end. As I mentioned earlier, just because a for
mation shows rising volume is no reason to ignore it. Only 4 formation failures
(or 9%) and 41 successful formations have rising volume trends.
Table 14.3 shows the statistics for flags and pennants when the prevailing
price trend is upward, which means the trend leading to the formation is rising
Table 14.2
General Statistics for Flags and Pennants
Description
Throwbacks
Pullbacks
Downward volume trend
Flags
144
1 30 consolidations,
14 reversals
10 or 1 3%
8 or 1 2%
7 or 1 0%
12 or 20%
1 1 days
112 or 78%
Pennants
106
97 consolidations,
9 reversals
12 or 19%
1 5 or 34%
8 or 1 6%
5 or 1 7%
10 days
95 to 90%

Table 14.3
Statistics for Flags and Pennants When Price Trend Is Rising
Description
Percentage beginning near 12month
price low (L), center (C), or high (H)
Price trend duration
Average rise leading to formation
Average rise after formation
Most likely rise after formation
meeting or exceeding price target
(measure rule)
Flags
17%, C18%, H76%
62 days
19%
19%
20%
63%
Pennants
L7%, C9%, H84%
S3 days
22%
21%
15% to 20%
58%
even though the chart pattern may have a falling price trend within the con
fines of its boundary.
Where in the yearly price range do flags and pennants usually form? Most
flags begin life near the yearly high, with 76% of the formations falling into
this category (I divided tie yearly price range into thirds). For pennants, 84%
are within onethird of the yearly high.
The trend duration, from the prior minor low before the formation
begins to the minor high after the formation ends, is 62 days for flags and 53
days for pennants. Since flags and pennants are halfmast formations, and
should you enter a trade after a flag or pennant completes, you should be out
of the trade in about a month, on average. This quick investment turn means
you can make a decent profit in a short time, then look elsewhere for another
trade.
To assess how steep the price rise leading to the formation is, I calculated
the average for the chart patterns. Both flags and pennants have rises near
20%, as measured from the low at the trend start to the highest high in the
formation. I include these values in Table 14.3 because it is important to select
flags and pennants that form after a large, quick price move.
A similar move, as you would expect, completes after the breakout. Flags
(19%) and pennants (21%) show rises that almost exactly match the gain lead
ing to the formation. As such, these formations live up to the nickname ofhalf
mast formations (they mark the halfway point).
Since the numbers are averages, it pays to check what the most likely rise
is by using a frequency distribution of the gains. For both flags and pennants,
the most common rise is in the 15% to 20% range, about where you would
expect.
The measure rule, which says that the move after a flag or pennant com
pletes will meet or exceed the trend leading to it, succeeds 63% of the time for
Statistics 223
flags and 58% for pennants. I consider values above 80% to be reliable. While
both values are over 50%, meaning that these chart patterns do act as halfmast
formations most ofthe time, the values suggest that you should be conservative
when gauging the eventual price move.
Table 14.4 shows behavior statistics for flags and pennants in downtrends.
Where in the yearly price range do flags and pennants in a declining price
trend occur? Both types typically form in the center third of the range, sug
gesting that prices are near the yearly high then begin heading lower. As they
tumble into the center third of the yearly price range, a flag or pennant forms.
Then, after the formation completes, prices continue lower.
The average trend duration, at 50 days for flags and 52 days for pennants,
needs explaining. It is the average time from the beginning of the price trend
to the end. It is measured from the nearest minor high before the formation
forms to the minor low afterward. Where a trend starts and where it ends is
sometimes difficult to ascertain, so I adopted the minor highlow scheme but
allow some variation when necessary.
The average decline leading to a formation is 18% for flags and 17% for
pennants. It is measured from the highest price where the trend begins to the
lowest price at the start of the formation. The decline after a formation com
pletes is similar (17%) for both formation types.
Taken together, we discover that both chart patterns appear to form in
the center of a decline (roughly 17% on either side) that occurs over about 6
weeks, on average. Such a large decline in such a short time suggests a profit
opportunity. Trading tactics are discussed in the next section.
The most likely decline is 15% (measured from the high at the end ofthe
formation to the trend low) for flags and 25% for pennants. Although the pen
nant value is large, it simply means that there are more samples in the 25%
range but the prior ranges have a fair number of samples, too. The many small
numbers pull the overall average downward.
Table 14.4
Statistics for Flags and Pennants When Price Trend Is Declining
Description
Percentage beginning near 12month
price low (L), center (C), or high (H)
Price trend duration
Average decline leading to formation
Average decline after formation
Most likely decline after formation
(measure rule)
Flags
L24%,C60%,H17%
50 days
18%
17%
15%
61%
Pennants
L23%, C66%, H11%
52 days
17%
17%
25%
52%

The measure rule, in which the price trend after a formation meets or
exceeds the prior trend, is met 61% and 52% ofthe time for flags and pennants
in a downtrend, respectively. I consider values over 80% to be reliable so these
results fall short. Most alarming is that pennants perform so poorly. Ifyou are
considering trading a pennant in a declining price trend, be aware that prices
may not reach the target.
Trading Tactics
Table 14.5 shows trading tactics for flags and pennants. Consult Figure 14.8 as
I review the tactics listed in Table 14.5. The measure rule gauges the minimum
price move. It is the same for both flags and pennants. First, determine where
the trend begins, which is usually the minor high (for downtrends) or low (for
uptrends) preceding the formation. Figure 14.8 shows the trend beginning at
point A. Subtract the low at the formation start (point B at 423
/4) from point A
(47'/z), giving a difference of 43
/4. Subtract the difference from the high at the
formation end (point C at 43) to give the target price of 381
/4. Prices reach the
target 13 trading days after they move below the formation trendline.
When trading flags and pennants, you must first be sure you have a valid
formation. Use the identification guidelines outlined in Table 14.1 to ensure
that you have correctly identified a flag or pennant.
Use the measure rule to gauge the amount of profit likely from the trade
and weigh the amount of profit against the possible risk of failure. Look for
support and resistance levels where price trends were repulsed in the past.
Many times prices will pause or turn around at these junctions. These values
become the risk points for a trade. You can compare the risk with the reward
by computing the current price with the measure rule target and the first or
second level of support or resistance. A ratio of reward to risk should be four
to one (or higher) for highly profitable trades.
For the stock shown in Figure 14.8, the potential reward is 43
/4 (that is, 43
38V4). The first resistance level is at 44 and there is another at 45 (assuming
the trade goes against you and prices rise). The risk is one or two, that is, 44
43 or 4543. The ratio, at 4.75 to 1, suggests this formation is worth trading,
Table 14.5
Trading Tactics for Flags and Pennants
Measure rule Calculate the price difference between the start of the trend and the
formation. Prices should move at least this amount above (if in an
uptrend) or below (for downtrends) the end of the formation.
Wait for breakout Once prices move outside the trendline boundaries, place the trade.
Sample Trade 225
Murphy Oil Corp. (Petroleum (Integrated), NYSE, MUR)
Sep93 Dec Jan 94
Figure 14.8 Flags and pennants measure rule. Use the measure rule to gauge the
decline in this stock. Take the difference between the prior minor high (point A) to
the formation low at the start (point B). Subtract the value from the high at the
formation end (point C) and the result is the expected minimum price move.
providing you limit your losses. A stop placed at 44V8 or so, slightly above the
first resistance level, works well.
Take a position in the stock after a breakout, once prices move outside the
formation boundary. As prices near the target price, as predicted by the mea
sure rule, consider closing out the trade. Since the statistics regarding the suc
cess of meeting the predicted price target are so poor for these formations, be
ready to close out the trade sooner than expected. If you wait for prices to
reach the target, you might turn a profitable trade into a losing one.
Sample Trade
For example, let us say you are considering shorting the stock shown in Figure
14.8. Since the price trend is downward and it is a flag formation, the statistics
suggest that 61% of the formations will meet their price targets, on average.
That is a poor showing and deserves caution.
As die chart pattern forms, you monitor the price closely by not only
charting the endofday price but also checking it at midday. When you dial
into your broker for a midday price quote and discover that prices have moved
outside the bottom trendline, you decide to pull the trigger. You sell short and
receive a fill at 42, just above the closing price of 41 '/2.

You follow the stock closely as prices decline. You look back through the
prior year's trading history and discover two support levels at about 40 and 39.
You believe, and hope, that the stock will fall through the first support level but
the second one may be more difficult. It is, after all, closer to the 38'/4 target
price and more robust than the first level.
When the stock moves sideways at the first support level, you check your
work and reexamine the fundamentals and technical indicators. Everything
seems good so you remain in the trade.
Eventually the stock pierces the first support level and declines to the sec
ond one, where it gets stuck. It closes at 39 but the next day moves up. So the
following day you decide to close out your position, believing that the risk of a
price rise now far exceeds the possible gain. Your short sale covers at 39 and
you receive almost $3 a share. That is not a bad profit for a hold time of just 2
weeks. On an annualized basis, the return is ... wonderful!
15
Flags, High and Tight
RESULTS SNAPSHOT
Appearance
Failure rate
for breakout
Average rise
Volume trend
Throwbacks
A consolidation region of several days to several
weeks long after a stock doubles in price
32%
17%
63%, with most likely rise between 20% and 30%;
44% have gains over 50%
Downward
47%
Having recently completed the chapter on flags and pennants, I was surprised
to discover an abundance ofhigh, right flags. Even more surprising is their per
formance. The 32% failure rate is poor, but ifyou wait for an upside breakout,
the rate drops to 17%. Reliable formations have failure rates below 20%, so
high, tight flags score well providing you wait for the breakout.
The average rise, at 63%, is among the highest I have seen for any for
mation. A frequency distribution suggests that die most likely rise is a more
sedate 20% to 30%. However, 33% ofthe formations have gains over 90% and
44% have gains above 50%. A 50% gain in about 2 months is a formation worth
exploring!
227

228 Flags, High and Tight Identification Guidelines 229
Tour
Figure 15.1 is a classic example of a high, tight flag. The quick rise from the
low point at 14 to the flag high at 303
/4 takes less than 2 months. The volume
trend is downward throughout the formation. After the slight pause, the stock
continues rising. In another 2 months, it reaches a peak of 120.
The high, tight flag is a play on momentum. When a stock doubles in a
short time, it usually takes a breather and consolidates. When it does, it gives
the trader the opportunity to buy the stock before the rise resumes. How do
you correctly identify a high, tight flag?
Table 15.1
O'Neil Identification Characteristics of High, Tight Flags
O'NeilCharacteristic Discussion
Substantial rise
Flag duration
Flag correction
A rise lasting less than 2 months carries prices upward
approximately 100% to 120%.
Prices move sideways usually for 3 to 5 weeks.
During the flag phase, prices drift down a maximum of 20%.
Note: These guidelines select formations with an average gain of 69%, but only 6
formations out of 81 qualify.
The phrase, high, tight flag is a misnomer as the formation usually does not
resemble a flag formation at all. Sometimes prices move up slightly as the flag
progresses, such as shown in Figure 15.1, but more often prices spike down
briefly (a day or two) then return and move downward or horizontally before
breaking out and heading up.
The formation was popularized by WilliamJ. O'Neil in his book, How to
Make Money in Stocks (McGrawHill, 1988). In his brief introduction to the for
mation, he identifies many characteristics that high, tight flags share. Table
15.1 lists them.
Diana Corp. (Telecom. Services, NYSE, DMA)
V . .,,,,,'<''"'''*''
Jan 96
Figure 15.1 A high, tight flag that sees prices rise from about 30 to 120 in 2
months.
O'Neil's description is quite specific but it is interesting for what is does
not say. He does not mention that prices should be moving horizontally for an
extended time before the stock doubles. Some have said that this is a prerequi
site, yet the charts accompanying the O'Neil text show at least one stock in a
steady uptrend well before the 2month rise to the flag begins. The chart
shows a price low of about 26 while the flag forms at nearly 100.
I view this extended price rise as a key. It suggests that you need only look
for a stock that doubles in 2 months, then consolidates. By extension, you
could have a stock double, consolidate (forming a high, tight flag), then dou
ble and consolidate again, forming another higher flag. I found several stocks
in the database to which this situation applies. In my selections of this forma
tion, I made no assumptions about the prevailing trend (in other words, the
chart pattern need not form from a long, flat base).
Some analysts suggest volume should trend downward during the flag
phase then spurt upward when prices break out of the formation. Again,
O'Neil does not state this as a prerequisite and I do not consider it in my selec
tion criteria. However, the statistics show a downward volume trend does
improve performance.
The last omission of interest is how to trade the formation. Presumably,
once you spot a high, tight flag you would buy into the situation. Unfortu
nately, with a 32% failure rate, you may be taken to the cleaners on numerous
occasions if you follow this approach. It is safer to wait for a breakout before
trading this and most other chart patterns.
How did I select the flags in this study? I programmed my computer to
identify all stocks that have a minimum price rise of 100% in 2 months or less.
Then I manually went through each stock and looked for a nearby consolida
tion region. If the region was close to the 100% price gain, then I accepted it
as a high, tight flag. I ignored the flag duration and correction guidelines out
lined in Table 15.1. The statistics later would show that these guidelines do
contribute to performance but only to a minor degree.
Figure 15.1 passes all the O'Neil guidelines, whereas Figure 15.2 does not
(if you apply them strictly). The stock in Figure 15.2 reaches a low of 5'/4 in

Amdahl Corp. (Computers & Peripherals, ASE, AMH)
|un94 |an95 Feb
Figure 15.2 If interpreted strictly, this high, tight flag misses all but one of the
O'Neil guidelines. It sports a rise of 95% in less than 2 months (measured from the
low marked L) leading to the flag. The flag descends 22% in 38 days before break
ing out and rising 33% above the highest high in early September.
earlyJuly then starts moving up. In early September, it reaches a price of 10'/4,
just shy of doubling. Admittedly, the 95% price gain is less than a strict inter
pretation of the O'Neil guidelines, but it comes close. The high, tight flag
slopes downward for 38 days, 3 more days than die maximum and declines by
22%, 2% over the threshold. You could argue that the numbers are close
enough to the O'Neil guidelines to qualify as a high, tight flag. I accept it as a
flag but not under the O'Neil criteria for performance testing purposes. By
necessity, I changed the fuzzy phrases, such as approximately, into hard rules.
Then I compared the performance ofthose flags passing his guidelines with all
the high, tight flags. A discussion of statistics appears later in this chapter.
Returning to Figure 15.2, the gain from this flag registers 33%, well short of
the average for all high, tight flags.
Table 15.2 shows the guidelines that should be used in selecting and eval
uating high, tight flags. First and most importantly there must be a short, quick
rise. A handful (10) of the flags in the database have rises of less than 100% but
none are below 90%. Most of the stocks make the journey in less than 2
months with the longest taking an extra week (67 days). Again, the key is a
short, quick doubling of the stock price.
Once the stocks are selected on a pricerise basis, then look for the near
est consolidation area. Most of the time, it will be quite near. In my selections,
I did not care how long die stock consolidated nor how far the flag descended
Table 15.2
Identification Characteristics of High, Tight Flags
A rise lasting less than 2 months carries prices upward at least
90% (shoot for a doubling of the stock price). Stocks with 2
month rises over 115% perform best.
Locate a consolidation area, an area where prices pause in
the prevailing uptrend.
The volume trend in the flag should be receding for best
performance.
Note: My guidelines result in nearly as good performance as O'Neil's but more than six
times the number of formations qualify.
Substantial rise
Find consolidation
Receding volume trend
before turning upward. All that mattered was that the consolidation area was
plainly visible to the casual observer.
The final identification guideline is not really for identification as much
as it is for performance. Flags widi a receding volume trend handily outper
form those without. However, I would not ignore a high, tight flag simply
because volume is rising. Rather, I would recognize that its performance may
be subpar.
Focus on Failures
Investing in a stock showing a high, tight flag is not without risk. Figure 15.3
shows a flag that suffers from what I call a 5% failure. Almost a dozen forma
tions (17%) have upside breakouts but do not continue rising by more than
5%. Since one would expect a stock to move up substantially after the break
out, I consider this behavior a failure of the stock to perform as expected.
Figure 15.3 shows a quick, nearly vertical rise, leading to formation ofthe
flag. As the rise falters, high volume tapers off. When prices head lower in the
flag portion of the formation (marked in this case by two downsloping trend
lines), volume recedes. The flag drifts lower for almost a month before break
ing out of the trend and heading up. After rising for just over a week to a new
high, the stock curls around and meanders lower. It throws back to near the
base ofdie formation, then moves horizontally for several months before drop
ping lower again.
Ifyou consider the highest point in the formation as the breakout point,
men this formation is a failure (because it fails to continue rising by more than
5%). However, if you mark the point where prices pierce the top trendline
boundary as the breakout point, then this formation is a success. Ifwe use such
breakout points in this study, then all 11 ofthe 5% failures disappear (in other
words, the formations are no longer failures).

232 Flags, High and Tight Statistics 233
Tesoro Petroleum (Petroleum (Integrated), NYSE, TSO)
Breakout Point
Jan 94 Aug
Figure 15.3 A 5% failure of a high, tight flag. Prices fail to continue moving up
by more than 5% above the high reached in the flag before heading down in fail
ure. These types of failures occur 1 7% of the time for high, tight flags.
Sometimes, however, the flag portion of the formation is irregular and
the breakout point is not clear, so I decided to use the highest point in the for
mation as the breakout point. This prejudiced the performance statistics some
what but is more conservative.
Incidentally, the high, tight flag shown in Figure 15.3 obeys all the
O'Neil guidelines. Unfortunately, prices rise by just 4% before tumbling.
Figure 15.4 shows a blatant failure of a high, tight flag to rise. At first, the
stock looks like a good candidate. From its low of 12!
/2 in early October, the
stockmoves up and hits a high of27 in early December—a double in 2 months.
Then prices drift lower in the flag phase .. . and continue moving down. Even
though it meets all the O'Neil guidelines, this stock fails to break out upward
and move higher. By late March, the stock hits a new low of 11 '/2 retracing all
the gains and a little more. The figure should emphasize how risky high, tight
flags can be. Unless you wait for the upside breakout, you may be setting your
self up for a losing trade.
Statistics
It is difficult to interpret the O'Neil guidelines to assess performance. At first,
I believed I read them wrong and thought the flag length was fine unless it
grew too long (over 5 weeks). Reading his guidelines again, it appears the flag
Medimmune Inc (Drug, NASDAQ, MEDI)
Figure 15.4 This high, tight flag meets all the guidelines, but it fails to
upward. By late March, it has given up all its gains and then some.
length should be between 3 and 5 weeks long. Others have interpreted the
guideline in a similar manner.
Table 15.3 shows die performance for the O'Neil selected high, tight
flags with various interpretations of his definition along with my own varia
tions. The first table entry stipulates that a flag occurs after the stock doubles
and rises by no more than 120% of its base value in 2 months or less, has a flag
length from zero to 5 weeks, and a flag that declines by no more than 20%.
There are 38 formations in the database that meet these criteria. Nine ofthem
are failures but the remaining 29 show a rise averaging 47%.
If you use the same guidelines and wait for an upside breakout (one that
rises above the highest high in the flag), then all the failures are eliminated and
the average gain rises to 64%.
When you place a minimum limit on the flag length, your performance
rises to 69%, but you only have six formations that qualify. I do not view this
as a positive step since this formation is scarce enough without making it less
likely to occur. Ifyou remove the minimum flag length and substitute a reced
ing volume trend, then the average gain becomes 68% with 20 formations
reporting in and no failures.
Scrapping many of the selection guidelines except for a minimum 90%
rise in 2 months or less, a receding volume trend, and an upside breakout, I
found the average gain to be 65% with 40 formations qualifying. Then I
changed the minimum price rise in 5% intervals. The performance dips at

Table 15.3
Performance Statistics under Various Selection Criteria
O'Neil Criteria Cain (%) Failures/Formations
100120% rise, up to 5week flag,
maximum 20% flag decline
maximum 20% flag decline, wait
for breakout
100120% rise, up to 35 week flag,
for breakout
47
64
69
9/38 or 24%
0/29
0/6
for breakout, receding volume
My Criteria
Minimum 90% rise, receding
volume, wait for breakout
Minimum 1 10% rise, receding
68
Gain (%)
65
63
63
75
84
93
0/20
Failures/Formations
0/40
0/36
0/35
0/29
0/24
0/15
95% and 100% but then rises steadily to an average gain of 93% at the 115%
interval. The performance deteriorates beyond this point.
You can see in Table 15.3 that the number of qualifying formations also
drops from 40 to 15. Of course, you must recognize that the selection criteria
is simply tuning the performance of the database. As such, your results will
vary.
If I can locate more high, tight flags by using fewer guidelines and not
suffer any meaningful performance degradation, then why not do so? Put
another way, I removed each selection criterion from the stocks that passed the
O'Neil guidelines and found their influence to be positive but less than 4%.
For example, when I removed the stipulation that a flag must have a maximum
20% correction, the performance drops from 64% to 63%. The guideline adds
value, but it limits the number ofstocks qualifying without significandy boost
ing performance.
Statistics 235
The only guideline diat improves performance when it is removed is the
100% to 120% price rise stipulation. The best range turns out to be between
110% and 140% for the stocks in this database.
What do all these statistics mean? I view an average price rise of 65% with
40 formations qualifying as better than a 69% rise with only 6 formations mak
ing the grade (because you have more opportunities to make a killing). The sta
tistics in Tables 15.4 and 15.5 refer to my guidelines outlined in Table 15.2,
not to the O'Neil criteria.
Table 15.4 shows the general statistics for high, tight flags that follow my
guidelines oudined in Table 15.2. I located 81 formations in 2,500 years of
daily price data—a rare formation indeed.
Most of them (78%) are consolidations of the prevailing trend. The
remainder are reversals and every reversal is also a failure.
The failure rate at 32% is quite high. However, when you wait for an
upside breakout, the failure rate drops to 17%. I view failure rates below 20%
to be reliable, so you really should wait for an upside breakout before taking a
position in a stock.
The average rise is an exceedingly high 63 %. This value does not include
the idea that the flag should show receding volume as a selection guideline to
improve performance. Ifyou include such a guideline, then your performance
rises to 65%, as measured from the highest high in the flag (or die highest high
leading to it). This penalizes performance since you could compute the results
using the breakout low instead of die flag high. Doing so reduces the number
of 5% failures from 11 to zero while boosting the average gain. You would still
have four regular failures, so even this method is not perfect.
Figure 15.5 is a graph showing the frequency distribution of gains. Ifyou
ignore the rightmost column for a moment, then the columns with the high
Table 15.4
General Statistics for High, Tight Flags
1991 to 1996
Failure rate
Most likely rise
81
26 or 32%
11 or 1 7%
63%
20% to 30%
20 days
60 or 74% down
Note: These general statistics follow the guidelines in Table 15.2 with the exception of a
downward volume trend.

236 Flags, High and Tight Trading Tactics 237
33%
40 50 60
Percentage Gain
Figure 15.5 Frequenty distribution of gains for high, tight flags. The gains over
90% skew the overall average upward.
est frequencies are 20% and 30%. I consider these two ranges the gain an
investor is likely to make. However, since 33% of the formations have gains
over 90%, the most likely gain may be higher.
The average formation length is 20 days, suggesting that many ofthe for
mations do not meet the O'Neil guidelines of 3 to 5 weeks for flag length.
Many ofthe formations (74%) have receding volume trends. Six out of21
failures (or 29%) have volume trends that are rising. About half (53%) of the
successful formations having rising volume trends also show below average
percentage gains. These statistics further support the notion that a receding
volume trend is beneficial.
Table 15.5 shows breakout statistics for those formations obeying my
guidelines. Most of the formations (81%) have upside breakouts and only 17%
of these fail to continue moving higher than 5%.
The percentage of throwbacks, at 47% of formations with upside break
outs, is not high enough to formulate any sort oftrading policy. Ifyou discover
a high, tight flag that is throwing back to the formation, then wait for the
throwback to complete. The throwback must reverse and prices must move up
before you buy in, otherwise the trend may not reverse and you will end up
with a loss. The time to throwback completion is 11 days, about average with
other formations in this book.
The number of days to reach the ultimate high is a very short 2 months
(70 days). This follows the quick 2month price rise that initiated the trend.
After the stock consolidates from a few days to several weeks, the upward trend
Table 15.5
Breakout Statistics For High, Tight Flags
Upside breakout
Downside breakout
Upside breakout but failure (5% failures)
Throwbacks
66 or 81 %
15 or 19%
11 or 17%
31 or 47%
11 days
2 months (70 days)
L0%, C8%, H92%
L0%, C34%, H70%
resumes. Thus, the high, tight flag behaves like a regular flag or pennant. It is
a halfmast formation (at least in terms of time—2 months before and 2
months after the formation).
Most of the formations (92%) occur near the yearly high. Dividing the
yearly price range into thirds and distributing the performance, we find that
those formations in the highest third of the price range perform best, with
gains averaging 70%. The poor showing, at 34%, of those formations in the
center third of the yearly price range offers a stark contrast. There are no for
mations in the lowest third, as you would expect.
Trading Tactics
Table 15.6 outlines trading tactics for high, tight flags. There is no measure
rule for high, tight flags. However, almost a quarter of the formations (13 or
24%) have gains after the breakout that exceed the rise leading to the forma
tion. In other words, the high, tight flag may be near the halfway point in the
rise.
There is really only one trading tactic for high, tight, flags. Figure 15.6
shows an example of this tactic. Take a position in the stock after it breaks out
Table 15.6
Trading Tactics for High, Tight Flags
Measure rule None. The formation can act as a halfway point. Use that as the
benchmark but be conservative in your estimate.
Buy after breakout If prices break out of the flag portion, buy the stock. If you can
not tell if a breakout has occurred, wait for prices to rise above
the highest high in the flag.

238 Flags, High and Tight Sample Trade 239
,1'!
A
m,
Figure 15.6 A high, tight flag with prices stairstepping higher. How do you
trade this high, tight flag? Buy into the situation once prices rise above the break
out level. You may buy sooner, once prices stage a breakout by piercing the top
trendline, but you increase your risk of failure.
of the flag formation. This breakout is not always clear. In Figure 15.6, if you
look closely, the stock appears to break through the top trendline 3 days before
the actual breakout. In many cases, you cannot draw a straight trendline
throughout the flag formation, making it difficult to gauge when a breakout
occurs. For this reason, I use the top of the formation as the breakout point
(shown in Figure 15.6 as the breakout level). Ifyou are skeptical of this guide
line, look back at Figure 15.2. Five days after the stock reaches a new high and
a flag begins forming, the stock spikes upward. Then, just as quickly, prices
turn down again. It is over a month before prices rise above the highest high
and a lot can happen in a month! Better yet, look back at Figure 15.4 and look
at the false breakout. If you had taken a position in the stock when prices
pierced the trendline, you would have ended with a loss. This trading tactic
would have prevented you from investing in the stock.
After you take a position in the stock, hang on for the ride. The rise usu
ally will not be a straightline advance but more like a staircase. The stock rises,
consolidates, then rises again. Figure 15.6 shows this behavior. In midSep
tember, just after the volume spike, prices retrace a bit before advancing again.
Then they move up to the flag and pause. Another advance and pause brings
higher prices into December. And so on. As long as the stock is stairstepping
higher, hang onto it.
Sample Trade
John is nuts. I say that in a friendly, good sort ofway. He is unreliable, sure, but
a blast to be around. Always effervescent, bubbling with enthusiasm, he trades
stocks just as he runs his life: carefree, pedal to the metal.
When he spotted the high, tight flag shown in Figure 15.6, he wasted no
time in taking a position. When prices pierced the down trendline, he bought
in big (above point A on the chart).
He placed a stop l
/s below the formation low at 5s
/s. Two days later, he
was stopped out.
"A billion here, a billion there, and pretty soon you're talking real
money!" he grunted as he told me about the trade.
He backed off for a few days and waited for the stock to climb above the
high (6l
/2). When it did, he bit and piled into the stock again at 61
/2. He con
sidered the bottom of the flag a support area, so that is the price he used as his
stop loss. This time, however, he used a mental stop, one that is not placed with
a broker but kept in his head. There is really no problem with a mental stop
providing an investor is willing to pull the trigger when prices hit the stop.
With unreliable John, that is always a problem.
Now and again, John looked at the price chart just to see how the trade
was doing. The stock climbed to a support zone at 8 and went horizontal for 3
months. Toward the end of that time, he raised his mental stop to 73
/4.
Then, the stock climbed again. It ignored the double top formed by peaks
in early January and late February, and so did John. By April the stock posted
a new high, quietly disclosing that the double top turned out to be false. As the
stock passed 13 in midApril, John started to pay attention. He saw it reach
13 'A and back off for a bit, sinking to a low of 11 '/s. Then it spurted up again.
John drew a trendline upward following the latest move and when prices
pierced the line, he called his broker. He sold at 135
/s, not close to the high of
157
/8, but "close enough for government work," he chortled. After commis
sions, he made 108% in slightly less than 8 months.

16
Gaps
RESULTS SNAPSHOT
Appearance
Area, common, or pattern
gaps
Uptrends
Downtrends
Gap Type
Breakaway
Continuation
Exhaustion
A gap appears in the price because the current low is
higher than die prior high (for uptrends) or the
current high is below the prior low (downtrends).
90% close in 1 week. The average close time is 6 days.
Cap Type
Breakaway
Continuation
Exhaustion
Close within a Week
(%)
1
11
58
Average Days
to Close
83
70
23
Close within a Week Average Days
to Close
6
10
72
86
43
17
240
Tour 241
There are five types of gaps, four of which I review in this chapter. The
emaining gap, the exdividend gap, is not considered because it rarely happens
f
nd has no investment significance. The exdividend gap usually occurs in util
ity stocks or stocks with highpaying dividends. On the day of dividend distri
bution, the price sometimes moves downward leaving a gap in the price chart.
Even though the price of the stock after distribution reduces by the dividend
amount, the day's trading range often fills the gap so no actual gap appears on
the chart.
I define closing the gap to be when prices return and span the gap com
pletely. The area or common gap closes quickest, with 90% ofthose gaps clos
ing within a week.
Listed in the Results Snapshot tables are die average days for the gap to
close. Sometimes gaps close quickly (such as exhaustion gaps) because they are
found near the ends of trends where prices reverse and fill the gap. Other gaps
take much longer since they mark the start of a strong trend (breakaway gap).
The continuation gap is a combination of the two because it commonly
appears in the middle of trends.
Tour
A gap appears in an uptrend price series when yesterday's daily high is below
today's low price. A downtrend gap is similar, being created when yesterday's
low is above today's high. In both cases, some type ofexuberance is driving the
stock to create a gap. It sometimes is nothing more than the stock being worth
less simply because ofa dividend distribution. At other times, the repercussions
are more severe. An earnings surprise, either positive or negative, often causes
a gap and the stock to rise by 10% or 15% or to decline by 30% or more,
depending on the severity of the news.
Figure 16.1 shows a plethora of different gap types. Area or common
gaps occur in areas of congestion, usually when prices are moving sideways.
They gap up or down and close quickly. Of all the gap types, area or common
gaps are, well, common—appearing all over the place. Breakaway gaps appear
at the start of trends. They, too, are quite numerous and accompany high vol
ume. Usually there is some fundamental event driving the stock, creating a
breakaway gap. Continuation gaps are relatively rare because they appear in
the middle of strong trends. Those trends themselves do not occur very often
and even less often do they contain a gap. Exhaustion gaps signal the end of
trends. They are the last jump up or down before the trend either reverses or
moves sideways.

242 Caps
Figure 16.1 Plenty of gaps appear in a daily price chart. The most numerous are
the area or common gaps.
Table 16.1 lists identification guidelines for gaps. Area, common, orpattern gaps
are all names for the same type of gap. The gap forms inside a consolidation
region. It is easy to spot as prices seem to hook around and close the gap in less
than a week. Figure 16.2 shows many examples ofthis hook feature. For exam
ple, you can see in late March that prices gap down and the next day the high
closes the area gap. A quick hook such as that is characteristic of area gaps.
Usually, few or no new highs (for uptrends) or lows (when prices gap lower)
occur immediately after the gap.
Volume may be high on the day prices gap but usually settles down
quickly. You can see this in late January. The volume spikes on the gap day
then returns to normal the next day.
Breakaway gaps highlight the start of a new trend. Volume rises substan
tially above the prior day and prices gap upward and continue rising (or falling
in the case of a descending price gap) forming new highs (or lows).
Consider the breakaway gap in earlyJanuary shown in Figure 16.2. You can
see prices moving up for 3 days accompanied by a rising volume trend. Then
prices level out and move horizontally for several weeks before gapping down in
an area gap. Two days later, another breakaway gap (not labeled on the chart)
appears and prices reach new daily highs for 3 days in a row before settling back.
The large breakaway gap in midApril, accompanied by a high volume
spike, might be an exhaustion gap. Since prices soon continue rising (making
Table 16.1
Identification Characteristics of Gaps
GapType Discussion
Area, common, or pattern
Breakaway
Continuation, measuring,
or runaway
Exdividend
Exhaustion
Occur in areas of congestion (trendless markets) and close
rapidly. Volume on the day of the gap may be high but
returns to normal in a day or two. No new highs (uptrends)
or lows (in downtrends) occur after the gap. A distinctive
curl as the gap closes is a key indication of this gap type.
Identifies the start of a new trend and usually occurs after
breakout from a consolidation region. Is accompanied by
high volume on the day of the gap, which continues for
several days. The trend continues long enough for several
new highs (for uptrends) or new lows (downtrends) to
occur after the gap.
Happens in the midst of a straightline advance or decline.
Prices continue making new highs or lows without filling
the gap. Volume is usually high, propelling prices in the
direction of the trend.
Is triggered by a dividend distribution. Prices move down
by the amount of the dividend and a gap appears if the
day's trading range does not close it.
Occurs at the end of a trend on high volume. The gap is
not followed by new highs or lows and the gap itself may
be unusually wide. After the gap, prices enter a consoli
dation region. Commonly occurs after a continuation gap.
The gap closes quickly, usually within a week.
new highs), I have chosen to label it a breakaway gap. The same situation
occurs in earlyJune on the way down. Volume spikes upward as prices make a
large gap. Usually large gaps are associated with exhaustion gaps, but prices
continue moving lower after just a few days, so, again, I call it a breakaway gap.
Continuation gaps occur in the middle of price trends. They do not hap
pen often since it takes a sharp rise followed by a gap and a continued rise in the
stock (the reverse for downtrends, too). In Figure 16.2, you can see two contin
uation gaps in August when prices zoom from a low of 23'/2 to 325
/8 in about 2
weeks. Two continuation gaps appear in the quick, sharp price rise on high, but
not unusually high, volume. The quick rise forms new highs and the gap
remains open (compare these continuation gaps with how quickly the area gap
closes). Of course, in a downtrend, prices gap downward and form new lows.
Exhaustion gaps commonly follow continuation gaps. The highest gap in
the August uptrend is an exhaustion gap. At first I thought it was another con
tinuation gap, but the gap is slightly larger than normal and prices pause for 2
days before making new highs. Those are some key factors associated with
exhaustion gaps. Excessively wide gaps are most likely exhaustion gaps when
they appear near the end of a trend. Two exhaustion gaps appear on the chart,

244 Gaps
Figure 16.2 Various gap types with area gaps illustrating the hook feature. Vol
ume pattern and position within the trend are the main keys to identify correctly
the different gap types.
one in August and one in September. The September gap closes quickly, which
is typical for exhaustion gaps.
Most exhaustion gaps occur on high volume; it is like the last gasp before
prices end the trend. You can see in Figure 16.2 that both exhaustion gaps have
high volume, but the September one takes the cake. Volume spikes upward
even as prices descend, then volume recedes but remains high for several days
after the gap. The high volume highlights the struggle of investors who want
to purchase the stock at a good price with those who are trying to get out ofthe
situation at the best offer.
Statistics: Area Gaps
Table 16.2 shows statistics for area gaps. After 25 stocks, I stopped logging the
formations since they are so numerous. I uncovered 174 in the 5year time
span. Since these formations offer little investment interest, many performance
statistics were not collected. However, 98% of the formations have gaps that
close within 1 year. Most ofthem (90%) close within a week. The average time
to close the gap is a very short 6 days.
Due to the short nature of gaps, I chose to compare the volume figures
using a 25day moving average ofvolume. On the day prices gap, the volume
is 135% of the average volume. The following day, and for succeeding days, it
is below average. This only emphasizes the belief that many area gaps may
Statistics: Breakaway Caps 245
Table 16.2
Statistics for Area Gaps
1991 to 1996
Number of gaps closed in 1 year
Average time to close the gap
Percentage of gaps closed in 1 week
Volume for gap day and next 5 days
compared with 25day moving average
174
171 or
6 days
90%
135%, 90%, 89%, 96%, 87%, 86%
Note: I needed only 25 stocks to record a representative sample.
show high volume during the gap but then volume fades quickly. The volume
pattern often becomes an important clue to the type of gap that is forming.
Statistics: Breakaway Gaps
Table 16.3 shows statistics for breakaway gaps. Ofall the various gap types, the
breakaway gap is perhaps the most important. I uncovered 199 of them in 50
stocks over 5 years, more than enough for a good sampling. Since breakaway
gaps occur near the start of a trend, the average rise for gaps appearing in
Table 16.3
Statistics for Breakaway Gaps
Description Uptrends Downtrends
Average rise or decline
Most likely rise or decline
Days to ultimate high, low
Percentage of gaps occurring
near 12month low (L),
Percentage gain or loss for each
Percentage of gaps closed in
1 week
Volume for gap day and next
5 days versus 25day moving
average
98
25%
10% to 20%
77 days
L21%,C27%,H52%
L26%,C31%,H24%
74 or 76%
83 days
1%
197%, 144%, 133%,
139%, 120%, 143%
101
20%
10% to 15%
52 days
L23%, C29%, H48%
L25%, C18%, HI 9%
67 or 66%
86 days
6%
260%, 183%, 156%,
139%, 126%, 113%

246 Gaps
uptrends is 25% and in downtrends 20%—both quite large for gaps. The most
likely rise and decline is similar, about 10% to 15% or 20%, depending on
whether the gap occurs in an uptrend or downtrend. I arrive at the most likely
rise or decline by using a frequency distribution of the gains or losses. The bin
with the highest frequency becomes the one with the most likely gain or loss.
This method helps evaluate the skewing ofthe average due to a number ofout
sized gains or losses. The most likely gain or loss gives a more realistic perfor
mance expectation.
The average number ofdays to the ultimate high or low is 77 and 52 days
for up and down trends, respectively. Both periods are quite short, indicating
that to profit from gaps, you have to act quickly.
Separating the yearly price range into three categories, high, center, and
low, provides a way to assess where most breakaway gaps occur. Most gaps hap
pen in the upper third ofthe yearly price range, even for those that trend down
ward. When the performance overlays the yearly price range, we find that the
best performing uptrend gaps occur in the center third ofthe yearly price range,
with gains averaging 31%. For gaps in a downtrend, those gaps occurring in the
lowest third ofthe yearly price range perform best, with an average loss of2 5%.
Breakaway gaps close within 1 year 76% ofthe time in uptrends and 66%
in downtrends. The average time to close the gap is similar for breakaway gaps
in up and down trends, at almost 3 months each. The number of gaps closing
understates the actual value since some gaps occur with less than a year remain
ing in the study. Many ofthem close anyway, but a few remain open at the end
ofthe study, and no attempt was made to determine if they close within a year.
Only 1 % (uptrend) and 6% (downtrend) ofthe gaps close in 1 week. This
should come as no surprise as breakaway gaps often start a trend. Prices move
up or down rapidly and do not look back. With average gains or losses of25%
and 20%, it takes quite a while for prices to return and close the gap—certainly
more than just a week in most cases.
You can see in Table 16.3 that the volume trend for breakaway gaps
remains high throughout the week after a gap. The day on which the gap
occurs, volume essentially doubles above the average and remains high. High
volume is a key factor in identifying breakaway gaps, so consult volume ifthere
is any doubt about your identification.
Statistics: Continuation Gaps
Table 16.4 shows statistics for continuation gaps. As gaps go, these are some
what rare, occurring only 160 times in 100 stocks over 5 years. Perhaps the
most remarkable facet of continuation gaps is that they occur near the middle
of a trend. As such, their average rise is about half that of breakaway gaps, at
11 % for both up and down trends (compared with a 25% and 20% for break
away gaps, respectively).
Statistics: Continuation Gaps 247
Table 16.4
Statistics for Continuation Caps
Days to trend high, low
Percentage gain or loss for each
Number of gaps closed in
1 year
Percentage of gaps closed
in 1 week
Volume for gap day and next
5 days versus 25day moving
average
Position of gap in time trend
(trend start to gap start)
Position of gap in price trend
(trend start to gap center)
85
11%
10%
14 days
, C17%, H72%
L10%,C6%, 1111%
74 or 87%
70 days
11%
223%,165%,144%,
143%, 135%, 133%
48%
75
11%
10%
11 days
L26%, C48%, H26%
L12%, C11%, H10%
71 or 95%
43 days
10%
242%, 149%, 118%,
99%, 105%, 97%
58%
50%
The most likely rise or decline is 10% for both up and down trends.
Again, a frequency distribution of performance removes any skewing caused by
large numbers in the average.
The days to the trend high or low is about 2 weeks, quite short when
compared with breakaway gaps. You might think that since continuation gaps
appear in the middle of trends, they would be about half the trend distance to
the ultimate high (when compared to a breakaway gap that starts a trend).
With continuation gaps, I did not search for the ultimate high or low but the
trend high or low—to prove or disprove that continuation gaps appear in the
middle of trends.
Many times a short trend ends even though the overall trend is upward.
In Figure 16.1, for example, the short uptrend in the middle ofJanuary ends a
week after it begins even though the upward bias of the stock carries prices
higher until midMay. The ultimate low and high on the chart occurs in mid
January (for the low) and midMay for the high. The trend low and high both
occur in January.
For uptrends, gaps occur in the upper third of the yearly price range,
whereas for downtrends, they happen in the center third. Overlaying the

248 Caps
performance figures on the frequency distribution of gaps in the yearly price
range shows that gaps occurring in the highest third ofthe price range perform
best with average gains of 11%. For downtrends, the best performing gaps
occur in the lowest third of the price range, scoring losses averaging 12%.
Nearly all the gaps (87% and 95%) close within 1 year. Again, some gaps
do not have a chance to close before the end of the study so the value may
understate the actual results. The average time to close the gap varies quite
substantially for gaps in uptrends (70 days) versus those in downtrends (43
days). It appears easier for a falling stock to turn around and rise, thereby clos
ing a downward gap, than it is for a stock in a bull market to suddenly decline
and go against the flow to close an uptrend gap.
The percentage of gaps closing within 1 week is just 11 % and 10%. This
is normal since continuation gaps occur and prices continue moving along the
prevailing trend and do not curl around and quickly close.
The volume figures in Table 16.4 compare with a 25day moving average
of the volume. On the day of the gap, volume is more than twice normal and
remains high.
Compared with the overall trend, I measured the position of the gap in
both time and price. Timewise, the gap occurs 48% of the way to the end of
the uptrend and 58% of the way to the end for downtrends. For both up and
down trends, the trend length is computed, in days, and compared with the
trend start to the day before the gap opens.
With price, the gaps occur 48% and 50% of the way from the trend start
to the gap center for up and down trends, respectively. The numbers support
the theory that a continuation gap occurs near the middle of trends. In up
trends, prices move a little further after the gap. Since the numbers are aver
ages, your result will vary.
Statistics: Exhaustion Gaps
Table 16.5 shows statistics for exhaustion gaps. I uncovered 159 exhaustion gaps
in 100 stocks, comparatively few for all gap types. From the gap to the end ofthe
trend, the gain is 6% for uptrends and the loss is 5% for downtrends. Since the
exhaustion gap marks the end ofthe trend, these low numbers make sense. It also
follows that the most likely rise, 3%, and decline, 3% to 4%, is small.
Where in the yearly price trend do exhaustion gaps occur? Most gaps in
an uptrend occur in the upper third ofthe yearly price range, whereas the low
est third of the range scores best for exhaustion gaps in a downtrend. This
makes intuitive sense as uptrends may set a new yearly high and downtrends a
new yearly low before the trend ends. The performance of the gaps over the
yearly price range was not measured because of the small average gains or
losses registered.
Trading Tactics and Sample Trade 249
Table 16.5
Statistics for Exhaustion Gaps
Number of formations in
100 stocks from 1991 to 1996
Percentage of gaps closed in 1 week
Volume for gap day and next 5
days compared with 25day
moving average
63
6%
3%
19%, C14%, H77%
62 or 98%
23 days
58%
280%, 144%, 120%,
121%, 108%, 95%
96
5%
Between 3% or 4%
L41%, C37%, H22%
94 or 98%
17 days
72%
295%, 172%, 127%,
108%, 105%, 117%
Nearly all (98%) of the gaps close in just 23 days (uptrends) and 17 days
(downtrends). The high closing value in such a short time reinforces the belief
that exhaustion gaps occur near the end of trends. The number of gaps closed
in a week ranges between 58% for uptrends and 72% for downtrends.
Table 16.5 shows gap volume and volume up to a week later. At the start,
volume is almost triple the average but drops to just slightly above normal a
week later.
Trading Tactics and Sample Trade
Table 16.6 lists trading tactics for gaps. To successfully trade gaps you have to
be quick, making sure to use stops, and you have to be ready to close out a trade
at a moment's notice. Still, they can be profitable. Consider what Gina did with
the situation shown in Figure 16.3.
As a seasoned investor, Gina knew all about gaps and practiced trading
them on paper until she was successful most of the time. The practice honed
her skills and pulling the trigger seemed rote. With a focus on limiting her
losses, she was growing confident that her trading style was one that would
allow her to succeed in the markets, so she took the leap and decided to trade
her system for real.
She followed the stock for quite a long time and was both familiar and
comfortable with the fundamentals of the company. When she noticed the
breakaway gap occur on May 10, she quickly checked the identification guide
lines. Volume was above average (although it may not be clear from the chart)

Table 16.6
Trading Tactics for Various Gaps
Area gaps These gaps are too shortlived to be traded profitably, consistently.
Breakaway gaps If high volume is present at the start of a trend, then trade with
the trend. Verify gap type by reviewing the identification
guidelines before trading.
Continuation gaps Continuation gaps usually mark the halfway point so you can
gauge the eventual price move. Measure from the trend start to
the gap center and project the difference from the gap center to
the predicted high or low.
Exhaustion gaps If an abnormally wide gap occurs or a gap occurs at the end of a
trend, then close out your position when the trend reverses. After
a trend reversal, consider trading the new trend (shorting the
stock if the prior trend was up). Violent reversals often follow
exhaustion gaps. Close out the trade the day after new highs (for
uptrends) or new lows (downtrends) fail to occur.
Stop loss The lower rim (for uptrends) or the higher rim (for downtrends) of
a gap is a good place to put a stop ('/a or so away from the rim).
Gaps provide nearterm support or resistance, so this works well
with those gaps that do not close quickly.
Apr 96 May |un jul
Figure 16.3 Cap trading. Gina bought the stock on the breakaway gap and sold
it a few days later for a $7,500 profit. Then she shorted the stock as the exhaustion
gap turned into a deadcat bounce.
250
Trading Tactics and Sample Trade 251
and a new upward price trend seemed to be forming. She called her broker and
bought 1,000 shares receiving a fill at 58 that day.
She placed a stoploss order at 57, H below the lower gap rim just to be
safe. If this turned out to be an area gap, she would probably be stopped out for
a small loss. During her paper trading days, she discovered that most gaps pro
vide nearterm support or resistance, so she was confident that her stop would
hold.
She watched the stock closely. Two days later the stock gapped again. It
could either be a continuation gap or an exhaustion gap, she decided. Volume
was heavy, about twice the 25day moving average, so that offered no clue. The
following day, when prices gapped again, she knew that the prior pattern was
a continuation gap.
Gina checked the price chart and using the center ofthe continuation gap
as a midpoint, she measured from the trend low (point A in the figure) to the
center of the gap. The difference was 5l
/i (that is, 60!
/4 543
/4). Adding the dif
ference to the middle of the continuation gap predicted that prices would top
out at 653
/4, so she placed a stop at 65H and moments later, the stock was sold.
That day, the stock climbed to a high of 66, slightly above the predicted price,
and closed the day at 631
A.
Not including commission charges, she made $7,500 in just 3 days. But
she was not done. The large daily price range on high volume when she sold
reminded her of a oneday reversal, but she was unsure. She decided to keep
her options open and look for an opportunity to sell short. She followed the
stock daily and when it closed below the support level at 61 she decided to sell
the stock short and received a fill at 59.
The next day she was surprised to discover that a large exhaustion gap had
formed, dropping the stock down to 49, a $10,000 gain overnight. Knowing
that the gap was in reality a deadcat bounce, she changed tactics and did not
immediately close out her position. Instead, she watched the stock bounce
upward for a few days then continue lower (as the formation predicts). Instead
of getting greedy, she decided to close out her position and received a fill at 45,
for an easy $14,000 profit in less than 2 weeks.
If you think Gina was lucky, netting over $21,000 in 2 weeks, you are
probably right. But her ability to correctly size up an investment opportunity
and act on it quickly while taking steps to minimize losses goes a long way to
explaining her luck. Some call it skill.
Gina is a serious investor who leaves nothing to chance. She did not just
jump in and start trading gaps after reading about it in some book. Instead she
researched the formation and developed a successful trading style that incor
porates gaps.

17
Hanging Man
RESULTS SNAPSHOT
Upside Breakout
Appearance
Failure rate
Average rise
Downside Breakout
Appearance
Failure rate
Average decline
An opening and closing price that is at or near the
daily high with a significantly lower intraday low
Since the breakout should be downward, in a rising
price trend, 67% have upside breakouts. The 5%
failure rate is 11 %.
40%, with the most likely rise being 10%
An opening and closing price that is at or near the
daily high with a significantly lower intraday low
Consolidation with shortterm (up to 3 months)
bearish implications
Not applicable. By definition, the breakout should be
downward. The 5% failure rate is 22%.
16%, with the most likely loss between 5% and 10%
The hanging man formation is really a candlestick formation adapted to non
candlestick charts. There are two main theories about how this formation
252
Tour and Identification Guidelines 253
works. One is that when the closing price is "near" (whatever that means) the
daily high, then there is an 80% chance the following day will see a higher
high. The statistics later in this chapter show that the best performance I could
come up with is between 55% and 57%. Perhaps the theory relates to security
types other than common stocks. For stocks held overnight anyway, the 55%
chance that tomorrow's high will exceed today's is slightly better than a coin
toss and is probably a wash if you compensate for the upward bias of the stock
market over time.
The other theory, and one that the 67% failure rate pertains to in the
Results Snapshot, is when both the daily open and close are "near" the daily
high but the stock trades "significantly lower" intraday in a rising price trend.
A hanging man formation under those conditions supposedly signals a bearish
trend reversal. This only works a third of the time—a huge disappointment.
I looked at the failure rates in the more traditional sense and came up with
better values. For those formations with upside breakouts, just 11 % ofthe for
mations fail to move higher than 5% before changing direction and heading
down. On the flip side, 22% of the formations with downside breakouts are 5%
failures. This is above the suggested maximum failure rate of 20% that I con
sider acceptable.
Hanging man formations with upside breakouts show gains averaging 40%,
which is quite good, but the most likely gain is 10%—comparatively poor. For
downside breakouts, the average loss is 16% with a likely loss between 5% and
10%. Ifyou decide to trade this formation, do not expect a large price move.
Tour and Identification Guidelines
Table 17.1 shows the identification guidelines for the hanging man formation.
Ifyou have the four daily prices available—open, high, low, and close—that is
all you need. The open, high, and close should all be the same. This means that
the horizontal bar on the chart will be at the top of the figure (it will look like
the capital letter T).
Table 17.1
Identification Characteristics of Hanging Man Reversal
Characteristic Description
Open = high = close
Significantly lower low
Rising price trend
The daily open, high, and closing prices must be the same
value. On an openhighlowclose chart, it looks like the
letter T.
The intraday low price must be significantly lower than the
intraday high price. This means 5% or more.
The price trend must be rising

254 Hanging Man
The intraday low should be significantly lower than the intraday high. I
interpret the phrase significantly lower to mean more than 5%. This arbitrarily
chosen value results in a decent sample size for the statistics. Ifyou do not like
5%, then choose another value. The results do not vary that much. Changing
the benchmark from 5% to 2.5%, 7.5%, or even 10% causes the failure rate to
vary from 59% to 75%. No matter how you slice it, it is still terrible—well
above the 20% maximum.
The final criterion for selecting hanging man reversals is to find them
during an uptrend. Since we are dealing with a formation that has a lifetime
(width) of 1 day, I do not think the trend need be a large one. As long as prices
have been rising noticeably, that is good enough for me.
Figure 17.1 shows what I am talking about; there are a number of hang
ing man formations highlighted. Each daily price move in the chart shows as a
highlowclose figure; the opening price is not shown. Those prices high
lighted by the black circles have the opening price at the daily high—a true
hanging man formation.
In early August, for example, you can see that a hanging man formation
surfaces a day after prices peak at 3. The open (not shown), high, and close are
all at the same value, 3. The daily low, at 2.81, is more than 5% below the daily
high (it is significantly lower by 6.3%). The prices leading die way up either do
not have the open at the daily high or do not have an intraday low 5% below
the high. They are not hanging man formations (since the chart does not show
opening prices, you will just have to trust me).
Figure 17.1 A true hanging man formation. The black dots mark the hanging
man formations. While only the high, low, and closing prices show in the graph, the
black dots mark the days in which the opening price also matches the intraday high.
The price trend, having spurted from a low of 1.88, forms a rising trend.
That is a key element in the formation; the reversal needs something to
reverse. PointA, however, shows that the price trend leading to it is downward.
Others, like point B, suggest the beginning of an uptrend (albeit, 2 days later).
Point C is a hanging man formation that just marks time. This is probably the
most common. Point D, arguably, is a successful hanging man reversal. Prices
rise for 3 days then move flat. It is a change in trend, from a rising trend to a
horizontal one or consolidation. The move from point D to B is downward,
further suggesting a true reversal of the upward trend (although stretched out
somewhat).
Of course, my favorite is the August reversal. Unfortunately, as you can
see by the many black dots, the formation rarely works as expected even dur
ing a rising price trend.
Focus on Failures
Figure 17.2 shows two types of hanging man failures. The first one, on the left,
I use to illustrate the beliefthat a hanging man formation should show a higher
high the next day. In this example, it does not. Prices continue their downward
trend. A higher high happens only 44% to 57% of the time, depending on
whether you include the opening price in the formation or not, respectively.
The Statistics section further describes this behavior.
Roberts Pharmaceutical (Drug,ASE, RPC)
Sep Oct
Figure 17.2 Two types of hanging man failures. A higher high is expected the
day after a hanging man formation, or it should result in a reversal of the upward
price trend.

256 Hanging Man
The right formation failure illustrates a failure of the price trend to
reverse. You can see that the hanging man formation has a high of 2 !3
/4 but 2
days later the high reaches 26'/2. This performance is hardly what I would call
a reversal of the upward price trend. In this example, the formation acts as a
continuation of the shortterm uptrend.
Failures of the second type, where a price reversal is predicted but does
not appear, happen 67% ofthe time. To flip this around, it is more accurate to
say that prices will continue moving upward twothirds of the time. With such
a poor showing, I can only infer that the hanging man formation does not
apply to stocks. Maybe it works better with other security types.
With such an astronomical failure rate, I rechecked my work. In a few
formations, the hanging man pattern appears a few days before a peak (such as
that shown in Figure 17.2). Most of the time the hanging man appears in the
middle or near the beginning of an uptrend. Prices simply do not reverse after
the formation—they continue moving higher.
There is one simple thingyou can do to drop the failure rate to zero. Sim
ply do not trade this one (or rely on it).
Statistics
Table 17.2 shows the first batch of statistics. Since many of the descriptions
about a hanging man formation say the opening or closing price should be near
the intraday high, I provided a range of distances to work from. The closing
price ranges from 0% to 25% below the intraday high.
Table 17.2 provides two types offormations, a Tbar and a hanging man.
A Tbar formation is one that looks like a hanging man except the opening
price is disregarded. In the hanging man formation, the opening price is the
Table 17.2
Statistics Showing Number of Times the Daily High Price Rises above TBar or
Hanging Man Formation the Following Day
Closing Price Below
Intraday High TBar Hanging Man
0
10
15
20
25
55.4
56.4
57.0
57.2
56.6
44.4
44.7
45.4
47.0
48.0
Note: A Tbar formation is a hanging man with the opening price ignored. A hanging man
allows the opening and closing prices to drift lower but has a significantly lower intraday low.
Statistics 257
same as the intraday high price with a significantly lower low (but a rising
price trend is not a requirement).
The statistics in Table 17.2 show how often the daily high price rises
above the Tbar or hanging man formation the following day. In the Tbar
column, the number of samples is huge—the smallest sample size is over
78,000. Of course, that is on 500 stocks over a 5year period of daily price data
(2,500 years). When the closing price is at the intraday high, a higher high
occurs the next day 55.4% of the time. Even when the closing price drifts
lower, the percentage remains nearly the same with a best case of 57.2%. This
is only slightly better than a fair coin toss (50%) and probably allows for the
historical upward trend in the markets over time.
I evaluated the hanging man column on those formations where both the
opening and closing prices are near the intraday high. In the case of 0%, the
open and close match the daily high. Only 44.4% of the time is a higher high
made the next day. Performance improves after allowing the open and close to
drift downward. Almost half (48%) the formations have higher highs the next
day. Incidentally, the minimum sample size is still a massive 27,600 plus.
I created Table 17.2 to specifically test the theory that a closing price near
the daily high suggests a higher high the next day. From the results shown in
die table, I consider the theory to be unreliable—certainly nowhere near the
predicted 80% success rate. A close near the intraday high is no guarantee of a
higher high tomorrow, as far as common stocks are concerned.
Table 17.3 shows general performance statistics for the hanging man for
mation. The price trend leading to the formation is generally ignored when
examining the formation (except for failure rates discussed in a moment). I sep
arated the chart pattern into two types: upside and downside breakouts. With
over 56,000 hanging man formations in 2,500 years' worth of daily price data,
I had to make some changes to drop the sample size to a reasonable amount.
I used the same procedure for the hanging man formation as I have for
other chart patterns (such as inside and outside days). For those stocks that
have more than 10 formations, I ration the number I accept. If a stock has 100
formations, for example, I accept 1 in every 10 so that 10 formations remain,
sprinkled throughout the 5 years of daily price data. You can see in the table
that these are about evenly split (by chance) between upside breakouts (274)
and downside breakouts (261). Those formations with upside breakouts are
predominantly reversals of the shortterm price trend, whereas those with
downside breakouts act as consolidations of the trend.
The failure rate, at 67%, is the worst I have seen for any formation. In
order for the formation to qualify as a failure, it must first be a hanging man
chart pattern passing these three criteria:
1. Intraday open, high, and close are all the same value
2. Intraday low is at least 5% below the high
3. Occurs during a shortterm (or longer) rising price trend

258 Hanging Man
Table 17.3
General Statistics for the Hanging Man Formation
from 1991 to 1996, limited to about
10 formations per stock
Failure rate (good formation: an
upward trend reversal)
5% failure rate
Average rise/decline for successful
formations
ultimate high/low
Percentage occurring near
12month low (L), center (C),
or high (H)
Volume day before to day after
versus 25day moving average
Percentage gain/loss for volume
1.5x 25day moving average
O.Sx 25day moving average
Number of high volume failures
(1.5x 25day moving average)
Number of low volume failures
(O.Sx 25day moving average)
274
104 consolidations,
170 reversals
104/155 or 67%
29 or 11 %
40%
10%
261
210 consolidations,
51 reversals
N/A
58 or 22%
16%
5% to 10%
2.5 months (78 days) 1.5 months (41 days)
L66%, C17%, HI 7%
L41%, C26%, H53%
107%, 115%, 102%
28%
45%
19 or 18%
42 or 40%
L77%, C1 6%, H6%
LI 5%, C12%, H17%
107%, 115%, 102%
16%
15%
N/A
N/A
After meeting the three conditions (it is just a hanging man formation in
a rising price trend), I examine the following days to see if the trend reverses
(in other words, a downside breakout). Most ofthe time (67%) prices keep ris
ing. There is no corresponding failure rate for downside breakouts because, by
definition, a hanging man formation that works has a downward breakout
(prices move lower the next day). There are only 155 formations (of the stocks
I examined) that meet the three criteria just described and of these, 104 have
upside breakouts (the breakouts should be down, so they are failures).
Since such a high failure rate is suspicious, at least in my mind, I looked
at all hanging man formations with the 5% failure rate in mind. A 5% failure
is when prices break out in a given direction and move less than 5% before
reversing and moving significantly in the other direction. For upside break
V Statistics 259
outs, this happens 11 % of the time, whereas downside breakouts have double
the failure rate (22%). I consider anything less than 20% to be acceptable.
To assess the average rise or decline of the two breakout types of hanging
man formations, I looked at the breakout direction and removed those 104 for
mations that fail. The remaining formations show gains of 40% for upside
breakouts and losses averaging 16% for downside ones.
Shown in Figures 17.3 and 17.4 are frequency distributions of gains and
losses, respectively. For upside breakouts, die most likely gain is less than 10%.
You can see in Figure 17.3 that 15% of the formations have gains over 90%
and a third of the formations have gains over 50%. These large gains pull the
average upward.
Figure 17.4 shows the results for downside breakouts. The horizontal
scale is in 5% increments and the tallest column, the largest loss, is 10%. Since
the 5% column is quite close, I consider the most likely loss to be between 5%
and 10%.
Successful formations with upside breakouts take 78 days to reach the
ultimate high. Downside breakouts reach the ultimate low in about half the
time, 41 days. Both numbers, in comparison, make sense because it takes
upside breakouts about twice as long to go twice as far.
I determine the ultimate high or low by a significant change in trend, an
adverse move of at least 20%. Finding the ultimate high for upside breakouts,
for example, means prices rise then decline by 20% or more (from the high).
On occasion, the 20% figure is excessive, and I override it when necessary. The
40 50 60
Percentage Gains
90 >90
Figure 17.3 Frequency distribution of gains for hanging man formations with
upside breakouts. The most likely gain is 10%, the tallest column on the chart.

260 Hanging Man
20 25 30
Percentage Loss
Figure 17.4 Frequency distribution of losses for hanging man formations with
downside breakouts. The most likely loss is between 5% and 10%.
same methodology applies to downside breakouts. Such a large swing accom
modates normal price behavior and quickly identifies significant trend changes.
Does a hanging man formation occur near the yearly high or low? Both
upside and downside breakouts occur most frequently within a third of the
yearly low. When you overlay performance on the yearly price range, the best
performing upside breakouts are those occurring within a third of the yearly
high, scoring average gains of 53%. Downside breakouts are more evenly split
but the highest return made by those formations appears within a third of the
yearly high (with an average 17% loss).
The volume during a hanging man formation is higher than average. You
can see in Table 17.3 that the formation itself has a volume trend that is 15%
above normal (115% of the 25day moving average). Both the day before and
the day after the formation show high volume.
Do hanging man formations with high volume produce superior results?
No. When volume is 50% (1.5x in the table) or more above average, the aver
age gain is only 28%, well below the 40% scored for upside breakouts. How
ever, when the volume is 50% below average, the performance improves
dramatically: 45%. For downside breakouts, the performance hovers around
the 16% average gain.
Does volume relate to the failure rate? To answer this question, I sepa
rated the formations with upside breakouts into three categories: volume that
is 50% above average, 50% below average, and everything else. The majority
ofthe failures occur between the two 50% ranges. When volume is 50% below
Trading Tactics 261
the 25day moving average, 40% of the formations fail. Only 18% fail with
high volume. These numbers suggest that you should be watchful of a hanging
man formation in a price uptrend that shows below average volume. It may
continue moving up instead of reversing.
Trading Tactics
After careful consideration, I cannot recommend trading this formation. The
primary belief behind this chart pattern is that prices will reverse the uptrend.
They do not. Just a third of the formations reverse, whereas the others see
prices continue higher.
The only advice I can offer is when you are considering buying or selling
a stock and see a hanging man formation. It is probably best if the opening
price is well below the intraday high but the stock closes at the high, suggest
ing upward momentum. The following day, there is a very slight tendency to
post a higher high. So, ifyou are selling, you might wait for the new high or at
least follow the stock closely throughout the day.
Ifyou are buying just before the close and the price is at or near the intra
day high (again, with the opening price near the intraday low), you might take
comfort in believing tomorrow's price will be higher—at least sometime dur
ing the day. The odds suggest a higher price, but the odds are not much better
than a coin toss, certainly not worth betting the farm on.

18
HeadandShoulders
Bottoms
RESULTS SNAPSHOT
Appearance
Failure rate
Average rise
Volume trend
Throwbacks
Percentage meeting
See also
Threetrough formation with center trough lower
than the others
5%
Downward; usually higher on the left shoulder than
the right
52%
83%
HeadandShoulders Bottoms, Complex
I find it easier to pick out tops than bottoms. Perhaps this is because I spend so
much time worrying about when to sell. Placing a trade is easy but getting out
is the tough part. In my quest to sell at the appropriate time, I have often over
looked the buy side: bottom reversals. Headandshoulders bottoms are just
such a formation. They are quite easy to spot and can be very profitable.
262
Tour 263
The Results Snapshot highlights statistics for this bullish reversal. Like
the top version of the formation, the bottom sports an exceedingly low failure
rate of 5%. Only a few formations either fall or climb by less than 5%. Those
that do experience an upside breakout continue rising by an average 38%.
Like many bullish formations, the headandshoulders bottom meets its
price target often: 83 % ofthe time. I consider values above 80% to be reliable.
Tour
What does the formation look like? Figure 18.1 shows a good example of a
headandshoulders bottom. The stock starts rising in November 1993 and
peaks during February, where the figure begins. From that point, the stock
moves downward and makes a lower low in late March before moving up. The
turn marks the left shoulder. The stock declines again and reaches a new low
during late April, forming the head. The right shoulder appears as the stock
recovers then continues moving down along the trendline (shown in Figure
18.1 as the neckline).
The stock advances above the neckline and stages an upside breakout.
However, the rise does not last long. Prices soon decline below the level of the
right shoulder. The stock moves sideways over the next 4 months. Then the
stock enters another headandshoulders bottom and the upside breakout
proves more lasting. By midAugust 1995, the stock is trading just below 60.
Arrow Electronics Inc. (Electronics, NYSE, ARW)
Feb94 Apr May |un
Figure 18.1 A headandshoulders bottom. Two shoulder troughs surround a
lower head. Volume is usually higher on the left shoulder than on the right shoulder.

264 HeadandShoulders Bottoms
The headandshoulders bottom shown in Figure 18.1 has a somewhat
unusual volume pattern. Volume is usually highest on the left shoulder, dimin
ished on the head, and even lower on the right shoulder. The rise from the
head to the right shoulder accompanies a rise in volume as does the actual
breakout.
In contrast, the formation shows little increase in volume during the rise
from the head to the right shoulder. Volume on the breakout is unexciting and
that helps explain why the stock stalls. Upward momentum fails to happen
quickly enough to propel the stock higher; the stock rounds over and heads
back down.
Figure 18.2 shows a headandshoulders formation on a weekly time
scale. I chose this chart to show you the typical trend of headandshoulders
bottom reversals. They usually form after an extended downtrend in prices. As
a reversal, once they complete, prices rise.
Why do headandshoulders bottoms form? The formation represents a
struggle to find the bottom, the lowest price that represents the best value. As
the stock descends during February 1994, investors nibble at the stock in
increasing numbers. Volume climbs even as the stock descends until it spikes
upward for 1 week during formation of the left shoulder. Buying demand puts
a crimp on the downward slide and prices move up but only for a week. The
following week, prices move lower. Again, volume spikes as the stock makes a
Alien Telecom Inc. (Telecom. Equipment, NYSE, ALN)
A S O N D 9 5 F M A M ) I A S
Figure 18.2 Headandshoulders bottom formation on a weekly time scale. It
takes several months before this headandshoulders bottom stages an upside
breakout. Volume is characteristic: highest on the left shoulder, diminished on the
head, and exceedingly low on the right shoulder.
new low and this becomes the head. The smart money is accumulating the
stock in anticipation of an eventual rise or a change in the fundamentals. The
stock moves up on receding volume then retreats and forms the right shoulder.
Volume on the three troughs is diminishing. The left shoulder has very
high volume, the head exhibits somewhat less volume, and the right shoulder
records the lowest volume up to that point. Only after prices start moving up
from the right shoulder does volume spike upward.
Breakout volume, depending on where you determine the breakout
occurs, is unconvincing. In late August, prices move decidedly above the neck
line and stage a definitive breakout. Even so, it is not until 2 weeks later that
volume advances noticeably.
Table 18.1 encapsulates the identification guidelines for a headandshoulders
bottom. Consider Figure 18.3, a headandshoulders bottom. The formation
does not appear at the end of a longterm downtrend but at a shortterm one
(up to 3 months). The uptrend begins the prior June with another headand
shoulders bottom. The formation reverses the slight shortterm downtrend
but continues the longterm uptrend.
Overall, the formation sports the three telltale troughs: left shoulder,
head, and right shoulder. The left shoulder is at about the same price level as
Table 18.1
Identification Characteristics of a HeadandShoulders Bottom
Shape A threetrough formation with the center trough below the
other two. It looks like a headandshoulders bust flipped upside
down. The three troughs and two minor rises should appear
well defined.
Symmetry The left and right shoulders should be opposite one another
about the head, somewhat equidistant in both time and price.
There are wide variations but the formation is noticeably
symmetrical about the head.
Volume Usually highest on the left shoulder or head and diminished on
the right shoulder.
Neckline A line that connects the rise between the two shoulders. A
piercing of the neckline signals an upside breakout. Ignore the
neckline if the slope is too steep. In such a case, use the highest
rise between the shoulders as the breakout level.
Upside breakout The breakout is upward, usually on high volume that powers
prices upward. A low volume breakout is not an indicator of an
impending failure.

266 HeadandShoulders Bottoms "
3 Com Corp. (Computers & Peripherals, NASDAQ, COMS)
40
Lett Shoulder Head Right Shoulder
Feb 95 Apr May jun |ul
Figure 18.3 A rare headandshoulders consolidation of the primary uptrend.
the right one and appears to be about the same width. Such symmetry is com
mon in headandshoulders formations (tops, bottoms, and the complex vari
ety). If the left shoulder is sharp or pointed, the right shoulder will be too.
The head is below both shoulders by a reasonable amount. By this char
acteristic I mean the formation is not a triple bottom—three troughs at about
the same price level.
In Figure 18.3, the left shoulder suddenly declines for 3 days, then
reverses and climbs to a minor high. Similarly, the rise between the head and
right shoulder climbs almost to the height of the rise between the left shoulder
and head then descends to the right shoulder. All five features, the three
troughs and two minor rises, appear well defined and distinct. The features are
important as you scan your charts looking for headandshoulders bottoms.
Symmetry is another important key to selecting a valid headandshoul
ders bottom. The right side of the formation usually mimics the left side. The
right shoulder declines to about the price level of the left shoulder and die dis
tances of both from the head are similar. Of course, there are many variations,
but symmetry should make a headandshoulders bottom stand out from a
sequence of any three depressions.
Volume represents anodier clue to the validity of a bottom. The left shoul
der typically has the highest volume, followed by the head, with diminished vol
ume on the right shoulder. Thus, overall, the volume trend is downward; higher
on the left side of the formation than the right—until the breakout.
The neckline is an imaginary line connecting the two rises between the
shoulders and the head. It can slope downward or upward. In wellformed for
mations, the slope of the line is not very steep, but a steep neckline should not
be a disqualifier of a headandshoulders bottom (see Figure 18.1—it has a
rather steep neckline).
Irregular volume patterns should also not disqualify the formation. Fig
ures 18.1 and 18.3, for example, have volume that is highest at the head.
Breakout volume is usually high as it pushes prices above the neckline.
However, in a quarter of the formations where prices continue higher, break
out volume is well below the day before the breakout. We see in the Focus on
Failures section that high breakout volume accompanies most failures. As a
rule, volume will rise on the day of the breakout, but it need not.
Focus on Failures
Like most formations, there are two types of failures. The first type, shown in
Figure 18.4, is a failure of the headandshoulders bottom to pierce the neck
line and move higher. As you would expect, the formation appears after a
downtrend in prices. The highest price peak is partly visible in the upper left
corner of Figure 18.4. From the high of 383
/4, prices fall to the low at the head,
2ll
/4, a decline of 45%. When the bottom forms, it should signal a trend reversal.
An interesting thing about the formation in Figure 18.4 is that the left
shoulder is almost the same shape as the right. Only a dollar separates the
Mar 94 Apr May |un |ul Aug Sep Oct Nov Dec
Figure 18.4 Failure of a headandshoulders bottom to stage an upside breakout.

268 HeadandShoulders Bottoms V '
two shoulder lows and the head is well below both shoulder troughs. The
right shoulder is somewhat farther away from the head than the left. This
characteristic is typical.
Volume is suspiciously low throughout the formation. The left shoulder
and head register about the same level of volume. The right shoulder volume,
however, is higher than the other two. Of course, an irregular volume pattern
is no reason to discard a formation—but it serves as a warning.
After the right shoulder forms and prices begin rising, volume tapers off
rapidly and the attempt to pierce the neckline fails. The rally attempt does not
even come close to the neckline.
Looking at the overall formation, there is no one item that signals an
impending failure. There is some suspicious activity, principally the abnormal
volume pattern, but nothing to deter an investor.
Figure 18.5 shows a slightly different type of failure. This is what I call a
5% failure. The two shoulders and head appear well formed and distinct. The
left shoulder looks different from the right, but the twin rises between the
shoulders are similar. The price level of the two shoulders is not suspiciously
out of line.
Volume is unusual. The only heavy volume appears near the head as
prices rise away from it toward the right shoulder. The right shoulder volume
looks like something you would want to tackle with your shaver: annoying but
not high enough to be alarming.
Airborne Freight (Air Transport, NYSE, ABF)
'I
Figure 18.5 A 5% failure in a headandshoulders bottom. Prices must rise by at
least 5% before the formation is a success. A 5% rise should take prices to 393
/s but
it does not happen.
Statistics 269
Prices advance smartly after the right shoulder forms. Once prices rise
above the stairstep incline, they zoom upward for 3 days and then stop. The
stock moves essentially sideways for 2 weeks before starting back down.
Although this formation does have an upside breakout, prices fail to rise
by more than 5% above the neckline. Prices should reach 393
/8 to meet the 5%
threshold, but they do not. The result is a failure of the 5% rule: Prices must
rise by more than 5% after a breakout or the formation is a dud.
I went through the various failures in the database and examined them to
see if there is any truth to the notion that low volume breakouts are subject to
failure. I found that this simply is not true. Of the 18 formation failures, only
8 (44%) occur after a low volume breakout. However, the sample size is small
(30 samples usually provides reliable results). These numbers conveniently
bring us to die next section: Statistics.
Statistics
Table 18.2 contains general statistics for headandshoulders bottom forma
tions. There is a good assortment of formations, 330 to be exact, in 2,500 years
of daily price data. Most of the formations, 85%, act as reversals, whereas the
remainder are consolidations ofthe prevailing trend. Almost all the formations
(95%) perform as expected. This means they break out upward and continue
higher by more than 5%. The average rise after an upside breakout is 38%.
Table 18.2
General Statistics for HeadandShoulder Bottoms
Number of formations in 500 stocks from 1991
to 1996
Failure rate
Most likely rise
Number of successful formations showing
downward volume trend
Average rise of upsloping neckline versus down
sloping neckline
Average rise of higher right shoulder versus higher
left shoulder
330
18 or 5%
38%
20% to 30%
258 or 83%
193 or 62%
38% versus 40%
36% versus 41 %
Note; With a 5% failure rate and an average rise of 38%, the headandshoulders bottom
is a formation worth considering.

270 HeadandShoulders Bottoms V'
Since large numbers can skew averages, I computed a frequency distrib
ution of the percentage gains. Figure 18.6 shows the results. The lopsided
bellshaped curve seems to be typical for most formations and the headand
shoulders bottom is no exception. Notice the large number ofgains over 90%.
These tend to pull the overall average upward. The frequency distribution
shows the most likely rise is between 20% and 30%, down slightly from the
average gain of 38%.
In the Trading Tactics section of this chapter I explain the measure rule.
It is a way for a formation to predict the minimum price move. For the head
andshoulders bottom, the prediction succeeds 83% of the time, a comforting
number (I view anything above 80% to be reliable).
The formation length from left shoulder trough to right shoulder trough
is 21
/: months. This understates the formation length somewhat because it does
not include the run down to the left shoulder and the rise to the breakout. Oc
casionally, it can take several weeks or months before prices stage a breakout.
About twothirds (62%) of the formations show a downward volume
trend. The slope of the linear regression line between the shoulder lows mea
sures this. The result fits with casual observation in that the left shoulder usu
ally has the highest volume, followed by the head and greatly diminished right
shoulder volume.
Does the slope of the neckline or the shoulder height predict the magni
tude of the resulting rise? I looked into these questions and found results dif
ferent from what I expected. For necklines in which the slope of the line is
Figure 18.6 Frequency distribution of gains for headandshoulders bottoms. The
most likely rise is between 20% and 30%.
Statistics 271
upward (the rise on the left is below the rise on the right), the stock has an aver
age gain of 38%, versus 40% for those stocks with downsloping necklines.
In a similar manner, I looked at the higher of the two shoulders. Does a
higher right shoulder suggest a larger gain? No. The average rise ofstocks with
higher right shoulders is 36% while those with higher left shoulders gain an
average of 41%.
For both necklines and shoulder height the results are not statistically sig
nificant. This means the results could be due to chance, or there could be some
veracity to the difference.
Table 18.3 shows statistics related to the breakout. Nearly all (98%) of
the headandshoulders bottoms have upside breakouts. There are only eight
formations that have downside breakouts. Of those formations with upside
breakouts, 10 fail to rise by more than 5%.
Throwbacks, when prices break out upward then return to the neckline,
occur 52% of the time. This suggests some hesitancy of the formation to rise.
It takes, on average, 11 days for the stock to return to the neckline and com
plete a throwback. This seems to be about average for many formations in
this book.
Is a formation more likely to throw back after a low volume breakout?
No. I separated the formations into two columns: those that have breakouts
with high volume (over 125% of the prior day) and those with low volume (less
Table 18.3
Breakout Statistics for HeadandShoulders Bottoms
Upside breakout
Downside breakout
Throwbacks
Is a formation more likely to throw back after
a low volume breakout?
12month low (L), center (C), or high (H)
Percentage gain for each 12month lookback
period
Successful breakouts on high volume
Successful breakouts on low volume
Formation failures on low volume
322 or 98%
8 or 2%
10 or 3%
167 or 52%
11 days
No
7 months (215 days)
L29%, C41%, H30%
1.37%, C34%, H44%
163%, 125%, 104%, 95%, 97%, 95%
1 64 or 74%
59 or 26%
8 or 44%
Note: The vast majority of breakouts are upside breakouts.

272 HeadandShoulders Bottoms ^
than 75% of the prior day). Then I matched the throwbacks with the two
columns. The results split evenly at 49% for each column. The results suggest
that a throwback is independent ofbreakout volume.
After a breakout occurs, it takes about 7 months to reach the ultimate
high. However, a frequency distribution ofthe time to reach the ultimate high
shows that most formations land in the shortterm category (up to 3 months).
I therefore classify this formation as having shortterm trading implications.
Where in the yearly price range does the formation occur? Most break
outs from a headandshoulders bottom occur in the center third of the yearly
price range. The breakout happens near the top of the formation and explains
why there are not more occurrences in the lowest third of the price range.
When we distribute the percentage gains over the same yearly price range, we
find that formations with breakouts in the top third of the yearly price range
tend to gain the most, 44%. This finding suggests that momentum players
grab hold of the stock and bid it up.
Average breakout volume is 63% above the prior day (163% of the total)
but rapidly recedes over the course ofa week. I looked at breakout volume as a
function of high volume (125% of the prior day) and low volume (75 % of the
prior day). Almost three out of four breakouts occur on high volume. Still, that
leaves 26% of the formations with successful upside breakouts on low volume.
Even though a formation may break out on low volume is no reason to suspect
the formation will ultimately fail. As mentioned previously, I looked at the 18
formation failures and 44%, less than half, have failures occurring after a low
volume breakout.
Trading Tactics
Table 18.4 discusses trading tactics for headandshoulders bottoms. Use the
measure rule to predict the minimum price move once prices break above the
neckline. In Figure 18.7, the head marks the lowest price in the formation.
Subtract its price from the value of the neckline at that point. In this example,
the head has a daily low price of 13l
/s and the neckline, measured vertically, is
at 17'/2. Add the difference, 43
/s, to the price where the stock closes above the
neckline. This occurs on March 28.1 use its daily low price of 15l
/2 on that day
to get a target price of 197
/s. Prices reach the target in midJuly.
Ifyou can determine that a headandshoulders bottom is forming, then
there is no need to wait for confirmation (that is, for prices to close above the
neckline) before placing a trade. With a failure rate of 5%, your guess will get
you in at a lower level and yield higher profits. However, this all hinges on the
validity of a headandshoulders bottom. If you guess wrong, you could see
your profits rapidly turn into a loss. Ifyou are unsure whether the price series
is indeed a headandshoulders bottom, wait for prices to move above the
neckline before investing.
Table 18.4
Trading Tactics for HeadandShoulders Bottoms
Measure rule
Do not wait for confirmation
Stop Loss
Watch for throwback
Compute the formation height by subtracting the value
of the lowest low reached in the head from the neck
line, measured vertically. Add the difference to the point
where prices pierce the neckline. The result is the target
price to which prices will rise, at a minimum. For steep,
upsloping necklines, substitute the rise between the
head and right shoulder (that is, the highest price in
the rise) for the neckline breakout price.
If you can determine that a headandshoulders
formation is completing, consider buying the stock.
This formation rarely disappoints and the rise is worth
betting on. However, you must be sure that a head
andshoulders bottom is present. Otherwise, wait for
prices to rise above the neckline.
Place a stoploss order below the lower of the two
shoulders. Often, prices drop to the shoulder lows
before meeting support. Raise the stop as prices climb.
If you miss the upside breakout, wait. Half the time, the
stock will throw back to the neckline. Once it does, buy
the stock or add to your position.
Alaska Air Croup Inc. (Air Transport, NYSE, ALK)
Sep94 Oct Nov Dec |an 95 Feb Mar Apr May |un Jul Aug Sep Oct Nov
Figure 18.7 A headandshoulders bottom. Compute the measure rule by sub
tracting the lowest low from the neckline vertically to find the formation height.
Add the difference to the point where prices close above the neckline. The result is
the target price to which the stock will climb, at a minimum. A broadening top
appears in July.
273

274 HeadandShoulders Bottoms v*
Also, since about half of all bottoms throw back, you can wait for a throw
back before placing the trade. Although this will get you in at a higher price,
the likelihood of the trade being profitable also rises. Ifyou have already placed
a trade, consider adding to your position once a throwback completes and
prices move higher.
The two shoulders are common support areas. Figure 18.7 shows an
example of this. The lower of the two shoulders, in this case the right shoulder,
supports the stock in late October.
After placing a trade, consider setting a stoploss point l
/s or so below the
lower of the two shoulders. Should prices decline, they often turn back before
declining below the shoulder lows. If this is too far away from the purchase
point, place your stop '/s below at the closest support zone. Raise your stop as
prices climb.
Sample Trade
Some people might consider Bob unlucky, but he has an adoring wife and two
children. Employed as a blue collar worker in a nearby auto plant, he is happy
when he is working. Unfortunately, strikes by the union have taken their toll
on his savings and he has been looking for ways to supplement his income.
Ever since he was a boy, Wall Street has held his fascination. He has
wanted to play the market and when he saw the headandshoulders bottom
pictured in Figure 18.7, he decided to deploy some of his savings. He bought
at 16, the day after prices pushed through the neckline.
For over a week, he did all right. Prices slowly moved up and reached a
high of 165
/s, then reversed. The stock threw back to the neckline and contin
ued lower. Suddenly, he was losing money. Should he sell and take a loss or
hang on because he knew it was going higher?
He decided to tough it out. The stock bottomed at 14'/2 and quickly
recovered. It reached a higher high, then moved sideways for over a month,
drifting slightly lower. Bob was not worried because he was making money. It
was not a lot, but with patience, he knew he would do okay.
During the summer, things heated up for the airline and the stock took
off. Almost on a daily basis, it soared higher, making new highs. A bearish
broadening top appeared but Bob did not know about such things. He felt
giddy in the thin atmosphere in which the stock was flying. The stock entered
the clouds at 213
/s.
When the airline stock hit turbulence in midSeptember and headed for
the ground, Bob could not believe it. The stock was plummeting and all he
could do was watch his profits spin lower like the stock's altimeter. He talked
it over with his wife and they decided to hold on. "It'll come back to its old
high and when it does, I'll sell it," he grumbled.
Sample Trade 275
The stock continued down. Soon, his profits gone, he was posting losses.
He maintained his firm stance that he would not sell until the price climbed
back to the old level.
During October, things changed. The stock pulled up just before nosing
into the ground, at 135
/8, and not only leveled out, but started climbing again.
In a month he was at breakeven.
At the start of the new year, a descending broadening wedge took prices
lower as it widened but turned out to be a bullish omen. In midJanuary, on
unremarkable volume, the stock turned the corner. Volume climbed, helping
prices reach a higher altitude.
As the stock closed in on his target of 213
/8, Bob called his broker and
placed an order to sell at that price. In late February, the stock began a straight
line run. It soared through 2 !3
/8, hitting his sell order but kept climbing. In just
over a month it reached 30.
Bob no longer invests in stocks.

19
HeadandShoulders
Bottoms, Complex
RESULTS SNAPSHOT
Appearance
Failure rate
Average rise
Volume trend
Throwbacks
Percentage meeting
Surprising findings
See also
An inverted headandshoulders formation with
multiple heads, shoulders, or both
Longterm (over 6 months) bullish reversal
6%
Downward
47%
82%
Formations with downsloping necklines perform
better. High volume breakouts propel stocks to
perform better.
Cup with Handle; Double Bottoms; Headand
Shoulders Bottoms; Horn Bottom; Rounding Bottom
I find that a complex headandshoulders bottom is more difficult to recognize
than a normal headandshoulders bottom but not alarmingly so. After all, if
you can locate a normal headandshoulders bottom, then there is a decent
276
Tour 277
chance that you are also looking at a complex one. If you look to the left and
right ofthe two shoulders, you might see additional shoulders. Multiple shoul
ders are one indication ofa complex formation. But before I delve too far into
pattern recognition, let me briefly review the important snapshot statistics.
The failure rate at 6% is outstanding. Only 15 formations out of almost
240 fail to perform as expected. The average rise is a reassuring 37%, and 82%
of the formations experiencing an upside breakout meet or exceed their price
targets. These figures are all excellent and they imply that this formation is
worth trading.
Two interesting findings relate to the formation appearance. When the
neckline slopes downward, the stock performs better, with gains averaging 39%
versus 34% (for those formations with upsloping necklines). Formations with
high volume breakouts also perform better, with gains averaging 39% versus
32% for lowvolume breakouts. We explore these results in the Statistics section.
Tour
There are two types ofcomplex headandshoulders bottoms: those with mul
tiple shoulders and those with multiple heads. Consider the chart in Figure
19.1, a complex headandshoulders bottom. The chart pattern has two left
shoulders, a single head, and two right shoulders. If you were scanning your
Figure 19.1 A dual shoulder complex headandshoulders bottom. Notice the
horizontal neckline and throwback to it. The formation is part of a rounding bot
tom chart pattern.

278 HeadandShoulders Bottoms, Complex J
charts for normal headandshoulders bottoms, this one would probably pop
up. The left and right shoulders are well defined and the head is below them.
As you widen your view, you see an additional pair of shoulders; the left shoul
der is about the same distance from the head as the right one. The two outer
most shoulders are near the same price level too.
Looking at all the shoulders and the head together, the chart is a good
example of a complex headandshoulders bottom. However, the volume pat
tern is unusual as it is heavier on die right than on the left. Most of the time,
the left shoulders show higher volume than the right pair.
Ifyou ignore the various labels, you can see a rounding bottom. Although
the volume pattern is not a characteristic bowlshaped pattern, the gentle turn
of prices (if you connect the minor lows) supports a bottom formation. How
ever you choose to classify this pattern, the bullish reversal is clear.
Shown in Figure 19.1 is a throwback to the neckline, a common occur
rence for the headandshoulders family, especially the complex variety.
Although it takes a week or two before prices really begin moving up, the stock
climbs to a high of 325
/s before retracing its gains.
Compare Figure 19.1 with Figure 19.2, a complex bottom with two
heads. Overall, the formation is quite symmetrical. There are two shoulders
and two heads. A neckline connects the highs in the formation and projects for
ward in time until prices close above it. The penetration of the neckline is the
breakout point.
Charming Shoppes (Retail (Special Lines), NASDAQ, CHRS)
Left Shoulder Head Head Right Shoulder
Jul 91 Aug Sep Oct Nov Dec |an 92 Feb Mar Apr
Figure 19.2 A dual head reversal. Volume on the left side of the formation is
higher than on the right.
v
In Figure 19.2, the breakout in midNovember quickly throws back to the
neckline and moves lower for a week or two. The stock rises but throws back
again before finally breaking away and heading higher. Bylate March the stock
reaches a high of 165
/g, well above the head low of 93
/i6.
Volume on the left side of the formation is heavier than on the right. In
this regard, the formation is more typical than that shown in Figure 19.1.
Are there certain characteristics that make headandshoulders bottoms easy to
identify? Yes, and they are outlined in Table 19.1. As discussed before, there
are two general types of complex headandshoulder bottoms: those with mul
tiple shoulders and those with multiple heads (rarely do you have both). Figure
19.3 shows a complex bottom with multiple shoulders. The head is distinctly
below the shoulders, far enough below to distinguish the chart pattern from a
triple bottom. In this case, there is a normal headandshoulders bottom
flanked by an additional pair of shoulders. The overall formation appears sym
metrical. The two left shoulders match the two on the right in distance. Fig
ure 19.3 shows a far right shoulder that is higher than its corresponding left
Table 19.1
Identification Characteristics of Complex HeadandShoulders Bottoms
Shape
Symmetry
Volume
Near horizontal neckline
Upside breakout
A headandshoulders bottom with multiple shoulders,
multiple heads, or (rarely) both. The head is lower than
the shoulders but generally not by very much.
The tendency for the shoulders to mirror themselves
about the head is strong. The price level of the
shoulders and time distance from the shoulder to head
is about the same on either side of the head. The
shoulders also appear to be the same shape: narrow or
wide shoulders on the left mirror the right.
Usually higher on the left side than the corresponding
shoulders on the right. Overall, the volume trend
recedes.
Connects the highest rise on the left and right of the
formation center. Most formations have near horizontal
necklines.
A breakout occurs when prices close above the neckline.
For those cases with a steep, upsloping neckline, use
the highest price between the head and rightmost
shoulder as the breakout price.

280 HeadandShoulders Bottoms, Complex Focus on Failures 281
Figure 19.3 A complex headandshoulders consolidation. The trend resumes
moving up once the formation completes.
one. However, the basic symmetrical pattern is typical for nearly all complex
headandshoulders bottoms.
Figure 19.3 also shows the usual volume pattern: The two left shoulders
show higher volume than the two right ones. Overall, die volume trend is a
receding one.
The neckline connects the highest peak on the left with the highest peak
on the right. Most of the time the line is nearly horizontal. Although this is
subjective, a scan of all the formations indicates that 74% obey this guideline.
Many of the formations shown in this chapter have near horizontal necklines.
For those with steep necklines (that slope upward), consider using the
highest high in the formation as the breakout price. Using a steepsloping
neckline to gauge the breakout point is risky. Prices may never close above the
neckline.
Once prices close above the neckline, a breakout occurs. Quite often,
prices throw back to the neckline and perhaps move lower before ultimately
continuing higher. Figure 19.2 shows an example of this behavior during late
December when prices plunged from a high of 11'/2 to 97
/i6, a decline of almost
20% in just 2 days! When the decline ended, prices recovered quickly.
The formation shown in Figure 19.3 is unusual because it acts as a con
solidation of the uptrend. Prices from November through January climb
steadily and then resume moving up after the breakout. The formation is a
consolidation region, where prices move horizontally for a spell.
Most of the time you will find complex headandshoulders bottoms at
the end of a downtrend. Figure 19.1 is an example of this. Although I discuss
statistics later, the study reveals that 75% of the formations act as reversals of
the prevailing trend and most of the heads (the lowest low in the formation)
occur near the yearly low.
Focus on Failures
Ifmaking money in the stock market is important to you, it pays to study your
failures. The lessons you learn will serve you for many years. When you look
at your failures as a group, you may begin to see trends. Such is the case with
chart formations.
Although there are only 15 failures out of 239 formations, 66% of the
failures act as consolidations of the trend. Of course, this is really no help at all
since you can only determine if the formation is a consolidation or a reversal
after the breakout. Many of the failures occur after an extended runup in
prices (then prices backtrack to the formation). After the breakout, the ultimate
low is nearby, usually within 10% below the lowest price reached during for
mation of the head. There are a few cases where the decline is over 25%, so
you should still place a stoploss order to limit your losses.
Figure 19.4 shows a typical failure of a complex formation to reverse the
downtrend. The stock peaks during September 1991 at a price of 1063
/g. From
that point, it is a slow decline at first but picks up speed after the minor high
during midJuly 1992. By the followingJanuary, the stock reaches alow of457
/8
and forms the dual head.
After the headandshoulders formation completes, prices do climb, but
only to 57'/8. Prices squeeze above the neckline and close there for just a hand
ful of days before sliding below the neckline in early March. Ultimately, the
stock reaches 405
/s in August.
The volume pattern is nearly perfect for a headandshoulders formation.
The left shoulder shows tremendous volume. Volume diminishes at the dual
heads, and the right shoulder shows even less volume. Breakout volume is ane
mic and may explain why the formation fails. On closer examination, I found
that only 3 ofthe 15 failures (20%) show low volume breakouts. It appears that
breakout volume is not a predictor of the success or failure of a formation.
After all, the statistics show 41 successful lowvolume breakouts.
I count any formation with prices that fail to rise by more than 5% as a
failure. About half the failures fall into this 5% failure category (the other
seven formations have downside breakouts). Figure 19.4, for example, falls
under the 5% rule. The breakout is upward, but it fails to climb very far before
reversing direction. Once prices decline below the head, I know that there is no
hope and mark the formation a failure.

IBM (Computers & Peripherals, NYSE, IBM)
Sep92 Oct Dec )an93
Figure 19.4 A failure of the complex headandshoulders formation to climb
more than 5% after an upside breakout, just 6% of the formations fail in this man
ner or have downside breakouts.
In sum, I found no reliable clues that indicate an eventual failure of a
complex headandshoulders bottom. This should not be alarming since fail
ures represent only 6% of the formations. In essence, you should be able to
trade this formation withoutworrying about a possible failure. Yes, failures will
occur, but with a 94% success rate, why worry?
Statistics
Table 19.2 outlines general statistics for complex headandshoulder bottoms.
I uncovered 239 formations in 500 stocks over 5 years. This number offorma
tions is on the low side but quite respectable for a somewhat rare formation. Of
the formations I reviewed, the vast majority (181 or 76%) act as reversals ofthe
prevailing trend, meaning that once the formation breaks out, prices move in
the direction opposite to that before the formation began. In the vast majority
of cases prices head higher.
The failure rate is 6%. Nearly all the formations I looked at have upside
breakouts in which prices rise by more than 5%.
After a breakout, the average rise is 37%. However, the most likely rise is
lower—20% to 30%. Figure 19.5 shows how I arrived at this range. I created
V Table 19.2
General Statistics for Complex HeadandShoulders Bottoms
Description _____ Statistic
1991to1996
Failure rate
Most likely rise
Average rise of formations with downsloping
necklines versus upsloping necklines
Average rise of formations with higher right
shoulders (furthest on left versus furthest on
right) versus higher left shoulders
Near horizontal neckline
239
15 or 6%
37%
20% to 30%
184 or 82%
150 or 63%
39% versus 34%
38% versus 36%
176 or 74%
Note: Complex headandshoulders bottoms have a 94% success rate and stocks climb
37% after an upside breakout.
40 50 60
Percentage Gain
80 90 >90
Figure 19.5 Frequency distribution of gains for complex headandshoulders bot
toms. The chart shows the most likely gain is 20% to 30% for the stocks in the
database.
283

284 HeadandShoulders Bottoms, Complex "
a frequency distribution of the percentage gains for the stocks in the database.
The chart shows two columns with the highest frequency, 20% and 30%, close
together and well above the surrounding columns. You can also see how
the over 90% column tends to pull up the overall average. The tendency for
large numbers to skew an average adversely is the reason I use a frequency
distribution.
The measure rule predicts the price move after a breakout. The rule
involves computing the formation height and adding the value to the breakout
price (see Trading Tactics). The result is the expected minimum move. For
complex headandshoulders bottoms, 82% of the formations in the database
meet or exceed their price targets. I consider anything above 80% to be reliable.
The average formation length is about 3l
/2 months, which understates
the actual length because it measures the distance between the two outermost
shoulders. The descent to the left shoulder and rise to the breakout point are
not included in the average.
I use the slope of the linear regression line on volume data to determine
whether volume is trending up or down. In almost two out of three cases
(63%), the volume trends downward. Many times it is easiest to see this trend
by viewing the shoulders. The left shoulders will have higher volume than the
corresponding right pair.
Some analysts suggest that the neckline slope and shoulder height show
the strength ofthe formation. I looked into this and found that formations with
downsloping necklines have a tendency for larger gains than upsloping ones.
The average price rise is 39% for formations with downsloping necklines and
34% for upsloping ones. The difference is statistically significant, meaning
the results are probably not due to chance.
In a similar manner, I looked at the shoulder height. Do formations with
a higher right shoulder rise further? Yes. The average rise is 38% versus 36%.
By higher right shoulder, I mean a right shoulder that does not decline as far as
the left one. Although the difference is not statistically significant, it does stand
to reason. When prices do not decline as far as they did in a prior minor low,
then the trend is in the process ofchanging (from down to up). Investors notice
this strength and purchase the stock.
Is there a relationship between the gain after a breakout and the height of
the right shoulder? Not that I could determine. I calculated the gains for the
stocks in the database and the right shoulder height, both expressed as a per
centage, and graphed the results. I expected to see small right shoulders with
outsized gains. The scatter plot revealed that the relationship is essentially a
random one.
Do complex headandshoulders bottoms have mostly horizontal neck
lines? Yes, with 74% falling into that category. This is a subjective measure and
I define the term mostly horizontal slope to mean less than 30%. The results sup
port using the characteristic as an identification guideline.
Statistics 285
Table 19.3 shows breakout statistics. Most of the formations (97%) have
upward breakouts with the remainder showing downward ones. Eight forma
tions have upside breakouts that fail to climb by more than 5 %. Together with
the 7 that have downside breakouts, this accounts for the 15 formation failures.
In almost half (47%) the formations, prices return to the neckline within
30 days (if they take any longer, it is not a throwback but normal price action).
These are called throwbacks and it takes only 12 days, on average, for prices to
return to the neckline.
For those formations with successful upside breakouts, it takes 8 months
to reach the ultimate high. This places the formation in the longterm cate
gory. However, I use a frequency distribution of the time to reach the ultimate
high to get a more realistic distribution. The results do not change. Most ofthe
formations take over 6 months (which is the threshold between intermediate
and long term) to reach the ultimate high.
Where in the yearly price range do formations occur? Most of the for
mations (44%) have their breakout in the center third of the price range. This
makes sense because prices rise from the shoulder lows to the neckline. That
rise alone often shifts the formation into a higher category. Mapping the per
centage gains over the yearly price range shows that there is little difference in
performance: 34% to 38%. The best gains occur when the breakout is within
a third of the yearly high. It seems to me that the momentum players grab hold
of the stock and bid it up faster when it is near the yearly high.
Table 19.3
Breakout Statistics for Complex HeadandShoulders Bottoms
Upside breakout
Downside breakout
Throwbacks
For successful formations, days to ultimate
high
12month low (L), center (C), or high (H)
lookback period
Average rise of high volume breakouts
versus low volume breakouts
232 or 97%
7 or 3%
8 or 3%
108 or 47%
12 days
8 months (241 days)
l_20%, C44%, H37%
L34%, C37%, H38%
163%, 138%, 115%, 103%, 98%, 99%
39% versus 32%
Note: Nearly all the breakouts are upward and reach the ultimate high in 8 months.

Another interesting statistic I developed relates to high volume breakouts.
Do high volume breakouts propel stocks higher? Yes, with gains averaging
39% versus 32% for low volume breakouts. The differences are statistically
significant.
In the calculation, I used the breakout day plus the next 2 days and aver
aged the volume of those three together (since a high volume breakout might
be delayed). Then I compared the volume with the day before the breakout. I
considered values over 25% above the benchmark as high volume and 25%
below the benchmark as low volume. I computed the percentage gains for
those stocks with high volume breakouts and compared them to those stocks
with low volume breakouts to derive the results.
Trading Tactics
Trading tactics are outlined in Table 19.4. The measure rule predicts the
expected minimum price move and is best explained by an example. Figure
19.6 shows a complex headandshoulders bottom on a weekly time scale with
the head reaching a low of 13!
/2. Directly above that point, the neckline has a
value of 185
/s. The difference, S'/s, is the formation height. Add the difference
to the breakout point (17) to get the minimum price move (22 VB).
Table 19.4
Trading Tactics for Complex HeadandShoulders Bottoms
Measure rule
Do notwaitfor confirmation
Stop loss
Watch for throwback
Compute the formation height by subtracting the
lowest low reached in the head(s) from the
neckline, measured vertically. Add the result to the
breakout price where prices pierce the neckline.
The value is the minimum target price.
If you can determine that a complex headand
shoulders formation is completing, consider
placing a long trade or cover any short
commitments.
Stocks sometimes decline to the lowest of the right
shoulders then turn around. Look for support areas
near the shoulders. Place a stoploss order below
the lowest shoulder or head.
Buy or add to the position during a throwback.
Wait for prices to finish falling before placing the
trade as prices sometimes throw back and continue
moving down.
Trading Tactics 287
It took just 2 weeks after the breakout to reach the target, but the stock
was not done climbing. It moved sideways for almost a year before continuing
higher. The stock reached a high of 393
/g, nearly triple the head low of 13 '/2 and
more than double the breakout price.
The chart in Figure 19.6 shows a complex headandshoulders bottom
that forms after nearly a 2year runup in prices. The formation marks a rever
sal ofthe 6month retrace. Once a breakout occurs, prices quickly climb to ful
fill the measure rule and then stall. Prices then move horizontally for almost 2
months before climbing to the next level. There the stock consolidates for 7
months before shooting upward in midJune 1995.
Since complex headandshoulders bottoms reliably break out upward,
there is little need to wait for the actual breakout. Once you determine that a
complex formation is present, buy the stock. Of course, the key is that you
must be sure a complex headandshoulders formation is present. Many times
this is made easier when the formation looks like the one in Figure 19.6, where
a normal headandshoulders pattern is flanked by two or more shoulders. If
you can identify the inner headandshoulders pattern, then you need only
widen your vision and look for additional shoulders.
With dual heads, the pattern is somewhat different. The dualhead for
mation usually has head lows that are less than a month apart. Two heads that
Alien Telecom Inc. (Telecom. Equipment, NYSE, ALN)
Left Shoulder^ .,' . — Right Shoulder
9 2 A S O N D 9 3 F M A M | J A S O N D 9 4 F M A M | | A S O N D 9 5 F M A M | | A S O N D 9 6 F M
Figure 19.6 Complex headandshoulders bottom on a weekly time scale. The
figure shows the target price found using the measure rule. Compute the forma
tion height from the head low to the neckline and add the difference to the break
out price. The right shoulders often offer support during future declines.

are close together usually distinguishes the formation from a classic double
bottom. Shoulder symmetry and a near horizontal neckline should put die fin
ishing touches on the formation identification.
Once you take a position in the stock, set your stoploss point. Many
times the various shoulder troughs will act as support levels. Ifyour headand
shoulders formation is near the yearly low, then there is a very good chance
that prices will either turn around at the head or decline slightly below it (by
10% or so) before bottoming out. From that point, prices climb higher.
If your formation is not within the lowest third of the yearly price range,
then sell the stock once prices drop below the head. Prices falling below the
head signal a formation failure and it is best to cut your losses instead of pray
ing that they will turn around. They will not.
After an upside breakout, almost half the time (47%), the stock throws
back to the neckline. Consider adding to your position or placing a long trade
once prices stop declining. You should wait for prices to rebound on a throw
back or else you could find yourself in a situation similar to that shown in Fig
ure 19.2. Prices throw back to the neckline then continue down for over a
week. Depending on when you bought the stock, you could have seen a near
10% price improvement if you had waited a few days.
Sample Trade
When the weather is nice, I like to take my bicycle out for a spin and give the
automobile drivers something to aim for. It was on one ofmy bike trips that I
met Melody. After I told her what I did for a living, she confessed that she was
a nightclub dancer and made oodles in tips. I was unsure whether I bought her
story, but she looked pretty enough (wearing a bike helmet and sun glasses,
who can tell?).
Anyway, she told me about a trade she had made in the stock pictured in
Figure 19.6. The stock intrigued her because a trendline drawn from the high
est high in early October to just after the head marked a turning point. That is
where prices moved up enough to pierce the trendline.
Melody knew that prices usually retest the low before beginning an
extended move upward, so she followed the stock and watched it loop around
and dip to 14. Then she glanced sideways and noticed the other dip at 143
/s.
That is when she uncovered the headandshoulders bottom.
A neckline connecting the rises between the two shoulders was impossi
bly steep; there was no way she could apply the traditional measure rule to
determine a target price, so she decided to buy into the stock when prices
closed above the right shoulder high.
This occurred in late May and she received a fill at 17'/2. Taking a closer
look at the graph, she saw two more shoulders, one during early February and
Sample Trade 289
the mirror image in midMay, both at 16. Her simple headandshoulders bot
tom changed into the complex variety.
The realization did not affect her investment plans at all, but it made the
situation more interesting. She wondered if another pair of shoulders would
appear. Her suspicions were fulfilled during late July when another shoulder
developed. This one at 153
/4 mirrored the shoulder in midDecember 1993.
Soon, prices began moving up. They climbed above the breakeven point in
midAugust and staged an upside breakout. Now she was able to apply the
measure rule for the complex bottom and found the target was 22'/s.
Since she did not need the money immediately, she held onto the shares
as prices rose. She thought the stock had enough upward momentum to reach
die old high at about 29'/4, and she set her sights on that. As long as prices did
not drop below the purchase price, she would stay in the trade.
She saw the stock building a base between 21 and 26 and wondered what
to make ofit. A downside breakout was a real possibility, so she raised her stop
to 21—the height of the plateau in October—and at a price just below where
die base seemed to be building.
In midJune, just over a year after she placed the trade, prices zoomed up
and reached her sell point. The stock sold at 29. The stock continued climb
ing, but she needed the money for a down payment on a house.
I was so engrossed with her story and the way she told it that I did not
realize she had dismounted from her bicycle. She spoke of coming back to her
place and making some new chart patterns, then playfully thrust her hips into
mine.
I fell off my bicycle.

20
RESULTS SNAPSHOT
Appearance
Failure rate
Average decline
Volume trend
Fullbacks
Percentage meeting
Surprising findings
See also
Threebump formation with center bump taller than
the others
7%
Slopes downward with highest volume on left
shoulder, followed by head and right shoulder,
respectively.
45%
63%
A downsloping neckline or a lower right shoulder
(versus left shoulder) predicts a larger decline, but
results are not statistically significant.
HeadandShoulders Tops, Complex
Of all the formations in this book, die headandshoulders top is perhaps the
most popular. This sterns, in part, from its reliability. With 93% ofthe forma
tions breaking out downward and continuing to move down, there is no need
to wait for a breakout before trading. In that regard, you can save yourself
290
Tour 291
more money ifyou own a stock and decide to sell or place a short sale sooner,
garnering larger profits. Headandshoulders tops' popularity also stems from
their recognizability. The characteristic three bumps with the higher center
bump make the formation easy to spot.
Two surprising findings are that the slope of the neckline and a lower
right shoulder predict a more drastic price decline after the chart pattern.
Unfortunately, the results are not statistically significant, meaning that they
may (or may not) be due to chance. They are still interesting nonetheless, and
the Statistics section describes the findings in detail.
Tour
Figure 20.1 shows a good example of a headandshoulders top. The three
bumps are clearly visible with the center bump being the highest of the three.
The left shoulder usually appears after an extended uphill run. The entire for
mation seems to stand alone when viewed in the context of a year's worth of
daily price data. This standalone characteristic makes the headandshoulders
top easily identified in a price series.
Figure 20.1 shows the highest volume occurring during the head. More
often the left shoulder will have the highest volume, followed by the head, with
greatly diminished volume during formation of the right shoulder. The iden
Great Atlantic and Pacific (Grocery, NYSE, GAP)
Head
Mar 93 Apr May Jun Aug Sep
Figure 20.1 A headandshoulders top formation where the center peak towers
above the other two. A pullback to the neckline occurs in almost half the formations.

292 HeadandShoulders Tops Identification Guidelines 293
tification guidelines are flexible because volume characteristics vary from for
mation to formation.
A trendline drawn along the bottoms of the two troughs between the
three peaks forms the neckline. The line may slope in any direction but slopes
upward about 52% of the time and downward 43% of the time with the
remainder being horizontal. The direction of neckline slope is a predictor of
the severity of the price decline. We see in the Statistics section that the neck
line slope and die shoulder height are both related to the ultimate price
decline.
Why do these formations form? Pretend for a moment that you are a big
spender and represent what is commonly called the smart money. You are
searching for a stock to buy and believe that Toll Brothers (Figure 20.2) rep
resents an intriguing situation. You review the fundamentals and everything
looks good, so you start buying the stock in midJuly as prices descend. Your
buying turns the situation around: The stock begins rising.
Soon you have acquired all the stock you want and sit back and wait. As
you expected, the company issues good news and the stock begins making its
move. Other investors jump into the game and buy the stock, sending the price
higher. As the stock rises above 10, you decide it is time to sell. After all, you
have made 20% in about 2 weeks. Your selling causes the stock to pause then
begin a retrace of the prior action.
Sensing weakness in the stock, you stop your selling and monitor the sit
uation closely. Other momentum and buythedip players, believing that this
Figure 20.2 Volume pattern of this headandshoulders top obeys the general
characteristics: highest on formation of the left shoulder and weakest on the right
shoulder. The downsloping neckline suggests an especially weak situation.
is a chance to get in on the ground floor of a further advance, buy the stock on
the retrace. The decline halts and the stock begins rising again.
As it rises, other momentum players make a bid for the stock or buy it
outright. Once the stock gets above 10, you begin selling it again, not heavily
at first because you have a large number of shares to dump. Still, the market
players notice your selling and the stock climbs just above 11 before heading
back down.
You dump your remaining shares as the stock begins tumbling. Volume
rises as other players sell their shares to unsuspecting buyers. The stock con
tinues moving down and slides back below 10. Believing the stock oversold,
demand picks up and sends the price moving up again for the last time.
You watch the action from the sidelines, content with the profit you have
made. The stock climbs to 103
/4 on the right shoulder. Lacking support, the
rise falters on weak volume and the stock turns down. Investors versed in tech
nical analysis see the headandshoulders top for what it is: a reversal. They
quietly take their profits and sell the stock. Others initiate short sales by sell
ing high and hoping the price falls.
Prices move down to the support level where prices declined the last time.
The stock pauses at the support level for a week and makes a feeble effort to
rise again. When the attempt falters, the stock moves down and pierces the
neckline. Volume picks up and the stock tumbles. Eventually, prices decline
back to where they began, just under 8.
In essence, the formation is a symbol ofshares being turned more quickly
as prices rise. Eventually the selling pressure squelches demand, sending prices
tumbling.
Are there certain guidelines that make identifying a headandshoulders top
easy? Yes, and Table 20.1 lists them. The identification guidelines are just
that, guidelines. The headandshoulders top formation can appear in a wide
variety of shapes. Consider Figure 20.3. Shown is a headandshoulders top
formation, but there are four shoulders and only one head. When a formation
appears with more than the standard two shoulders and one head, it is called a
complex headandshoulders pattern. Complex headandshoulders patterns
for both tops and bottoms have their own chapters but many appear in this
chapter's statistics. They are, after all, headandshoulder tops too.
The headandshoulders top formation usually appears at the end of a
long uptrend. Sometimes, when the prior uptrend is of short duration, the
reversal takes prices down to where they started the climb (see Figure 20.2). At
other times, the decline is usually short (up to 3 months) or intermediate (3 to
6 months), or can signal a change in the primary bullish trend. The actual
length of the decline cannot be predicted.

Table 20.1
Identification Characteristics of HeadandShoulders Tops
Shape
Symmetry
Volume
Neckline
Downside breakout
After an upward price trend, the formation appears as three
bumps, the center one is the tallest, resembling a bust.
The two shoulders appear at about the same price level.
Distance from the shoulders to the head is approximately the
same. There can be wide variation in the formation's
appearance, but symmetry is usually a good clue to the
veracity of the formation.
Highest on the left shoulder, followed by the head. The right
shoulder shows the lowest volume of the three peaks.
Connects the lows of the two troughs between the three
peaks. The line can slope up or down. Often used as a
trigger point (to buy or sell) once prices pierce the line.
Once prices pierce the neckline, they may pull back briefly,
then continue moving down.
Even though the formation shown in Figure 20.3 is somewhat odd, it
does have a symmetrical appearance. The two left shoulders are at about the
same price level as the corresponding two right shoulders. Each of the shoul
ders is approximately the same distance from the other and from its mirror
opposite. In the chart pattern, the head is centrally located. The symmetrical
appearance of a headandshoulders top formation is one of its key ideiitifica
Toys R Us (Retail (Special Lines), NYSE, TOY)
Figure 20.3 A complex headandshoulders top pattern. The chart shows the
wide variation that a headandshoulders formation can take.
Statistics 295
tion characteristics and helps separate any three bumps from a valid headand
shoulders chart pattern.
Volume obeys the general characteristic: It is higher on the left shoulder
than on the head and higher on the head than on the right shoulder. If you
consider just the three inner peaks in Figure 20.3, the volume pattern changes
somewhat since the left shoulder has volume diminished from that shown dur
ing the head. Even so, the volume on the left shoulder is still above the right
shoulder.
The neckline, as shown in Figure 20.3, connects the two troughs between
the three inner peaks. It slopes upward but need not do so (contrast with Fig
ure 20.2). The neckline serves as a confirmation point. Once prices pierce the
neckline, and assuming they do not pullback, prices continue moving down
in earnest.
A pullback to the neckline occurs almost half the time. It usually takes less
than 2 weeks to complete a pullback, but do not be fooled. The trend will re
sume downward shortly. However, a pullback does allow you one more oppor
tunity to exit a long position or institute a short trade. Take advantage of it.
Focus on Failures
Failures ofheadandshoulders formations are rare, but they do occur. Figure
20.4 shows an example of a failure. The wellformed formation has a head cen
trally located between two shoulders. The left and right shoulders are at the
same price level, 291
/g. Volume is highest on the left shoulder and lowest on the
right, as you would expect.
Why do prices fail to pierce the neckline at point A and head down? The
answer is not clear. The formation is perfect except that it fails to descend. It
acts as a consolidation or continuation of the upward trend. Not shown in the
figure, the prior two formations were descending triangles. These formations
usually break out downward but these did not. Both had upside breakouts and
both signaled a bullish uptrend. The two formations were clues to the strength of
the rise, but one could also argue that the appearance of a headandshoulders
formation would probably signal an end to the extended rise. It did not.
If there is a good side to this failure, it is that failures do not occur very
often. Only 6% of the formations (25 out of 431) I looked at consolidate as the
formation in Figure 20.4.
Statistics
Table 20.2 shows general statistics for the headandshoulders top formation.
The headandshoulders top pattern is a plentiful one, occurring 431 times
over the study period. In almost every occurrence it acts as a bearish reversal of

Hughes Supply Inc. (Retail Building Supply, NY5E, HUG)
Figure 20.4 A rare headandshoulders consolidation. The formation fails to con
tinue down after reaching point A. Symmetry and volume patterns offer no clue to
the eventual failure.
Statistics 297
the uptrend. Only 30 formations (7%) fail to continue moving down by more
than 5% or break out upward. That statistic means that prices usually decline
after a valid formation occurs and suggests that you need not wait for confir
mation ofa downside breakout. A discussion of this trading tactic follows in the
Trading Tactics section.
Once prices pierce the neckline, they continue moving down another
23%, on average. However, a frequency distribution of declines suggests the
most likely decline is less, about 15%. Figure 20.5 shows the relationship. The
bellshaped appearance of the chart is reassuring. I consider the tallest column
the most likely decline because it has the highest frequency (the highest number
of formations in that decline range).
I discuss the measure rule in the Trading Tactics section, but suffice it to
say that the rule measures the height from the head to the neckline and sub
tracts the result from the point where prices pierce the neckline. Prices meet
the target only 63 % ofthe time. I consider values above 80% to be acceptable,
so this method comes up a bit short.
On average, the formation is 2 months long (62 days), as measured
between the left and right shoulder peaks. Of course, this understates the actual
length of the formation since prices need time to rise up to the left shoulder
and decline to the neckline. Ifyou add the drop from the right shoulder to the
Table 20.2
General Statistics for HeadandShoulders Tops
1991 to 1996
Failure rate
Most likely decline
Average performance of lower right shoulder
versus performance of lower left shoulder
Average performance of formations with
downsloping necklines versus upsloping necklines
431
30 or 7%
23%
15% to 20%
254 or 63%
2 months (62 days)
24% versus 22%, but result is
not statistically significant
23% versus 22%, but result is
not statistically significant
Note: With a 7% failure rate, you need not wait for the breakout before placing a trade.
Figure 20.5 A frequency distribution of declines for successful downside break
outs in headandshoulders tops. The graph suggests the most likely decline is
about 15%, below the average decline of 23%.
296

breakout point (that is, the neckline), then the average length rises to 79 days.
I did not measure the time it takes prices to rise to the left shoulder peak.
Some analysts have suggested that the shoulder height implies a strong or
weak technical situation. By that I mean if the left shoulder peak is higher than
the right one, then prices are more likely to decline farther than if the heights
are reversed. Although I found this to be true, with a lower right shoulder
decline of 24% and a 22% decline for formations with a lower left shoulder,
the difference is not statistically significant. By that I mean the result could be
due to chance, or it could mean that there is indeed some validity to the notion.
I examined the relationship between the two shoulder prices. Do short
right shoulders mean a larger decline? I used a minimum price difference
between the two shoulders of 5%. Although the performance results (34%
average decline versus 19%) are significant, the sample size is not (I found only
II or 12 samples out of 431 because identification depends on the shoulders
being near in price to each other). Using a 4% price difference brings the
results down to 25% and 22% and increases the sample size to 35. However,
die results are nearly the same as that achieved without any qualifiers (those
shown in the table). If you can make a blanket statement about this behavior,
it might be that larger price differentials between the two shoulder prices
results in different performance. Significantly lower right shoulders suggests a
more substantial decline.
In a similar manner, some analysts have said that a downsloping trend
line shows a weaker technical situation than an upsloping one. This is also
true, with losses averaging 23% versus 22%, but again, the difference is not
statistically significant. Since so many have suggested that these relationships
do exist and the statistics do not contradict that notion, then it is more likely
to be a reliable indicator of pending price action.
Table 20.3 shows statistics related to breakouts. The vast majority (98%)
of breakouts are downward, whereas only 9 formations move upward. How
ever, there are 21 formations with downside breakouts that fail to move down
by more than 5%. Even so, this only represents 5% of the formations with
downside breakouts. The statistics suggest that a headandshoulders forma
tion will not only break out downward, but that prices will continue moving
down as well.
There are a large number ofpullbacks, at 191 or 45%. Fullbacks are when
prices drop below the neckline then return to it. Figures 20.1 and 20.3 show
examples of pullbacks. On average, it takes slightly less than 2 weeks for the
pullbacks to return to the neckline.
Once prices break out downward, it takes 3 months on average to reach
the ultimate low. This seems to be a typical decline rate for bearish formations.
Of course, sometimes the formations indicate a longer term trend change, but
you will usually find that prices reach the ultimate low within 3 to 6 months
(short to intermediateterm trading implications).
Statistics 299
Table 20.3
Breakout Statistics for HeadandShoulder Tops
Upside breakout
Downside breakout
Pullbacks
Percentage of breakouts occurring near the 12month
9 or 2%
422 or 98%
21 or 5%
191 or 45%
11 days
3 months (91 days)
L11 %, C40%, H49%
L22%,C22%,H21%
Note: Almost all the headandshoulders tops break out downward and reach the ultimate
low in about 3 months.
A frequency distribution of the days to reach the ultimate low confirms
die shortterm implication of a headandshoulders top. The vast majority of
formations (265 or 66%) are short term, with only 83 falling into the interme
diate term, and 53 taking more than 6 months to reach the ultimate low.
If you consider where the formations appear over the prior 12month
price range, then you can get a feel for where the best performing headand
shoulders patterns occur. First, I exclude all formations that begin within 1
year from the start of die study. Then die yearly price range is divided into
tliirds for each formation. Every formation sorts into one of die three bins,
depending on die price at die breakout. The results show diat most formations
(49%) break out near die yearly high, suggesting diat most headandshoulders
tops appear at die end of an uptrend.
Do formations that have a breakout near die yearly high perform better
(that is, decline further) dian diose breaking out near the yearly low? No. The
performance percentages distribute evenly at about 22%. This means prices
are likely to decline about 22% after a breakout regardless ofwhere diey occur
in die yearly price range. In other bearish formations, we have seen that die
best performing chart patterns occur near die yearly low.
Table 20.4 shows volume statistics. As I was searching for the formations,
I made no assumptions about die volume pattern. However, an analysis of die
statistics using linear regression shows there are definite trends. Almost two
out of three formations (62%) have receding volume trends. This downward
trend is clear in many of die charts that accompany this chapter.
When you look at the volume pattern of die three bumps, you discover
that high volume usually accompanies creation ofdie left shoulder. The volume

Table 20.4
Volume Statistics for HeadandShoulders Tops
Volume highest on which bump?
Second highest volume on which bump?
Volume lowest on which bump?
247 or 62%
Left shoulder, 49%
Head, 51%
Right shoulder, 74%
159%, 146%, 111 %, 99%, 99%, 98%
Note: Volume is highest on the left shoulder and weakest on the right shoulder.
is higher than that shown during the other two bumps. Formation of the right
shoulder usually shows the lowest volume. Volume at the head falls between the
other two bumps—it is usually lower than the left shoulder but higher than the
right one.
To make the volume assessments, I looked at each formation and logged
the volume pattern for the three bumps. For example, I assigned the highest
volume to one of the three bumps after looking at the chart pattern. This was
also done for middle and low volume ranges. After reviewing all the stocks, I
added up the numbers for the three categories (highest, middle, and lowest vol
ume) for each of the three bumps (left shoulder, head, and right shoulder). The
percentage numbers beside each entry in Table 20.4 tell how many bumps
have the associated volume characteristics.
For example, left shoulders have the highest volume, with 49% of them
falling in that category. Heads follow at 37% and right shoulders at 13%, all in
the highest volume category. The only surprise in the figures is the number of
hits in the lowest volume category. Seventyfour percent of right shoulders
have low volume, whereas only 14% of left shoulders and 12% of heads show
low volume.
Turning our attention to volume at the breakout, we find that the volume
rises on the breakout day by 59% above the prior day (or 159% ofthe total). This
is in line with other bearish formations. Shown in Table 20.4 are additional vol
ume statistics for the week after the breakout. Notice how the volume drops off
rapidly. Again, this decrease in volume is normal and emphasizes the belief that
prices can fall oftheir own weight and do not need high volume to decline.
Trading Tactics
Shown in Table 20.5 are trading tactics, and Figure 20.6 shows an example of
the measure rule as it applies to a headandshoulders top. If you ignore the
backward volume pattern, the formation looks fine. Each of the three bumps
Table 20.5
Trading Tactics for HeadandShoulders Tops
Measure rule Compute the formation height by subtracting the value of the
neckline from the highest high reached in the head, measured
vertically. Subtract the result from the breakout price where prices
pierce the neckline. The difference is the minimum target price to
which prices descend. Alternatively, compute the formation height
from the highest high to the daily low price in the higher of the
two troughs. Subtract the result from the daily high price in the
higher of the two troughs to get the target price. This method
boosts the success rate to 69% and does not rely on the neckline
or breakout point (useful for steep necklines).
Do not wait for Once the right shoulder forms and you are confident that a head-
confirmation and-shoulders formation is valid, sell your stock or sell short. With a
93% success rate, there is little need to wait for a confirmed
breakout before placing a trade.
Short stop For short sales, place a stop just above the lower of the two
troughs or just above the neckline, whichever is higher.
Watch for pullback Initiate a short sale or add to your position during a pullback. Wait
for prices to begin falling again before placing the trade as prices
sometimes pull back and continue moving up.
Arco Chemical Co. (Chemical Basic), IMYSE, RCM)
Figure 20.6 The measure rule as it applies to a headandshoulders top. Calculate
the formation height by subtracting the neckline price from the highest high, mea
sured vertically. Subtract the result from the high at the breakout. The result is the
minimum target price to which prices decline.
301

appear rounded and the overall formation is symmetrical. The measure rule
uses the formation height as a basis for computing the target price. In the
head, measure vertically down from the highest daily high until you intersect
the neckline. Subtract the value of the neckline from the highest high. The
result gives the formation height. In the figure, the stock reaches a high price
of 51 on September 13. Directly below that point is the neckline price at about
473
/s. The difference of 35
/s is the formation height.
Once prices pierce the neckline, subtract the formation height from the
daily high at the breakout point. In Figure 20.6, the high at the breakout is
481
/2> leaving a target price of 447
/s. Prices surpass the target when they decline
below the value in late November. Since the target serves as a minimum price
move, prices often continue moving down, as in Figure 20.6. However, only
63% of the formations meet or exceed the price target. I consider values above
80% to be reliable.
The measure rule, as just described, is the conventional way to compute
a target price. However, it does have a flaw. Consider Figure 20.7. Prices dur
ing the right trough recession decline to 273
/4, well below the higher trough at
31]
/4. A neckline joining the two is too steep. Prices never plunge through the
neckline and it is impossible to compute a target price using the conventional
method. Instead, compute the formation height by taking the difference
Figure 20.7 Headandshoulders top with steep neckline. There is no target price
using the conventional measure rule because of the steep neckline. Alternatively,
compute the formation height by subtracting the higher trough low (point A)
from the highest high. Subtract the result from point A to get the target price.
Prices meet or exceed the target 69% of the time versus 63% for the conventional
method.
Sample Trade 303
between the highest high in the head and the lowest low in the highest trough
(point A on the chart) in Figure 20.7. After finding the formation height,
subtract the value from point A to get the target price. In this example, the
highest high is at 335
/8 and the lowest low at the highest trough is 3lH, giving
a height of 23
/8. Subtract the result from 311
A to get a target of 287
/s. Figure 20.7
shows this value, and prices reach the target during midApril.
The alternative method has two advantages. First, it can always be calcu
lated and is somewhat easier to use since it does not rely on the value of the
neckline. Second, it is more accurate, achieving a success rate of 69%, mean
ing that more formations exceed the price target using this alternative method
rather than the conventional one.
Returning to Table 20.5, since the formation rarely fails, there is little
need to wait for a confirmed breakout. Instead, once you are sure a headand
shoulders top is forming and you want to maximize your profits, sell any shares
you may own or sell short (as close to the right peak as possible). This action
allows you to get out of a commitment sooner than waiting for the neckline
to be pierced.
Occasionally, prices will rebound at the neckline and move higher. Either
repurchase the stock at that point or close out your short. Since prices usually
pierce the trendline on their way down, do not be too quick to repurchase the
stock. It may bounce up at the trendline then continue down after moving hor
izontally or follow the neckline. Sometimes general market conditions or other
companies in the same industry can provide a direction clue. If they show
weakness, expect your stock to follow the crowd and prices to move lower.
If you sell short, place your stoploss order either just above the neckline
or above the lower of the two troughs, whichever is higher. Selecting a nearby
resistance point usually works well.
If prices pull back to the neckline, consider adding to your short position.
However, be sure to wait for prices to begin falling after a pullback. Occasion
ally, prices will pull back and continue rising.
Sample Trade
Kelly is not just a housewife; she is much more that. When her husband brings
home the bacon, she not only fries it but cleans up the mess afterward. She bal
ances the books and keeps tabs on their newborn.
She started investing years ago for fun. Now, it has become part of her
daily life. In the spare moments between chores, she is often staring at the
computer screen, reviewing the statistics of a prospective acquisition and let
ting her daughter bang on the keyboard.
Over the years she has been able to parlay their meager savings into a six
figure retirement portfolio. It was not easy and the mistakes were painful but
she viewed each failure as a learning experience.

The stock pictured in Figure 20.7 posed an interesting situation for her.
She was not keen on shorting a stock because her paper trades rarely worked
out. Still, she kept her eyes open and searched for good investment candidates.
This one piqued her interest.
The stock began its uphill run just before May 1993. It followed a gently
sloping trendline upward until late January when it stumbled. The stock
moved down to 261
/? before recovering, a drop of less than three points, but a
sign ofweakness. Kelly followed the stock closely and when the head appeared,
she made a note on her program that it might turn into a headandshoulders
top. "It just had that certain feel," she remarked. She was right.
The right shoulder plunge took prices lower than she expected but
quickly recovered to near the left shoulder high. She drew a neckline below the
two valleys and thought the line was too steep to serve as an anchor for the
measure rule, so she used the alternate measure rule and computed a target
price of just 287
/s. This did not seem right either, so she used the right shoul
der low to compute another target. This one turned out to be 217
/8, or the
height from the head to the right shoulder valley projected downward from the
valley low. That target would take prices back to the July level and it seemed
reasonable to her.
Still, something bothered her about the stock and she decided not to
trade it. When the doorbell rang, she left her daughter alone briefly to answer
it. Moments later, the phone rang. It was her broker confirming that the stock
sold short. Kelly ran to the computer to see her daughter standing on the
chair, beating on the keyboard with a wide but guilty grin on her face. Kelly
hoped it was only gas, but, no, she had indeed sold the stock short at 31.
After spending some anxious moments reviewing the trade, Kelly decided
to maintain the position. The number of shorted shares was just 100, an
amount she could live with. Prices quickly retreated to the neckline where they
found support. The stock bounced and when it moved above the right shoul
der low, she got concerned. After a few days, the stock leveled out and moved
sideways. In case this turned out to be the beginning of a measured move up,
she placed an order to cover her trade at 29. That would leave her with a small
profit but still allow her to participate if the stock declined.
Two weeks later, she had an answer. The stock tumbled for 5 days in a
row, then just as quickly recovered, only this time it formed a lower high. The
volatility was wearing her down so she placed an order with her broker to
cover her position when prices reached the old low. She was taken out when
prices descended to 223
4 on their way down to 20. After expenses, she made
about 25% on the trade. Her daughter got a big kiss for her help.
21
HeadandShoulders
Tops, Complex
RESULTS SNAPSHOT
Appearance
Failure rate
Average decline
Volume trend
Fullbacks
Percentage meeting
Surprising findings
See also
A headandshoulders formation with multiple heads,
shoulders, or both
8%
Downward
64%
67%
Formations with downsloping necklines or higher
left shoulders perform marginally better.
Double Tops; HeadandShoulder Tops; Horn Tops;
Rounding Tops; Triple Tops
Except for appearance, there is not much difference between a normal head
andshoulders top and a complex one. Add a dual head or a few extra shoulders
to a regular formation and you have a complex headandshoulders top. Both
formations have a volume trend that generally slopes downward between the
305

shoulders. The left shoulders often have higher volume than the correspond
ing right ones.
The failure rate for complex headandshoulders tops is very low (8%)
but slightly above the rate recorded for regular headandshoulders tops. Pull
backs have also moved up a notch and now appear in nearly two out of three
formations.
The average decline at 27% is on the high side for a bearish formation.
However, the most likely decline is just about evenly distributed from 10% to
30% (when viewed in 5% increments).
Two interesting findings deal with necklines and shoulder height. When
the neckline, which is a trendline joining the lowest minor lows between the
shoulders, slopes downward, the performance improves slightly from a decline
of 26% to 27%. Likewise, when the outermost shoulder is higher on the left
than the corresponding one on the right, the performance improves from 27%
to 28%. In both cases, the differences may not be statistically significant
(meaning they could be due to chance).
Tour
There are two basic varieties ofcomplex headandshoulder tops, as illustrated
in Figures 21.1 and 21.2. In Figure 21.1, the formation appears after an
Alcan Aluminium (Aluminum, NYSE, AL)
Aug 95
Figure 21.1 Complex headandshoulders top with dual heads. The stairstep
pattern of a measured move up forms the left shoulder and head. The twin peaks
take on the appearance of a horn top, and the resulting move down resembles
another measured move, albeit stuttered.
Tour 307
extended bull run that begins in November 1992 at a low price of 15'/4. The
stock climbs to a high of 365
/8 by midJuly 1995, then melts back to 285
/8 by
midOctober, forming a base for die head and shoulders. The stock rebounds,
creating the left shoulder. It pauses at die 3132 level by moving sideways, dien
spikes upward again in a sort ofmeasured move thrust. The measured move up
finishes shy of its target price by just over a dollar before die stock begins
retracing its gain. The peak serves as die first head.
After moving down a bit, die stock pushes upward and tags the old high,
then drops. Another head appears. Once prices slip from the head, they find
support at the first shoulder trough and rebound. The right shoulder takes
shape. After declining through the neckline, formed by a line joining the two
shoulder troughs, prices quickly pull back and move higher. They turn away at
the 32 resistance level and continue down in a straightline run to 283
/8.
Computing a line using linear regression ofthe daily volume over die for
mation (outermost shoulder to shoulder) indicates volume recedes. Although it
is difficult to tell from die chart in Figure 21.1, about two out of every three
complex headandshoulders tops show a receding volume trend.
Ifyou believe in the classic definition of a double top, you might consider
the twin heads a candidate for that formation. However, several flaws eliminate
this pattern as a candidate for a double top. First, the two peaks are too close
together on the time line. For a classic double top, the peaks should be at least
a month apart (my definition of a double top allows peaks to be closer
together). Also, the recession between the two peaks should take prices down
by 15% to 20%. Figure 21.1 shows a decline ofjust 5% from the highest peak,
well short of the goal.
I could further complicate the comparison by pointing out that the two
heads look like a horn top, but we are discussing complex headandshoulder
formations. Let me say that you find such behavior quite often in technical
analysis: Each formation can be viewed from several different perspectives.
Some analysts might see a complex headandshoulders formation while others
see a pair of measured moves—one the skewed mirror image of the other—
while yet others might see a horn top. The results are the same so there is no
need for concern: All point to a bearish situation.
Figure 21.2 shows a different type of complex headandshoulders top.
Multiple shoulders with only one head is the more common of the two vari
eties. Pictured is the type of technical pattern that rips die heart out of novice
investors. Imagine someone buying diis stock in October, just before the rise
begins. Prices quickly move from a low of about 13 to a high of 277
/s, a dou
bling of the stock price in a little over 3 months. On the way up, our novice
investor is thinking that picking stocks is an easy game; his selections are turn
ing to gold.
The first shoulder forms as prices touch 277
/8, then retreat to a low of 22.
The decline undoubtedly upset our investor pal. He probably told himselfthat
he would sell the stock once it returned to its old high.

Figure 21.2 Typical complex headandshoulders reversal. Multiple shoulders
with a single head in a rather flat formation round out the pattern. The volume pat
tern emphasizes that volume is usually higher on the left side of the formation than
the right.
In early January, prices zoom upward and make a smaller peak at 27'/2.
Since the rise is so steep, our intrepid investor thinks, why sell the stock when it
isgoing to go higher? He is right. Prices retrace a bit then move higher and form
the head at a price of 28s
/s. Once the head completes, things start to go wrong
for our buddy. He is swayed by glowing predictions on the Internet of the
stock moving up to 3 5 or 40 within a year.
At die top, prices round over and start down. They stop midway between
the troughs of the two left shoulders before making one final attempt at a new
high. Up to this point, there are several opportunities to sell the stock at a good
price. Did our novice investor take them? Probably not. Always optimistic that
prices will ultimately break out and reach higher ground, he does not see the
budding complex headandshoulders formation for what it is: a warning.
When prices drop below the neckline, our novice investor has just 2 short
days before things really get going. By the third day things are looking grim as
the stock closes at 23, near the low for the day. Prices quickly unravel and ulti
mately reach a low of 15, just a few dollars above the purchase price. That is
when our investor throws in the towel and sells the stock. Of course, this is
near the low and the stock ultimately climbs to 30 a year later.
How can our novice investor recognize the bearish reversal? Table 21.1 out
lines some identification tips of a complex headandshoulders top.
Consider Figure 21.3, another example of a multiple shoulder chart pat
tern. After a decline from a headandshoulders formation just off the left side
of the chart, prices decline until reaching bottom at the start of July. Then they
rise up, haltingly, and form a new headandshoulders formation: a complex
top. If you ignore the labels for a moment, the inner price action looks like a
rounding top. This smooth price rollover is common for complex headand
shoulder formations. Of course, the flat head shape for a multiple shoulder pat
tern (Figure 21.2) is also typical.
You can divide Figure 21.3 into a pure headandshoulders formation by
ignoring the outer shoulders. For singlehead formations, this is the easiest
way to correctly identify a complex headandshoulders top. First locate a reg
ular headandshoulders pattern then expand your view to include additional
shoulders. In this example, the head rises above the surrounding shoulders.
The two shoulders are usually equidistant, or nearly so, from the head. The
price level of the left and right shoulders is very nearly the same. Thus, the
Table 21.1
Identification Characteristics of Complex HeadandShoulders Tops
Shape
Symmetry
Volume
Neckline
Downside breakout
A headandshoulders top with multiple shoulders or, more
rarely, two heads. The head is higher than the shoulders but
generally not by very much.
The tendency for the shoulders to mirror themselves about the
head is strong. The price level of the shoulders and time
distance from the shoulder to head is about the same on either
side of the head. The shoulders also appear to be the same
shape: Narrow or wide shoulders on the left mirror those on
the right.
Usually higher on the left side than on the right and is usually
seen when comparing the shoulders on the left with
corresponding ones on the right. Overall, the volume trend
recedes.
Connects the lowest left shoulder trough with the lowest right
shoulder trough. When the line extends and intersects prices,
that signals a breakout.
When prices close below the neckline, a breakout occurs. For
those cases with a steep, downsloping neckline, use the
lowest trough price as the breakout point.

Figure 21.3 A more rounded appearing complex headandshoulders top. Left
shoulder 1 and right shoulder 1 could be considered part of the inner headand
shoulders formation. The inner headandshoulders looks like a rounding top for
mation.
symmetry of a complex headandshoulders top is more pronounced than a
regular headandshoulders formation.
In a regular headandshoulders formation, one shoulder may be higher
in price than the other or one shoulder will be much further away from the
head—a rather extended shoulder. That is usually not the case with the com
plex variety. Symmetry is paramount and a key identification element.
Moving to the outer shoulders, they also are equidistant from the head
and are very nearly at the same price level as well. Continuing the symmetry
example, the two peaks labeled left shoulder 1 and right shoulder 1 appear to
be shoulders of the same formation, although further away than the inner
grouping.
Ifyou consider the inner quad of shoulders as part of the head, then what
remains is a large, regular headandshoulders formation. This is denoted
by left shoulder 1, right shoulder 1, and the large, rounded head (composed of
five minor highs). Even the neckline supports this example as prices touch
the line several times before dropping through it in a 1day decline of about
four points.
If you could zoom in on the volume pattern, you would see it is margin
ally heavier on the left side of the formation than on the right, at least for the
formation bounded by the inner (higher) neckline. High volume on the left
side of the formation as compared to the right is typical for complex headand
shoulders formations and occurs about twothirds of the time.
The neckline joins the lows of the lowest trough and is interpreted the
same way as a normal headandshoulders top. Once prices pierce the neckline,
a downside breakout occurs and prices move lower. Volume typically rises on
a breakout and can remain high for several days, depending on the severity of
the decline.
Focus on Failures
Complex headandshoulders patterns suffer two types of failures. Both are
rare. Figure 21.4 shows the first type. I define a downside breakout as a close
below the neckline, or in the case of steep necklines, a close below the lowest
shoulder trough. Figure 21.4 shows prices declining below the neckline only
once on May 8 but closing above it. From that point, prices rise and move
above the highest head and an upside breakout occurs.
The formation itself is well formed. It has two heads at about the same
level and two shoulders also near the same price level. Symmetry throughout
the formation looks good, too, as the shoulders are equidistant from the head.
Volume appears heavier on the left shoulder than on the right, as you would
expect. Only during the decline from the right head to the right shoulder does
Figure 21.4 A complex headandshoulders failure to reverse. Prices fail to close
below the neckline before moving above the formation top and staging an upside
breakout.

volume rise. In short, there is no real indication that this formation will fail to
continue moving down, but it does.
Figure 21.5 shows a slightly different picture. The multipleshoulder for
mation appears less balanced. The shoulders are somewhat less even and not
equidistant from the head. However, it is a complex headandshoulders top
with a downside breakout. Unfortunately, prices fail to continue moving down.
The stock suffers a 1day drop of $2, but then rises in an ascending broaden
ing wedge pattern. The wedge is a bearish pattern that breaks out downward
but it too fails to descend very far. Within a few months, the stock is again
making new highs.
Returning to the complex headandshoulders pattern, I regard the for
mation as a failure because prices fail to move down by more than 5% after die
breakout. A 5% decline should take prices to 49s
/g, but the actual decline is well
above that.
Are there any clues to the failure of this formation? The volume pattern
is flat. It shows no tendency to diminish over time. A receding volume pattern
is not a hardandfast rule so I do not consider this pattern to be that unusual.
To answer the question, I see no real clues as to why the formation does not
continue moving down. A closer examination ofthe fundamentals on the com
pany may provide some clues.
Minnesota Mining and Manufacturing (Chemical (Diversified), NYSE, MMM)
May 94 Dec Jan 95
Figure 21.5 Another failure of a complex headandshoulders top. This one fails
to decline more than 5% below the breakout point. An ascending broadening
wedge takes shape in late November and December.
Statistics 313
Statistics
As alarming as the two figures may appear (Figures 21.4 and 21.5), one has to
balance failures with the realization that they do not occur very often. Table
21.2 shows that only 11 out of 141 formations fail to move down by more than
5%. That gives a 92% success rate. In the vast majority of cases I studied, the
formation acts as a trend reversal. Once prices break out downward, they con
tinue down by more than 5%. In their quest to reach the ultimate low, prices
decline by an average of 27%.
The most likely decline is somewhat harder than usual to determine. Fig
ure 21.6 shows a frequency distribution of declines with a 10% interval
between bins. The figure shows the most likely decline is 20%. Subdividing the
interval into 5% bins shows the frequency for many bins is within a few hits of
die others. The 5% chart suggests the declines can range from 10% to 30%,
almost equally.
The Trading Tactics section discusses the measure rule in detail, but it
involves computing the formation height and subtracting it from the breakout
value. The measure rule provides a minimum price target that hits 67% ofthe
time. This is lower than the 80% I like to see, so one should consider the pos
sibility of falling shy of die target before placing a trade.
Description
Table 21.2
General Statistics for Complex HeadandShoulder Tops
Statistic
1991 to 1996
Failure rate
Most likely decline
Average performance of formations with
downsloping necklines versus upsloping
necklines
Average performance of formations with higher
left shoulder (farthest on left versus farthest
on right) versus higher right shoulder
141
11 or 8%
27%
20%
87 or 67%
3 months (83 days)
89 or 63%
27% versus 26%
28% versus 27%
Note: The complex headandshoulders pattern takes longer to form and declines further
than regular headandshoulder formations.

30 40 50 60
Percentage Decline
Figure 21.6 Frequency distribution of declines for complex headandshouiders
tops. The most likely decline is 20%.
The formation is about 3 weeks longer (83 days total) than a regular head
andshoulders formation. This should come as no surprise since it requires
more time for another set of shoulders or heads to appear.
Almost twothirds (63%) of the formations have a downward volume
trend. A line formed using linear regression on the volume data makes the
determination. Most of the time the line slopes downward and indicates vol
ume is receding.
Some analysts suggest that a downsloping neckline implies an especially
bearish situation. I found this to be true, but die difference (with an average
decline of 27% versus 26%) may not be statistically significant. In the calcula
tion, I used the lowest of the two shoulder troughs to determine the slope of
the neckline.
In a similar manner, high left shoulders, when compared to their right
counterparts, suggest a more bearish outlook. Again, I found this to be true
with formations that have higher left shoulders suffering a decline averaging
28% versus 27%. For this study, I used the outermost shoulders as the bench
mark. Since the two numbers are so close to each other, the difference may not
be meaningful.
Table 21.3 shows breakout statistics. There are only three upside break
outs, with the remainder being downward (138, or 98%). Once a downward
breakout occurs, the formations must continue moving down by at least 5% or
else they are 5% failures. Eight formations (6%) fall into this category.
Statistics 315
Table 21.3
Breakout Statistics for Complex HeadandShoulders Tops
Upside breakout
Downside breakout
Fullbacks
ultimate low
12month price low (L), center (C), or
high (H)
Percentage loss for each 12month
lookback period
3 or 2%
138 or 98%
8 or 6%
88 or 64%
10 days
16%, C47%, H48%
L22%,C28%,H26%
175%, 170%, 138%, 120%, 112%, 106%
Note: Nearly all the formations (98%) have downside breakouts.
Fullbacks, which are prices that break out downward but return to the
neckline within 30 days, are quite numerous at 88 (64%) of formations, sug
gesting that ifyou miss the breakout, wait. A pullback may occur, allowing you
to place a short sale at a higher price. Of course, if the stock fails to pull back,
then look elsewhere for a more promising situation. Never chase the stock as
you will buy into a situation that may soon reverse and go against you.
The average time for the pullback to return to the neckline is 10 days.
Many of the formations have pullbacks occurring just a few days after a break
out. The average is skewed upward by several outliers that take over 3 weeks
(but less than a month) before returning to the neckline.
On average, it takes 3 '/2 months to reach die ultimate low. When we com
bine this statistic with those in Table 21.2, we discover that these formations
are wide, suffer a more severe decline, and take longer to reach the ultimate
low. In diat regard, the complex headandshoulders formation is more pow
erful dian a regular headandshoulders reversal.
Most of the formations occur in die center third (47%) or upper diird
(48%) of die yearly price range. Only 6% of die formations break out near die
yearly low. The results make sense in that die price trend rises to die formation,
effectively elevating die breakout point out of the lower regions. If you filter
die percentage loss for each of die formations over their yearly price range, you
see diat die largest declines occur in the center third of the price range. The
formations in that category decline by an average of28%. Quite close to this are

316 HeadandShoulders Tops, Complex ^^
formations with breakouts in the top third of the yearly price range, with an
average decline of 26%.
Volume is unusually high for several days after a breakout. You can see
the trend in Table 21.3. Volume starts out high, 75% above the prior day (or
175% of the total volume), but recedes to near the average by the following
week.
Trading Tactics
The measure rule predicts the minimum expected decline. Look at Figure 21.7
as an example of the measure rule outlined in Table 21.4. Compute the for
mation height by subtracting the difference between the highest high (315
/g)
from the value of the neckline directly below the highest high (27). Subtract
the result (4s
/g) from the breakout price (253
/4), which is where declining prices
pierce the neckline. Prices drop below the target price of 21l
/s in early March.
This is the conventional measuring rule and it is successful 67% of the
time. That is below the 80% success rate I like to see for measure rules. The
conventional method also has a flaw when the neckline slopes steeply. Under
such circumstances, prices may never pierce the neckline and yet the stock is
tumbling. Fortunately, the neckline rarely slopes steeply in complex forma
Trading Tactics 317
Table 21.4
Trading Tactics for Complex HeadandShoulders Tops
Measure rule
Do notwaitfor confirmation
Short stop
Watch for pullback
neckline value from the highest high reached in the
head, measured vertically. Subtract the result from the
breakout price where prices pierce the neckline. The
result is the minimum target price. If the formation looks
like a mountain suddenly appearing out of a flat base,
prices may return to the base. See Figures 21.2 and 21.5.
If you can determine that a complex headand
shoulders top formation is completing, consider acting
immediately. Place a short trade or sell any long
commitments. This formation rarely disappoints and the
decline is above average. Should prices rebound at the
neckline, reestablish your position.
Look for resistance areas about the neckline troughs.
Place a stop just above the higher shoulder trough. The
shoulder tops and head also represent good locations
for stoploss orders.
Place a short sale or add to the position during a
pullback. Wait for prices to begin falling again before
placing the trade as prices sometimes pull back and
continue moving up.
Figure 21.7 A complex headandshoulders top formation. For the measure rule,
compute the difference between the highest high and the neckline, measured ver
tically, and subtract the result from the breakout price. A broadening top appears
from December through February and a deadcat bounce follows.
tions. However, ifyou do run across a formation with a plunging neckline, use
the lowest of the shoulder troughs as the breakout price. Compute the forma
tion height in the normal manner, but subtract the height from the trough
value to get a target price.
A slightly more conservative measure rule computes the formation height
from the highest high to the higher of the neckline troughs. Then subtract the
result from the higher ofthe neckline troughs and the result is the target price.
This method results in 76% ofthe formations meeting the measure rule, which
is still short of the 80% I like to see.
Using the chart pattern shown in Figure 21.7 as an example, the highest
high is at 3 l5
/g and the higher of the two neckline troughs is at point A, 27'/2
(using the daily low price). This gives a formation height of 4'/s (315
/s 27l
/i)
and a target price of 233
/s (or 27'/2 4'/s). Prices reach the target during the
large 1day decline on March 2.
If the headandshoulders formation does not occur after an extended
upward trend—the situation shown in Figure 21.1—then prices will probably
drop back to the low where the formation began. In Figure 21.1, the low just
before prices start rising to the formation is about 28. That is the level to
which prices return after the breakout.

At other times, the formation will suddenly shoot up from a base, then
just as quickly return to the base. Figures 21.2 and 21.5 show examples of this
situation.
Since the complex headandshoulders top is so reliable, once you are rare
you have a valid pattern, take advantage of it. Ifyou own the stock, sell it. Ifyou
do not own the stock, short it. For short positions, look for areas of strength—
resistance levels—and place a stoploss order just above that level. Common
resistance levels are the shoulder troughs, shoulder tops, and head top. Should
the price rise above the highest high, immediately cover the short position as
it is an upside breakout and likely to continue soaring.
Be aware that prices sometimes break out downward, then regroup and
rise above the neckline before plunging down again (see Figures 21.1 and 21.3).
These extended pullbacks and regular pullbacks are good places to initiate a
short sale or add to a position.
Sample Trade
Henry runs a small hedge fund. He considered buying into the stock shown in
Figure 21.7 but needed more bullish evidence. Two weeks later he got his sig
nal. Prices pushed up through a longterm, downsloping trendline in August
1994. Other indicators he uses on a daily basis confirmed the buy signal, so he
bought shares for his fund at an average price of about 174.
As the stock climbed, he followed its progress and checked periodically on
the fundamentals. The ride up was not an easy one because the stock began
acting oddly from October through early December. At one point during that
period, it looked as if a broadening top was forming, but the price action
changed enough that the formation fell apart.
The stock bounced between 21 and 25 several times then pushed its way
to higher highs. Henry suspected the end was near, so he began taking a closer
look at the fundamentals. He was so engrossed with his research on the com
pany that he failed to notice a pattern forming. Over drinks with his fund man
ager friends, he shared with them what he had dug up about the company. The
news was not good.
"So that's why it's making a broadening top!" one remarked. Henry fur
rowed his brow and pictured the price action in his mind and there it was, a
broadening top, just like his friend had said.
The next day Henry pulled up the chart and looked at it more closely. He
saw higher highs and lower lows (see the zoom out in Figure 21.7), character
istic of a broadening pattern. Coupled with his fundamental research on the
company, he knew it was nearing time to sell, but not yet. He wanted to sell at
the top, when prices tagged the top trendline.
In early February, when prices attempted to reach the previous high, they
fell short, dipped down for a few days, and tried again (the two right shoulders).
Sample Trade 319
The second rise was even shorter than the prior one, signaling weakness, so
Henry started selling immediately.
The failure to sail across the formation and touch the top of the broad
ening formation meant it was a partial rise. A partial rise in a broadening top
usually means one thing: A downside breakout will follow.
By the time the stock pulled back up to the base of the two right shoul
ders (point B), Henry had sold his holdings. As he was getting ready to leave his
office for home, something on his computer screen caught his eye. The broad
ening pattern had changed into a complex headandshoulders top. There were
the two left shoulders balancing the two right ones with a head perched in the
center.
Henry discussed the new situation with his mentor and his fund manager
buddies, then decided to short the stock. By the time prices reached the long
term up trendline, he had a tidy sum sold short.
Two days later, prices tumbled. They dropped 20% or $5 a share in 1 day
and continued down. In less than a week, they were at 16 before finding some
support, a plummet of 36%.
Henry had studied the behavior of deadcat bounces, and he pulled out
his notes and brushed up. He knew the stock would bounce upward, usually
within a week, then trend lower.
True to form, prices moved up a bit (to IS'/s), but it was not the smooth,
rounded bounce he expected. In the coming days, prices moved lower, so
Henry quit complaining, but he watched the situation closely.
The stock bottomed out at about 16 and trended horizontally. To him, it
looked as if the stock was building a base and preparing for an upward move,
so he covered half his short position.
In late April, when prices jumped up to 18'/2, he immediately covered the
remainder of his position. In the pub that evening with his buddies, he was all
smiles. He was feeling so good that he decided to pick up the tab.

22
Horn Bottoms
RESULTS SNAPSHOT
Appearance
Failure rate
Average rise
Surprising finding
See also
Two downward price spikes separated by a week on
the weekly chart
11%
Abnormally long price spikes perform better.
Double Bottoms, Pipes
I first discovered this formation while pondering a result from my study of
double bottoms. Double bottom formations with bottoms closer together per
form better than those spaced widely apart. What would happen if you con
sidered formations that have bottoms only a week or so apart? I tested the idea
and discovered that the formation performs well. The failure rate at 11 % is
well below the 20% maximum rate I allow for reliable formations.
The average rise at 37% is below the usual 40% gains posted by other
bullish formations. Such large gains do not happen overnight. A frequency
distribution of the time it takes to reach the ultimate high shows most horn
bottoms falling into the longterm category (taking over 6 months to reach the
ultimate high).
320
Tour 321
I reviewed a number ofcriteria to improve the performance ofhorns. The
criterion responsible for the most improvement is when the horn length is
longer than most spikes over the prior year (at least twice as long). The aver
age gain rises to 43%.
Tour
Figure 22.1 shows what a horn bottom looks like. After peaking in late Decem
ber 1993, prices plummet from a high of 503
/4 to the horn low at a base of 303
/4.
On the left side ofthe horn, prices have a large weekly price range of about $7.
High volume makes the week appear like a oneweek reversal (with the same
attributes as a oneday reversal but over the course of a week), signaling a pos
sible trend change.
The following week, prices close lower but nowhere near the left horn
low. Then, 1 week later, prices spike lower again but close near the high ofthe
week and just Vs below the prior close. The horn bottom is complete: A dou
ble price spike separated by 1 week marks the turning point. From that point,
prices move up and more than double from the horn low in about a year and a
half.
Tandy Corporation (Retail (Special Lines), NYSE, TAN)
9 3 D 9 4 F M A M | | A S O N D 9 5 F M A M | J A S O N
Figure 22.1 A good example of a horn bottom. Two downward price spikes, sep
arated by a week, look like a steer's horn flipped upside down.

322 Horn Bottoms
How do you correctly identify horn bottoms? Table 22.1 shows identification
characteristics. In essence, the characteristics define the shape of an inverted
horn that is clearly distinguishable from the surrounding price action.
Consider Figure 22.2 that shows two horn bottoms. What might strike
you first about the left formation is its similarity to Figure 22.1. The stock
shown in Figure 22.2 tumbles from a price of almost 25 in August 1991 to a
low of just 61
/2 at the right horn spike in less than 2 years. Then prices recover
and move higher.
Horns are visible on weekly charts. Although they appear on daily charts,
weekly charts make selection easier. For the formation on the left in Figure
22.2, the chart shows two long, downward price spikes separated by a week.
The low of the center week stays well above either of the spike lows, empha
sizing the inverted horn shape of the formation.
Looking back over the months, you can see that there are no downward
spikes that come near the length ofthe horn spikes (as measured from the low
est low to the lower of the two adjacent weeks). The twin horns mark an
unusual event, one that an investor should pay attention to.
Table 22.1
Identification Characteristics of Weekly Horn Bottoms
Weekly chart, downward spikes
Small price variation
Clear visibility
Large overlap
High volume
Use the weekly chart and locate two downward price
spikes separated by a week. The two spikes should be
longer than similar spikes over the prior year and be
well below the low of the center week. It should look
like an inverted horn. Abnormally long spikes result in
better performance.
The price difference between the two lows of the
horn is small, usually 3
/s or less, but can vary up to $1
or more for higher priced stocks.
In a downtrend, the horn lows should be well below
the surrounding lows, especially to the left of the
formation for several weeks (or months). Usually,
horns appear near the end of declines but also
happen on retraces in uptrends (where visibility is less
clear to the left).
The two spikes should have a large price overlap
between them.
The left spike shows higher than average volume
(54% of the time) and the right spike shows below
average volume (52% of the time). However, these
are only guidelines and volume varies greatly from
formation to formation.
The price difference between the two horn lows is usually small, about 3
/s
or less. Higher priced stocks can show larger differences (but on a percentage
basis, the differences are minimal). In Figure 22.2, the difference between the
two spikes is just $0.25 (for the left formation).
For the left formation, the horn appears after a downward price trend,
allowing clear visibility to the left of the formation, as no downward trends or
price outliers obscure the view. This is important in that the formation should
stand alone and not be part of a congestion region. It should mark the turning
point of a downward price trend.
Price overlap is quite good for the formation, with the two horn spikes
nearly the same size but the left one shifts upward.
The formation shown on the right of the chart in Figure 22.2 is what a
horn bottom looks like in an uptrend. There is a small price retrace, of 3 weeks'
duration, just as the horn bottoms out. These few weeks separate the formation
from the surrounding price action and allow easy recognition.
The two spikes share the same low price, 14'/2, and have good price over
lap (as the right spike almost completely overlaps the left one). You can argue
that the separation of the horn low from the surrounding weeks is not excep
tional when compared with the pipe formations in early December and late
June. That is certainly true, but most of the prices show remarkably even bot
toms, not a jagged coastline.
The volume trend on this formation is unusual as both horns show below
average volume. The left formation shows a different volume trend as both
horns sport above average volume. The statistics say that the left side of the
Figure 22.2 Two examples of horns, one in a downward trend and one in an
uptrend. Notice that the second pipe formation calls the turn exactly.

324 Horn Bottoms
horn bottom will have above average volume 54% of the time, whereas the
right will have below average volume just 52% of the time. Since both figures
are quite close to 50%, it really can go either way, so I would not place too
much emphasis on volume.
Focus on Failures
Even though horn bottoms sport a low failure rate (at 11%), they still have fail
ures. Consider Figure 22.3, a 5% failure or a horn bottom. A 5% failure is
when prices start out in the correct direction but falter (rising by no more than
5%), turn around, and head back down. The twin, downward price spikes look
good in that they are long and with good overlap. They form as part of a
retrace from the high, and prices usually return to form a second high (a dou
ble top), or perhaps move even higher. Ifyou believe that this stock will form
a second top, then this formation is probably not worth betting on as the price
appreciation potential is just not exciting enough.
If you look at this formation differently, you might suspect that it will
form a headandshoulders top. The left shoulder is already visible in late May
1994 and another shoulder could form as part of a mirror image, probably in
April or May of 1995. If that is the case, then you should also pass this one up
as the right shoulder might top out at about 16. This assumes prices continue
Figure 22.3 A horn bottom failure. Among other clues, similar length price spikes
(marked L) suggest this formation might be suspect. The descending triangle sug
gests lower prices.
Statistics 325
moving lower and probably stop dropping in the 10 to 12 range (forming the
neckline) before moving up to the right shoulder. Since prices should drop,
why buy now?
Another clue to this formation failure is the spikes themselves. Ifyou look
over the prior prices, you see several downward price spikes that rival the
length of the horn. These are warning signs that this horn might not be any
thing special.
The visibility is poor because earlier prices block the view. Usually, prices
have lows that move down in sync with the remainder of the prices (like that
shown in the February to April 1995 decline). A horn appearing in a sharp
decline should have good visibility to the left of the formation. In this case, the
horn bottom has competing prices to the left ofit, blocking its view. What this
tells us is that prices seem to form a base while their tops are declining. In other
words, a descending triangle is forming and the investor should be wary.
Taken together, there seems to be ample evidence that this horn might
not work out as expected. But, statistically, how often do horn bottoms fail?
Statistics
Table 22.2 shows statistics for horn bottoms. The chart pattern is quite plen
tiful, occurring almost 300 times in 500 stocks over 5 years. Of these forma
tions, 160 are consolidations ofthe prevailing trend, whereas the remainder act
as reversals. A closer examination of the data shows that 88% of the reversals
occur when prices are moving down—they reverse direction and head up after
the formation.
Most of the consolidations are also in a downtrend, with 70% showing
prices moving down over the prior month. The longerterm trend (over 3
months), however, is upward 65% of the time. In other words, the overall
Table 22.2
General Statistics for Weekly Horn Bottoms
to 1996
Failure rate
Most likely rise
Left, right horn volume versus 25day moving
average
296
34 or 11%
37%
20% to 30%
146%, 116%
9 months (265 days)

326 Horn Bottoms Statistics 327
price trend is up, then consolidates (moves down or retraces) for a month
before continuing up.
The failure rate is quite low at 11%, well below the 20% maximum for
reliable formations. Almost all the formations work as expected by moving up
more than 5%. The average rise is 37%, with the most likely rise falling
between 20% and 30%, a strong showing.
I computed the values using the average of the highest high and lowest
low the week after the formation ends (the week following the right horn
spike). This assumes that an investor buys the stock the week after noticing the
horn and receives a fill at the midrange value. It also assumes that he sells out
at the highest possible price (the ultimate high) before a significant trend
change occurs (a price change of 20% or more).
Figure 22.4 shows a graphic representation ofthe price gains. It is created
using a frequency distribution to eliminate any undue skewing of the average
by large percentage gains. You can see in Figure 22.4 that the rightmost col
umn is quite large, showing 11% of the formations with gains over 90%. The
remainder of the chart takes on the appearance of a bellshaped curve.
I classify the most likely gain as the column(s) with the highest frequency
(not including die rightmost column). In Figure 22.4, these are 20% and 30%,
so the most likely gain should fall in that region. However, the adjacent
columns, 10% and 40%, are nearly as high, so the range might be wider. Of
course, your results will vary.
Figure 22.4 Frequency distribution of gains for horn bottoms. The most likely
gains are in the 20% to 30% range if you ignore the rightmost column.
Volume for the horns, when compared to a 25day moving average of
volume, is 146% for the left horn spike and 116% for the right horn spike.
These figures emphasize that there is quite high volume on the left spike and
somewhat diminished volume on the right one, on average. A count shows that
54% of the left spikes have above average volume, whereas 52% of the right
spikes have below average volume.
For those formations that move higher by more than 5 %, it takes them 9
months to reach the ultimate high, on average. That is a comparatively long
time for a bullish bottom.
Table 22.3 shows some interesting statistics ofhorn bottoms. The bench
mark, the average gain for the formation, is 37% with a failure rate of 11%.
Table 22.3
Surprising Results for Horn Bottoms
Description
Average Rise Failure Rate
13 for both right and
left horns
Benchmark 37 11
Rise for horns with price differences
between lows 38
Rise for horns with no price difference
between lows 36
Performance of horns with lower right
(R) spikes versus horns with lower
left (L) spikes L37, R40
Rise when the right horn is an inside
week 33
Rise when right horn is an outside week 39
Performance for high volume left (L) L38, R37, B37
horns, right (R) horns, and both (B)
horns
Performance for low volume left
(L) horns, right (R) horns, and
both (B) horns L37, R37, B35
High left horn volume, low right
horn volume 40 7
High right horn volume, low left
horn volume 38
Linear regression price trend:
3 months up 37
Linear regression price trend:
3 months down 38
Large price spike (2x average) 43 12
High volume left, low volume
right, downsloping 3month
price trends 34 14

328 Horn Bottoms Trading Tactics 329
When comparing the low price between the two horn spikes, those for
mations with a price difference show gains averaging 38%, whereas those with
no price difference have gains of 36%. In essence, you can expect horns with
uneven low prices to do better. Since horns with uneven low prices perform
better, which is more profitable: formations showing a lower left horn or a
lower right one? Horns with a lower right spike perform better, showing a
40% average gain, whereas those horns with a lower left spike have gains aver
aging 37%.
I also looked at the high prices and discovered that inside weeks perform
poorly (with gains ofjust 33%). Inside weeks are just like inside days in that the
high is below the prior high and the low is above the prior low. The price range
is inside the range of the prior week (or day). I ignored the intervening week
(that is, the week between the two spikes) in the comparison.
I checked on outside weeks, where the right horn high is above die left
horn high and the right horn low is below the left horn low (no ties allowed).
Horns meeting the criteria show gains averaging 39%. Again, the comparison
ignores the center week.
For the following comparisons, I define high volume as being above the
2 5day moving average and low volume as being below the 2 5day moving
average. Do high volume horns score better? Yes, but the difference is mini
mal. High volume on the left horn works best, showing gains of 38%, whereas
low volume left horns have gains of 3 7%. Gains for horns with high or low vol
ume on the right horn ties at 37%. In other words, those formations with high
volume on the right horn show die same gains as those widi low volume on the
right horn. When volume is high on bodi horns, die chart pattern has an aver
age gain of 37%, and horns with low volume on both spikes have gains of 35%.
With high volume on bodi horns, the failure rate increases slightly to 13%.
Measuring the different combinations, high left and low right horn vol
ume shows a 40% gain and a failure rate of 7%. The opposite combination,
low left and high right volume, has gains of 38%.
I used linear regression on the closing prices to determine the existing
price trend leading to die formation. Do horn bottoms perform better after
prices have been trending down for 3 months or trending up? The results are
a wash, with a 37% gain for diose horns in a rising price trend and a 38% gain
for diose in a downtrend.
Do large downward price spikes perform better? Yes, with gains of 43%
but die failure rate rises slightly to 12%. To arrive at this determination, I
computed die spike length from die lowest low to die lowest adjacent low (on
either side) over the prior year leading to die formation. I averaged the values
together to get an average spike length and compared it to the horn length.
Think of the comparison using your fingers: I measured the difference
between your middle fingertip and die closer of the two adjacent fingertips.
Those horns that are at least twice die average lengdi perform better than
shorter ones.
Putting many of the tests together—high volume on the left horn, low
volume on the right one, and a downsloping price trend over 3 months—
shows gains of 34% and a failure rate of 14%. Since these various factors are
simply trying to tune the database for highest performance, your results will
vary. Scanning die table for the best value shows that when horn spikes are
abnormally long, at least double most spikes over die prior year, then perfor
mance improves.
Trading Tactics
Table 22.4 outlines trading tactics and begins with die measure rule. There is
none. Since horn bottoms have no measure rule, there is litde guidance that I
can suggest on when to take profits or how well a formation will perform.
However, once you identify a horn bottom using the guidelines outlined in
Table 22.1, then buy the stock.
Perhaps the most important key to horn bottoms is that they should con
tinue to look like horns (see "Clear visibility" in Table 22.1). Prices should
climb after the twin horn spikes. If you need to wait an extra week or two to
prove this occurs, then do so.
Separate the price trend leading to the formation into either an uptrend
or downtrend. Horns that appear late in uptrends may mark the end of the
upward price move. Prices continue moving higher (perhaps by 10% or so),
then stop. Of course, if your horn appears at the start of an uptrend, then
prices might well be on their way to a large gain.
Horns in downtrends are common. On the one hand, if the downtrend is
just a retrace of die prevailing uptrend, then refer to the uptrend guideline. In
Table 22.4
Trading Tactics for Horn Bottoms
Measure rule None
Identify Use the characteristics outlined in Table 22.1 to correctly identify a
horn bottom. The week after the right horn is key. Prices should
climb smartly and the weekly low should not be anywhere near the
horn low (in other words, the horn should still look like a horn and
not be encroached on by the succeeding price action).
Uptrends Some horns appear near the end of uptrends, so watch for the
trend to change.
Downtrends Horns will usually not mark the end of the downtrend, but they
will be close. Prices might continue to drift down for $1 or so
(below the lowest horn low) then head upward.
Stops If you can afford the loss, place a stop $1 below the lowest horn to
reduce the chance that a retest of the low will stop you out.

330 Horn Bottoms
such a case, if the horn appears after an extensive advance, then the uptrend
may be nearly over. Invest cautiously or look elsewhere.
On the other hand, if the stock has been trending downward for a long
time (for months anyway), then the end might be in sight. The horn probably
will not mark the low exactly, but it should be close. Usually, horns appear a
month or so before or after the actual turning point.
If prices are trending down and you see a horn forming, you might wait
before buying the stock, just to be sure prices have really turned around. For
uptrends, you probably should buy into the situation immediately since prices
will only climb away from you and get more expensive.
Place a stop loss up to $1 or so below the lowest low. For lowpriced
stocks, this may mean taking a significant loss. In such a case, perhaps it is best
to skip the trade and look elsewhere. The reason for placing the stop well
below the lowest horn low is to allow prices time to turn around. Prices some
times curl around and retest the low (moving $1 or so below the horn low)
before recovering and trending upward.
Sample Trade
Mary saw the horn bottom forming in the stock pictured in Figure 22.5. To
her the chart suggested prices would continue moving up. Prices certainly
climbed above the right horn low (at 31) quite smartly, leaving the horn clearly
visible on the weekly chart.
9 2 N D 9 3 F M A M | J A S O N D 9 4 F M A M I J A S O N D 9 5 F M A M ) | A S
Figure 22.5 Horn bottom on weekly chart. Mary was stopped out during
throwback just before prices doubled.
the
Sample Trade 331
Looking back at the entire price chart, she saw prices begin climbing in
early October 1992 at a low of 85
/8. From that point, they soared to the current
price in several waves. Waves pushing prices higher took between 4 and 5
months, whereas those moving lower took 3 months.
When the horn formed, the 5month up pattern was in progress; that
is what she hoped anyway. She bought the stock the week after the horn
completed at 35. She hoped the horn marked an end to the short retrace and
prices would resume their upward trend. She was wrong. When prices curled
around, she placed a stop at 307
/8 —l
/s below the lowest horn low. She sus
pected that this might not be low enough, but a 12% loss was all she was will
ingto tolerate.
In early December, her worst fears were confirmed. She was stopped out
as prices plummeted from 33 to 29. Three weeks later the stock hit bottom at
28'/2 and turned around. From high to low, the decline lasted just over 3
months, as she predicted. Prices moved swiftly upward and topped out at 57,
exactly double the low.
Did Mary sell too soon and pass up her chance to nearly double her
money or did she use prudent money management to limit her losses? Those
questions are ones we all face with eventually. Do you know the answer?

23
Horn Tops
RESULTS SNAPSHOT
Appearance
Failure rate
Average decline
Surprising finding
See also
Two upward price spikes separated by a week on the
weekly chart
Either, with shortterm (up to 3 months) bearish
implications
16%
21%, with most likely loss about 10%
The failure rate drops to 6% when the volume on both
spikes is below the 25day volume moving average.
Double Tops; Pipe Tops
As you look at the Results Snapshot statistics, you have to ask yourself one
question: Why would you ever want to trade this formation? With a likely loss
of just 10%, you probably would be a fool to short a stock based on this for
mation. However, if you trade enough and with luck, your returns should
approach the average of21 %. The kicker is horn tops take less than 4'/2 months
to reach the ultimate low, on average. Annualizing the returns gives a score of
58%, enough to warrant a closer look.
A surprising finding is that when the volume on both spikes is low, below
the 25day moving average, then the failure rate drops to 6% while still main
taining a 21% average decline. That is something to keep in mind ifyou trade
this formation.
332
Tour 333
Tour
With many formations, there is usually a mirror image; with double bottoms,
for example, there are double tops. So it is with this formation. Chapter 22 dis
cussed horn bottoms and this chapter talks about horn tops. Having discovered
the bottoms, I wondered ifthe tops would work out as well. First, though, what
do horn tops look like?
Figure 23.1 shows an example of a timely horn top. If you read Chapter
22 on their bottom siblings, then horn tops should come as no surprise. A
horn top is an inverted version of the weekly horn bottom. A horn top sports
twin peaks separated by a week and is commonly found near the end of an
uptrend. Volume is usually heavy at both peaks but not by a huge margin above
the 25day moving average. After the right price spike, prices drop lower and
continue moving down, sometimes substantially.
In Figure 23.1 the stock begins its rise to the horn in midJune 1993 at a
price of 203
/8. At the peak, prices reach a high of 32s
/8, a gain of 60% in 2
months. With such a sharp gain in so little time, a consolidation or congestion
region is likely. Instead, the horn top marks a change in trend. Combined with
the earlier top, the double top is a bearish signal.
After the twin peaks of the horn appear, prices drop to 26, then pull back
to the formation base, generally following an upsloping trendline. Then prices
94 F M
Figure 23.1 Horn top showing twin peaks. The stock drops almost in half after
the horn top. Note the weekly time scale. The two peaks in April and August rep
resent a double top.

334 Horn Tops
head down again. In the beginning ofJanuary, prices reach bottom at 163
/4 for
a decline of almost 50%.
Table 23.1 shows identification characteristics that make horn tops easy to
recognize. Consider the horn shown in Figure 23.2. Use the weekly chart to
facilitate identification. Look for twin price spikes that are separated by a week.
The two spikes should be well above the surrounding prices (clear visibility)
and a good distance from die high in the center week. Ifyou look back over the
price history for the year, the two spikes should stand out and be larger than
most other spikes. In Figure 23.2, you can see that the price spike in late Sep
tember is the only real competition (for the period shown, anyway).
The hightohigh price variation between the two spikes is usually small,
about 3
/s or less. Occasionally, some horn tops will show large price discrepan
cies, but these usually are associated with higherpriced stocks. In the study, no
formation had a peaktopeak price difference of more than 11
A. Figure 23.2
shows a price differential of /g point. It is unusual for the two peaks to end at
the same price so expect some price variation.
Price overlap between the two spikes should be large. This guideline
means that you should discard any formation that has a long left spike but a
very short right one (or vice versa) or two spikes that have few prices in com
mon. Figure 23.2 shows two long spikes with much oftheir length overlapping.
Table 23.1
Identification Characteristics of Horn Tops
Weekly chart, upward spikes
Clear visibility
Large overlap
Low volume
Use the weekly chart and locate two upward price spikes
separated by a week. The two spikes should be longer
than similar spikes over the prior year and tower above
the high of the center week. It should look like a horn.
The price difference between the two horn highs is small,
usually or less, but can vary up to $1 or more for
higherpriced stocks. Do not expect the horn highs to
end at the same price; that only happens 20% of the
time.
The horn highs should be well above the surrounding
highs and the best performing reversals appear at the
end of a long uptrend.
The two spikes should have a large price overlap.
For best performance, look for below average volume on
both horn spikes.
Homestake Mining (Cold/Silver Mining, NYSE, HM)
N D 95 F M
Figure 23.2 Horn tops with unusually tall price spikes. This horn exceeds
clearance of an earlier spike.
the
The final guideline is not really a guideline at all as much as it is an obser
vation. When the volume on each spike is below the 25day moving average,
then the formation tends to have significantly fewer failures (the rate drops
from 16% to 6%).
Focus on Failures
There are a variety of reasons why a particular formation fails. Some break out
upward and never look back, whereas others begin moving down, falter, then
recover and climb significantly. The latter case, the socalled 5% failures, are
predominant with horn tops. Prices fail to continue down by more than 5%
before recovering and heading higher. However, with a failure rate of 16%, the
rate is still less than the 20% maximum for reliable formations.
Unfortunately, a 5% failure does not tell the complete story. With horn
tops the most likely decline is just 10%, hardly enough to cover the cost of
making a trade. Still, that performance is about par for bearish reversals in a
bull market.
What can be learned from examining the failures? There are 30 failures.
Of those formations, 15 precede a meaningful decline in the stock by an aver
age of 2.7 months. Most warn about a coming decline by forming less than 2

336 Horn Tops Statistics 337
months ahead of schedule. That is worth knowing. If a horn fails to call the
turn in a stock you own, be alert to a possible trend change coming soon.
You can improve your investment performance ifyou consider the over
all environment for the stock. Look at the horn top pictured in Figure 23.3,
After trending down for a year, the stock pierced the downsloping trendline,
signaling a trend change, and moved higher. Then the horn top formed.
In situations like this, after a long downtrend, a stock usually bounces up,
curls around, and retests the low (but not always). So when the horn formed, it
probably signaled the price top before the retest. Although not shown in Fig
ure 23.3, the stock did move lower, but it first bobbled up for a few months (to
63l
/i). Ultimately, the stock dipped to 49'/s before recovering.
Before you invest in this formation, you have to place emphasis on the
piercing ofthe down trendline. It suggests that prices will rise (a piercing is one
indicator of a trend change). With a price rise imminent, why would you con
sider shorting the stock? Even though the horn reversal suggests prices will
decline (and they do, in the short term, but only by a dollar or so), does it merit
a trade?
When there is serious conflict or doubt about a situation, then look else
where for a more promising trade. Sure, you might miss making a killing now
and again, but you do not want to end up on death row after your trading cap
ital runs out from all the losses you have been taking.
Figure 23.3 Horn top appearing after an extended downtrend. It is probably best
to ignore such horn tops. A pipe bottom marks the turning point.
Statistics
Table 23.2 shows statistics for the horn top. I uncovered 188 formations in 500
stocks over 5 years. As far as formations go, horn tops probably rank as some
what rare, but they are plentiful enough to be a viable trading vehicle.
The formation is almost evenly split among consolidations (92) and
reversals (96). The best performing formations are reversals, in which the horn
top accurately signals a trend change with prices moving lower.
The failure rate (16%) is good. I consider formations with values below
20% to be reliable. The average decline is 21% for successful formations.
However, the most likely decline is just 10% as shown in Figure 23.4. A fre
quency distribution allows a presentation ofthe losses without the undue influ
ence of large losses skewing the average. You can see that the highest column
is the 10% category, followed by 15%. Figure 23.4 suggests that your losses
will probably fall around these two numbers. Clearly 40% of the formations
show losses totaling 15% or less. Of course, if you are an optimist, it also
means 60% of the formations score better than a 15% decline. That is really
not bad for a bearish formation in a bull market.
Table 23.2 shows the average horn volume. The left horn has slightly
higher volume on average than the right spike, and they both exceed the 25
dayvolume moving average.
For those formations that move lower by more than 5% (anything less
is a failure, and excluded), they reach die ultimate low in about 4'/z months
(134 days).
Table 23.3 shows the benchmark, 21% average loss for the 84% of for
mations in die study that move significantly lower (by falling more than 5%).
The remainder of the table highlights different studies to determine the per
formance of various features of the horn top.
Do horns with uneven high prices perform better than those with the
same high price? No. Horn tops in which the highest price is the same on both
Table 23.2
General Statistics for Horn Tops
Number of formations in 500 stocks from 1991 to 1996
Failure rate
Most likely decline
Left, right horn volume versus 25day moving average
188
30 or 16%
21%
10%
1 38%, 125%

Figure 23.4 Frequency distribution of losses for horn tops. The most likely decline
is 10%, with 40% of the formations showing declines of 15% or less.
Table 23.3
Surprising Results for Horn Tops
Description
Average
Decline (%)
Failure
Rate (%)
Benchmark 21
Decline for horns with price differences
between highs 21
Decline for horns with no price difference
between highs 23
Percentage rise for high volume left (L) L21, R22, B21
horns, right (R) horns, and both (B) horns
Percentage rise for low volume left (L) L22, R21, B21
horn, right (R) horn, both (B) horns
High left horn volume, low right horn
volume 21
Low left horn volume, high right horn
volume 22
Linear regression price trend 3 months up 22
Linear regression price trend 3 months
down 20
Large price spike (3x average) 25
16
18 for both right
and left horns
6 for both right
and left horns
20
43
Note: Large spike length showed the largest loss but nearly a tripling of the failure rate.
338
Trading Tactics 339
price spikes perform better, with losses averaging 23% versus 21% for forma
tions with uneven high prices.
What are the volume characteristics of horns? The left spike, when
accompanied by low volume (below the 25day volume moving average), per
forms marginally better than when attended by high volume (22% versus
21%). The reverse is true for the right spike, with 22% losses for those with
high volume versus 21% for those with lowvolume. Performance is 21 % when
both the right and left spike show high volume (and a similar result for those
horns showing low volume on both spikes). When both spikes show below
average volume, the failure rate decreases to 6%, well below the benchmark
16% rate. Different combinations such as high left and low right volume on
the spike of a horn top show similar losses, namely 21%. The reverse, low left
spike volume and high right spike volume, scores a 22% loss. For all the dif
ferent combinations, there is not a large enough difference between the num
bers to bother with a test for significance. Does it really matter whether the
average loss is 21% or 22%? Probably not. Incidentally, the phrases high vol
ume and low volume are comparisons with the 25day moving average. High
volume is when the formation is above the moving average, and low volume is
when it is below the moving average. Think of the phrases not as high or low
but as above average and below average.
I use linear regression on the weekly price data over a 3month span to
assess the price trend leading to the formation. For those horn tops with rising
price trends, the average performance is 22% versus 20% for those with down
ward sloping trends. Again, not enough of a difference to get excited about.
Formations showing large price spikes outperform the benchmark with a
25% loss. Unfortunately, the loss accompanies a significant deterioration in
die failure rate, which climbs to 43%. There are only 28 formations with spikes
that are at least three times the average spike length over the prior year. In case
you are wondering how spikes twice the normal size do, they show losses aver
aging 22% accompanied by a failure rate of 27%.
As you look over the values in the table, you can probably surmise that it
really does not matter what the horn looks like or how much volume accom
panies the formation. In many cases, prices decline, on average, about 20% to
22%. However, to reduce the risk of a failed trade, it is probably wise to choose
formations with below average volume on both spikes (thereby reducing the
failure rate from 16% to 6%).
Trading Tactics
Table 23.4 outlines trading tactics for horn tops. With a likely decline of just
10% (but an average decline of 21%), should you trade this formation at all?
Sure, but it all depends on how it is used. If you own stock in a company that
has zoomed upward and forms a horn top, it is time to consider taking profits.

340 Horn Tops
Table 23.4
Trading Tactics for Horn Tops
Identify
Threat assessment
Low volume
Trend change
Use the characteristics outlined in Table 23.1 to correctly
identify a horn top on the weekly charts.
Look for an uptrend spanning many months. Such uptrends
often show horns near the end of the trend. If the horn top
appears near the end of a long downtrend, then it is best to
avoid it. Watch out for horns appearing after a downward
trend when the trend changes and starts moving higher.
Prices may decline but the decline is usually shortlived (as
in the rise between a double bottom).
The failure rate declines to 6% if below average volume
appears on both spikes. This is worth considering.
A horn top usually signals an approaching trend change,
usually in less than 2 months.
Perhaps the first and most important trading tactic is to be sure that you
have a valid horn top. Table 23.1 can assist you in your identification. After you
have correctly identified the horn, look at the surrounding price pattern. Is the
stock trending higher and has it been moving higher for months now? If so,
then the horn may signal an approaching top. Sometimes the horn calls the
turn exactly while at other times it is off by a few months (usually it precedes
the turn but sometimes it lags).
If the horn appears on a downtrend, that is fine. The formation is simply
implying that the trend will continue moving down. However, should die horn
appear after an extended downtrend, then the possibility of further declines
may be in jeopardy. This is especially true if the horn appears after a long
downtrend when prices are beginning to move up again (they are retracing
some of their losses). Often the retrace signals a trend change. You will be buy
ing into a situation in which you believe prices will fall and retest the low, but
prices will dip slightly then head higher.
Be cautious when selling short after a long downtrend and be especially
cautious if the downtrend has ended and the horn seems to mark the end of an
upward retrace. Do not expect prices to fall far. They do occasionally but the
majority of the decline has already passed.
In the Statistics section I point out the interesting result that spikes with
volume below the 2 5day moving average show substantially lower failure rates
while keeping the average decline at 21%. That is worth considering before
you place a trade.
Even if a horn top fails and prices either drop by less than 5% or move
directly higher, there is a chance that the horn is a premature signal of a bear
Sample Trade 341
ish trend change. Therefore, whenever you see a horn top, be aware that the
end of an uptrend may be just a few months away.
Sample Trade
You might be saying, "Those bromides are all fine and good, but how do I
really trade it?" Consider the situation in Figure 23.5 faced by Sandy. She
watched the complex headandshoulders formation take shape in a stock she
owned. Volume on the far left shoulder was higher than during the head, as
expected, and volume on the far right shoulder was further diminished. All in
all, it was setting up to be a dire situation. When she spotted the horn top
forming on the right shoulder, she knew die end was near.
Two weeks after the right horn spike appeared, she sold her holdings in
the stock for a tidy profit. But she was not finished. The following week, she
shorted the stock at 34'A. Her calculations using die measure rule for the head
andshoulder formation indicated a decline to 16. This was higher than the 12
predicted by die measure rule because she used her sell point instead of die
breakout point (which was not reached yet so she had no idea what it was).
Widi the stock selling at 34, a drop to 16 seemed too optimistic. Still, she
watched widi glee as die stock plummeted. In rapid order it fell to about 25
before meeting support. Sandy scanned the weekly chart looking for support
Figure 23.5 A horn top appears as part of the right shoulder of a complex head
andshoulders formation.

342 Horn Tops .'
levels and noticed one during October at about 24:
/2 and one during June at
around 233
/4. Together, these two levels and the round number of25 probably
spelled a difficult time for the stock to continue lower.
When the stock headed back up in late May, she decided to close out her
position at 27. She pocketed about $7 a share in just over 4 months. When she
returned to the chart a year later, she saw that it did descend lower as the
headandshoulders formation predicted. During early November, it reached
a low of 20'/8, still well above the calculated 16 target she used and the 12 pre
dicted by the measure rule. The flip side of this is that after she sold the stock,
it climbed to 303
/4, just 10% below her sell point.
24
Inside Days
RESULTS SNAPSHOT
Upside Breakouts
Appearance
Failure rate
Average rise
Surprising findings
Downside Breakouts
Appearance
Failure rate
Average decline
Surprising findings
A daily price range narrower than the prior day
Short term (up to 3 months) bullish consolidation
56%
Many, for both up and down breakouts. See Table
24.3 in the Statistics section.
A daily price range narrower than the prior day
Short term (up to 3 months) bullish reversal
51%
10%, with most likely loss between 0% and 5%
Many, for both up and down breakouts. See Table
24.3 in the Statistics section.
This formation had me perplexed for the longest time. The question I was
struggling to answer is, how do you measure success or failure ofthis 2day for
mation? The answer seems obvious now, but it was not when I started
researching this chart pattern.
343

344 Inside Days
The theory behind the formation says an investor can expect a large price
move the day after an inside day; that is how I ended up measuring success or
failure. You can see in the Results Snapshot that the performance (the failure
rate) is essentially random—near 50%. About half the formations have large
price moves immediately after an inside day and half do not.
Some analysts may argue that if you try to depend on a large move the
very next day, you are subject to disappointment. One must be openminded
and have patience, a large move will surely come along eventually. My
response is, how long should I wait? Two days? A week? A month? When will
the large price move occur?
Yes, several of the formations do experience a large price move 2 or more
days after an inside day, others do not. I decided to measure the performance
improvement when a large move is expected within 2 days. The results? The
upside failure rate improves to 46% (from 56%), whereas downside failures
decrease to 43 % (from 51%). That is an improvement but still well shy of the
20% maximum for reliable formations.
If I waited a week for a large price move, there would be even more
improvement. Of course, waiting a year would show an even larger improve
ment. So I ask again: How long should I wait? I settled on the more conserva
tive measure and expect a large price move the next day. You are certainly
free to continue the analysis for various time periods to see how the pattern
performs.
I separated inside days into two types: those with price trends moving
upward (upside breakouts) and those moving lower after the inside day (down
side breakouts). The determination of whether the formation is a reversal or
not depends on die trend before and after the formation. Inside days with
upside breakouts usually continue the upward, shortterm trend, whereas those
with downside breakouts seem to mark the end of the upward trend (acting as
reversals).
The average move is small (13% and 10%) and the most likely gain or
loss is less than 5% for each breakout direction. With other types of forma
tions, any move less than 5% classifies as a failure (the socalled 5% failure).
However, since the price moves from inside days are usually of small impact
and duration (many post lower lows the day after the inside day only to begin
moving higher; or the reverse: a higher high followed by a downward move).
Thus, the average gain or loss includes the many small moves as well as the
larger ones.
Table 24.1 outlines the identification characteristics and Figure 24.1 shows
numerous examples of inside days. An inside day is socalled because the daily
price range is inside the range of the prior day. At least, that is how I chose to
Table 24.1
Identification Characteristics of Inside Days
Lower high The daily high must be below the prior daily high. No ties allowed.
Higher low The daily low must be above the prior daily low. No ties allowed.
Daily range The daily high cannot equal the daily low. In other words, there
must be some sort of daily price range.
interpret the definition. Some technical analysts allow ties of the daily high or
low, but I do not. I also imposed a further restriction: The inside day needs a
daily price range. In other words, the intraday high and low cannot be the same
price. There is no real significance to this except for thinly traded stocks.
Thinly traded stocks have hundreds ofinside days over the 5year period under
study. It is bad enough that even with the stipulations shown in the table, the
number of formations is well over 100 for many stocks in the database. In the
Statistics section, I detail how I trimmed die number of formations to a more
manageable level.
The many black dots in Figure 24.1 highlight inside days. Although I des
ignate the inside day to be the day with the narrower price range, die chart pat
tern is really a combination of 2 days. As an example, consider the inside day
highlighted by die text in die figure (late October).
Alcoa (Aluminum Company of America) (Aluminum, NYSE, AA)
Oct 95 Nov Dec Jan 96 Feb Mar
Figure 24.1 Numerous examples of inside days, which are highlighted by the
black dots. The one in late October marked the start of a measured move up
formation.

346 Inside Days
The first day ofthe chart pattern has a high of 50% and a low of48H. The
following day—the inside day—the range is 50 to 485
/s. The inside day has a
lower high and a higher low than the prior day, and the daily price range is not
zero (meaning that the high and low are different values).
The primary beliefis that there will be a large price move after the inside
day. You can see in Figure 24.1 that the day after the chart pattern prices move
up smartly. One could argue that the inside day acts as a reversal of the down
ward price trend (while the inside day that happens 3 days before is a consoli
dation of the downward trend—prices continue moving down after the
formation). Prices rise from a close of491
/4 to 54'/8, pause for a week, then con
tinue up to reach a high of 59 !
/8 in late November. Astute readers might rec
ognize the price pattern as a measured move up.
One has to ask the obvious question and that is, is an inside day a real
chart pattern or is it just a name attached to a random event? That is a good
question and one for which I am not sure I have the answer. If you believe
in the strict version of the theory, that a large price move follows the day
after an inside day, then the formation is just a random event. With failure
rates over 50%, it is more accurate to say a small price movefollows the day after
an inside day.
Ifyou are not so strict and simply say that a large price move follows (with
no time limit), then the price pattern works out better. This is really no sur
prise when you consider that we begin with a comparatively narrow price
range. Eventually we will get either a wide price range or prices will move up
or down. The real question is, what does the term large price move mean?
I define the term to mean that prices the day after an inside day either
climb or descend beyond the high or low posted the day before the inside day
(ofthe 2day formation, that is the day with the larger price range). If this does
not happen, then a wider intraday price range, one that appears to be a multi
ple of the inside day's range, is also acceptable. Ifboth these conditions are not
met, then the formation is a failure.
Focus on Failures
When considering a formation such as an inside day, the normal gauges offail
ure really does not apply. The notion of a 5 % failure, when prices move less
than 5% in the direction of the breakout, is meaningless when the formation
lasts just 2 days and the ultimate move might constitute a higher high or lower
low the very next day.
Instead, I chose to rate the formation on how well it obeys the theory that
prices make a large move the day after an inside day. Why the time limit? Two
reasons really. First, eventually a stock is going to make a large move, so plac
ing a time limit in which to expect a move makes sense. Second, we are in the
business of making money. Our job is to find formations that make the most
money in the shortest possible time, so a time limit in which a formation must
perform also makes sense.
Exactly what constitutes a failure (alternatively, what does a large move
mean)? As mentioned in the Identification Guidelines section, a failure is when
prices do not move very far the day after an inside day. Figure 24.2 highlights
two failures. The first failure comes after an inside day in late January. The
inside day has a price range of 15'/4 to 143
/4. The following day the range is
exactly the same. Clearly, a large price move does not immediately follow the
inside day. The second failure is similar in that die daily price range does not
change much from the inside day, nor do prices move very far. Yes, the low is
lower than the inside day but by less than 3
/i6—not a very compelling move.
Contrast the two failures with the move following the inside day in mid
August. Prices gap (a breakaway gap, incidentally) upward and the daily low is
above both the highs of the prior 2 days (which makes up the inside day forma
tion). This inside day leads to a large price move.
What about a formation that is not so obvious? For that I needed to
establish some rules. I consider the formation a success if the day after an inside
day has a higher high or lower low than the day before the inside day (and not
by just a little either). In other words, a price trend should develop. Ifprices fail
Acclaim Entertainment Inc. (Computer Software & Svcs, NASDAQ, AKLM)
|an 93 Mar Oct
Figure 24.2 Two failures among the many inside days (black dots). A large price
move does not follow the two inside days and so they are failures. A breakaway
gap appears after the August inside day.

348 Inside Days
to move higher, then a larger intraday price range, one that is a multiple of the
inside day, constitutes a winning combination.
For example, point A in Figure 24.2 shows what I consider to be a large
price move after the inside day. The daily low of 14'/2 matches the low of the
day before the inside day and the high matches the inside day's high. The price
range looks like a large price move when compared with the inside day.
Perhaps my rules are too strict because the failure rates are abnormally
high. In the next section, Statistics, I discuss the numerical results.
Statistics
Figures 24.1 and 24.2 illustrate one important fact: Inside days are plentiful.
They are so numerous that I chose to limit the number of formations I would
examine (many stocks have over 100 formations in 5 years). Fortunately, this
formation is easily recognized by the computer with complete accuracy.
Instead of tabulating the performance of just five stocks (for about 500
formations), I chose to spread the selections over many stocks. To do this I
limited the number of formations to 10 per stock by counting the number of
formations in the stock and skipping the appropriate number.
For example, if a stock has 20 formations, I tabulate every other forma
tion. A stock with 100 formations means that I use every tenth pattern. Not
only does this technique spread the formations over the 5year span, but it
enlarges the number of stocks from about 5 to 52. The large number of for
mations tabulated and the spreading of the formation in both time and across
many stocks ensure that the sample pool is diverse and representative of the
performance of any one formation.
I separated the formations into those with prices trending up after the
formation (upside breakouts) and those with trends moving down (downside
breakouts). An upside breakout means a higher close or a higher high the day
after the inside day. A downside breakout means a lower close or a lower low.
Ifnone ofthe three (high, low, or close) change, then I examine additional days
until a trend determination can be made.
Table 24.2 shows performance statistics for inside days. I included 243
formations with upside breakouts and 277 with downside breakouts from 520
formations in 52 stocks over 5 years. Formations with upside breakouts usually
act as consolidations of the shortterm upward trend (151 versus 92 reversals),
whereas a slight majority of formations with downside breakouts (123 consol
idations versus 154 reversals) mark a reversal of the upward trend.
The failure rate, which gauges whether there is a large move the day after
an inside day, is 56% and 51% for upside and downside breakouts, respec
tively. I consider failure rates below 20% to be acceptable, so this is well into
Statistics 349
Table 24.2
General Statistics for Inside Days
Description
Upside
Breakout
Downside
Breakout
from 1991 to 1996, limited to 10
formations per stock
Failure rate
Average rise/decline of successful
formations
ultimate high/low
Percentage of formations occurring
near 12month low (L), center (C),
or high (H)
Percentage gain/loss for each 12
month lookback period
Volume day before to day after versus
25day moving average
243
151 consolidations,
92 reversals
137 or 56%
13%
05%
L24%, C24%, H52%
L27%, C6%, H11%
140%, 83%, 93%
10%
22%
277
123 consolidations,
154 reversals
141 or 51%
10%
05%
2 months (65 days) 22 days
LI 7%, C34%, H49%
L11%, C11%, H8%
140%, 83%, 93%
13%
9%
the twilight zone. Thus, it is more accurate to say that most formations make
a small move the day after an inside day.
The average gain is only 13% and die average loss is 10% for die two
breakout types. This is well below die 40% for upside breakouts and 20% for
downside breakouts that formations commonly share. The reason for this poor
showing is probably because ofthe way the ultimate high or low is determined.
In the case of the ultimate high, die rule is diat when the trend changes, the
ultimate high is the highest high posted between the formation and die trend
change. Additionally, if prices break out upward dien dip below the formation
low, that truncates die period used (the reverse is true for downside breakouts).
With a formation as short as an inside day, there is not much latitude
before prices move below die formation low. Most other formations have a
wider price range that allows the stock plenty of room to recover before drop
ping below the formation low and calling an end to the search for die ultimate
high. The narrow formation height sometimes penalizes performance.

350 Inside Days
Both upside and downside breakouts have likely gains of less than 5%.
Figure 24.3 shows a frequency distribution ofthe gains (formations with upside
breakouts) and losses (downside breakouts) for the inside day chart pattern.
Both upside and downside breakouts have the highest column in the zero to
5% range. Almost half (49%) of the formations with downside breakouts have
losses less than 5%. Twothirds (or more) of the formations have returns less
than 10%. Such meager returns suggest it is unwise to depend on this forma
tion for any substantial lasting move.
Oddly enough, it takes almost three times as long to reach the ultimate
high after an upside breakout (65 days) than the low in the downside breakout
(22 days). A review of die data shows that there are a number of long duration
climbs to the ultimate high. The large numbers unfairly pull the average
upward. Removing all tripledigit durations (of which there are 13) results in
an average climb to the ultimate high ofjust 13 days. A similar computation for
the ultimate low reveals prices reach the low in 1 7 days.
Do inside days occur near the yearly high, low, or somewhere in between?
A frequency distribution of the closing price to the prior yearly price range
reveals that those formations with upside breakouts occur predominately in the
upper third ofthe price range. A similar situation exists for downside breakouts
with 49% of the formations residing within a third of the yearly high.
Mapping the performance onto the yearly price range is only revealing
for upside breakouts. The best performing inside days with upside breakouts
Figure 24.3 Frequency distribution of gains and losses for inside days. Both
upside and downside breakouts from inside days have moves typically less than 5%.
Statistics 351
are those with closing prices within a third of the yearly low. They have gains
averaging 27%. The worst performing upside gain is for those with closing
prices in the center third ofthe yearly price range. They have average gains of
just 6%. Downside breakouts are unexciting, scoring losses within 3% of each
other. The numbers suggest that ifyou are going to trade inside days, then you
might pay attention to where they occur in the yearly price range. Choose ones
with upside breakouts near the yearly low.
The volume pattern is what you would expect. The day before an inside
day shows the highest volume at 140% ofthe 25day moving average. A higher
number of shares change hands when a large price range occurs as compared
to the inside day when the volume drops to 83% of the average.
I looked at the performance for inside days with high and low volume sep
arated by their breakout direction. For inside days with upside breakouts and
volume that is half the 25day moving average, they score gains of 22%, well
above those inside days with volume 50% (1.5x) above average. However, the
reverse is true for downside breakouts. Inside days with high volume perform
better than those with low volume. Therefore, the breakout direction and vol
ume pattern can be a clue to the performance of inside days.
Table 24.3 lists surprising findings for inside days. The failure rate
explains the first finding, that of whether the formation lives up to expecta
tions. The belief is that a large price move follows an inside day. Over half the
formations fail to show such a move.
The next several table entries determine whether the breakout direction
is predictable. Does the closing price the day before the inside day predict the
breakout direction? No. I divided the intraday price range into three parts,
25%, 50%, and 25% ofthe range. Then I tagged in which part the closing price
resides. I looked to see if an upside or downside breakout relates to whether the
closing price is in the top 25% or bottom 25% of the daily price range. The
results hover around 50%—essentially random.
Table 24.3
Surprising Results for Inside Days
Description Results
Do inside days predict a large price move the next day?
Does the closing price of day before inside day predict breakout
direction?
Does the closing price of the inside day predict the breakout direction?
Does the closing price of the inside day versus prior day predict
breakout direction?
Do inside days result in larger price ranges the next day?
The smaller the daily pricerange ratio between the day before and the
inside day, the larger the rise or decline
No
No
Yes
Yes
56% have
larger ranges
N/A

352 Inside Days
I applied the same test to the inside day instead of the day before and the
results change. When the close is within 25% ofthe daily high, 61% ofthe for
mations have upside breakouts. When the close is within 25% ofthe daily low,
60% of the formations break out downward.
Next, I compared the close of the inside day with the prior day's price
range. When the price ofthe inside day closes within 25% from the prior day's
high, an upside breakout occurs 58% ofthe time. That result is not much bet
ter than a random event (50%). However, when the close is within 25% of the
prior day's low, a downside breakout occurs 70% of the time. Thus, ifyou have
an inside day that closes near the prior days'1
low, expect a downside breakout.
Do inside days result in large price ranges the next day? The average
daily price range is $1.05 the day before the inside day, then it drops to $0.47
during the inside day and climbs to $0.64 a day later. A frequency distribution
of the results says 44% of the days after an inside day have an equal to or
smaller daily price range. Alternatively, 56% have larger daily price ranges.
The last finding in the table deals with the ratio of two daily price ranges.
I computed the daily price range the day before the inside day and divided it by
the daily price range ofthe inside day. Then I compared the ratios with the ulti
mate gain or loss. A scatter plot of the results indicates that chart patterns with
smaller ratios perform better than those with larger ones. Figure 24.4 shows
Figure 24.4 Price range versus ultimate loss. The graph shows the ratio of the
price range the day before to the inside day's range compared with the resulting
percentage decline for downside breakouts. The graph suggests smaller price
ratios perform better.
Trading Tactics 353
what I mean. Ofthe two charts, one for upside and one for downside breakouts,
the figure shows the more pronounced effects of the two; it is for inside days
with downside breakouts.
At first, Figure 24.4 looks like a random splattering of dots. But, upon
reflection, you can see that as the pricerange ratio gets smaller, performance
improves. By this I mean that there are more dots in the lower portion of the
graph extending to the right than in the left portion extending upward. The
graph for upside breakouts is similar, but the effect is less prominent.
Why is this the case? I compare the situation to symmetrical triangles.
Those formations are like coils, tightly winding until the spring releases and
prices shoot upward or downward in an explosion of activity. An inside day is
just a 2day triangle.
The analysis considers the ratios of the price range for the inside day
with the prior day. What if the real effect is the formation height? Do shorter
formations perform better than taller ones? Here the effect is just the reverse.
Formations with downside breakouts have a scatter plot that is essentially ran
dom with dots all over the chart. However, the chart ofupside breakouts shows
a slight tendency for shorter formations to perform marginally better than
their taller counterparts. By shorterformation, I am talking about the height
(daily price range from high to low) of the day before the inside day. Using
inside days instead of the day before shows an essentially random relationship
for both upside and downside breakouts.
Trading Tactics
Without degenerating into a discussion of the random walk and the likelihood
of tomorrow's price being higher is just 5050, there are scant trading tactics
that I can offer, certainly not enough to fill a table. If you still need a bromide,
then let me say this: Ifyou are going to rely on an inside day to place a trade,
then trade in the direction of the trend. Wait for prices to stage a breakout by
closing either above the top or below the bottom of the inside day. If the break
out is upward, then go long; ifit is downward, then short the stock, run to your
favorite place of worship, and pray!
In the Statistics section I briefly mention the next two suggestions. The
best performing inside days, with upside breakouts, occur in the lowest third of the
yearly price range (with gains averaging 27%, well above the 13% average).
Downside breakouts show no meaningful performance difference wherever
they break out in the yearly price range. Pay attention to the volume characteris
tics ofinside days. Upside breakouts on low volume and downside breakouts on
high volume often perform better than other combinations. See Table 24.2 and
the Statistics section for more information.

354 Inside Days Sample Trade 355
Figure 24.5 Inside day with a hanging man formation. Nathan successfully posi
tion traded this inside day for a $2,000 paper gain. A hanging man formation
appears in midDecember.
Sample Trade
Nathan has a conservative job in the banking industry but in his spare time he
likes to take out his aggression by paper trading stocks with the hope ofone day
becoming a fulltime position trader. When he saw the inside day develop in
Airgas, he decided to paper trade it from the short side (see Figure 24.5). Widi
the knowledge that the closing price of an inside day might suggest the break
out direction, he believed that the stock would tumble—or so he hoped. The
day after the inside day, he sold the stock short and received a fill at 14M>.
As predicted, the stock closed lower for the day but quickly retraced its
progress over the next 3 days. Just as he was about to cover his position and get
out, the stock reversed direction and turned down. Three days later it slipped
below 13, for a tidy paper gain on his 1,000 shares.
Then, die stock jumped upward (point A on the chart) but Nathan was
not aware of the sharp move until well into the following day. He drew a
downsloping trendline along the minor daily highs and saw that the upmove
had but just barely pierced the trendline. This signaled a possible trend change.
However, when he got a quotation from his broker, the stock was already mov
ing back down.
He flipped a coin and decided to hold on to the position a little longer. At
the end of the trading day, the stock closed lower, back below the trendline.
Realizing that his decision was little more than a crap shoot, he decided that
should the stock pierce the uptrendline and close above it, he would cover
his short.
Each day he plotted the stock and watched its progress. In midDecember,
when the stock closed above the trendline after a hanging man formation,
Nathan decided it was time to check out. He covered his short and received a
fill at 127
/id, making him about two grand richer. "Paper trades are easy!" he
snorted.

25
Island Reversals
RESULTS SNAPSHOT
Tops
Appearance
Failure rate
Average decline
Volume trend
Fullbacks
Percentage meeting
Bottoms
Appearance
Failure rate
Average rise
Volume trend
Throwbacks
Percentage meeting
356
Prices gap up to the formation then gap down at the
same price level, leaving an island.
13%
21%, with most likely decline less than 10%
Downward
65%
78%
Prices gap down to the formation then gap up at the
same price level, leaving an island.
17%
Downward
70%
85%
Tour 357
The performance ofisland reversals is perhaps surprising only for its mediocrity.
Failure rates for both tops (13%) and bottoms (17%) is quite good, below the
20% threshold that I view as the maximum allowable for reliable formations.
For tops, the average decline is a very respectable 21%, but the most
likely decline is less than 10%. This suggests you would not want to make a
habit of trading this formation. For bottoms, the average rise is 34%, reason
able for bullish formations but still not up to the caliber of other bottom
reversals. The most likely rise is less than 20%, about average for bullish
formations.
Fullbacks and throwbacks are prevalent, suggesting that the gap after the
island completes closes quite quickly. Investors can make use of this behavior
to delay their investments until the pullback or throwback completes and prices
resume their breakout direction.
Tour
Figure 25.1 shows what island reversals look like. The first island, shown on
the far left, is a oneday reversal. In the study ofislands, I did not tabulate such
narrow formations, but the figure shows an example of a small reversal. The
center formation is an island bottom. Prices gap downward in midSeptember,
reach a new low in early October, then gap upward later in the month. The
Aug 91 Sep
Figure 25.1 A oneday reversal, island bottom, and island top. You can see the
price change that results after each reversal.

358 Island Reversals
two gaps appear at about the same price level, 11'/2. From that point, prices
climb quite rapidly and reach a high of 2ll
/2, well above the lll
A price posted
the day before the breakout. Notice that the breakout is on very high volume.
The last island highlighted on the chart is near the top. It is an island top and
prices head down from about 20 to less than 8. This formation is a traditional
island top because of its compact size. I did not count it in my tabulations
because the gap on the right is less than the K point I used to filter such for
mations. Still it does emphasize a trend reversal with excellent timing. Not
highlighted is a very large island top. The island bottom shares the left gap
(point A) while the right gap is the large price decline in midJune (point B).
The large island is almost 8 months long.
Island reversals are easy to identify and Table 25.1 shows the identification
characteristics. Both types of reversals, tops and bottoms, are set off by gaps.
The first gap is usually an exhaustion gap (or a breakaway gap if the move is
strong enough) and the second one is a breakaway gap. The gaps appear at or
near die same price level but are typically not the same size. The large gap
marked as point B in Figure 25.1 (on the far right) makes this clear. That gap
has a size of almost 5 while the one on the left (point A) is only 3
/g wide.
As far as identification goes, the gap in midDecember 1991 (point C)
should not be paired with the gap in late October (point A) since they are not
at die same price level. Even diough the price pattern looks like an island
Table 25.1
Identification Characteristics of Island Reversals
Shape Gaps set off both island tops and bottoms and share all or part of
the same price level. Most times, prices move away from the gaps
leaving the island with a clear view of the opposing gap (the left
gap usually does not close quickly).
Rising trend Tops have prices that lead up to the left gap and fall away from
the right gap.
Falling trend Bottoms have prices that lead down to the left gap and rise away
from the right gap.
Volume Volume is usually high on the breakout day (the day prices make a
second gap and form the island) but need not be.
Time The island can be from one day (a oneday reversal) to several
months long. Some analysts have suggested that islands are quite
short, up to a week or two, with relatively flat price zones, but I
placed no such restrictions on them.
because it is set off by gaps, it is not an island reversal by the traditional defin
ition. The gaps must overlap prices or be quite close to one anodier. In this
study, all islands have price gaps widiin l
/s of one anodier and are at least H
point or larger. In addition, I did not consider oneday reversals (islands com
posed of 1 day) as part oftiiis study. I believe that such islands are too difficult
on which to base an effective trading policy.
Consider Figure 25.2, which shows several island reversals. The first two
islands happen as part of a retrace in a downtrend. The small island tops last for
about a week before prices resume their downward plunge.
The island bottom forms after an extensive decline that sees prices drop
from a high of 36'/2 to 12 in less dian 5 mondis. The exhaustion gap occurs on
very high volume and prices diat day have a very large 3 '/2 point trading range.
The gap remains open until prices gap upward in late January. Looking at the
bottom formation overall, it looks like a complex headandshoulders bottom.
Marked on die figure are the dual shoulders and duplicate heads. The volume
for die headandshoulders pattern is what you would expect: highest on die
left shoulder, diminished on the head, and quite low on the right shoulder.
Only after prices gap upward does volume spike higher for 2 days before recov
ering and trending downward.
The island top on the right of die figure is somewhat difficult to spot
because of the large gap on the left that matches the small one on die right. It
Figure 25.2 Several island reversals, some with short durations and some with
longer durations. The island bottom is also a complex headandshoulders bottom
that retests the neckline in April.

only takes a few days for prices to reach their high before easing down. When
prices gap lower at the end ofMarch, volume does not budge. This may signal
a weak trend and it turns out that prices do not make consistent new highs for
at least a year.
Focus on Failures
Failures come in all manners of depiction. Look at Figure 25.3, a chart of a
large island top. This failure is typical of many formations, especially those
with small declines. The island is unusual as it forms after a region of consoli
dation. The first gap is a breakaway gap since it breaks away from the consoli
dation region on high volume and prices move up. The second gap is an
exhaustion gap that closes quickly.
The figure shows a 5% failure. Prices head lower after the second gap on
the right, but decline by no more than 5 % before recovering and moving sub
stantially higher. Why? There is a common law, for lack ofa better term, that
says a reversal will only travel as far as the prior rise. In other words, a reversal
has to have something to reverse. In Figure 25.3, you can see that prices con
solidate for 2 months. When the island top appears, what is there to reverse?
The rise from December to midFebruary, when prices rise from roughly 55
to 65, unwinds. Prices move steadily lower at first then plunge and end back at
56. Beyond that, the prior year's worth ofsupport at that level is just too exten
sive to allow any further decline. There has to be something to reverse. Remember
that before you take a position in a stock, especially something such as an
island reversal that is known to be light on performance.
Figure 25.4 shows another failure. The island bottom is clearly visible on
the chart. The two gaps separate the main body of the island from the main
land with plenty of clear ocean. Over the course of the formation, the first gap
remains open. The second gap closes about a week after prices throw back to
the breakout point. Then prices continue lower.
How could you have known that this island bottom would rise by less
than 5% before sinking? The question is irrelevant because of throwbacks.
For island bottoms, throwbacks occur 70% of the time, on average. This high
rate allows an investor the luxury of waiting before investing. Once the stock
throws back to the formation, wait for prices to recover. Assuming they do,
jump in and buy the stock.
In the situation shown in Figure 25.4, the stock did not recover; it con
tinued down. Had the investor been paying attention, he would not have got
ten himself into this moneylosing situation in the first place.
Is there another reason to suspect this formation? Sure, and it is called a
trendline. If you draw a downsloping trendline along the peaks, beginning
with the tallest one, you quickly discover that the minor peak after the island
Oct93
Figure 25.3 Long island top that fails because prices drop by less than 5%. A
reversal needs something to reverse.
May 93 |un |ul Aug Sep Oct Nov Dec Jan 94 Feb
Figure 25.4 An island bottom failure. Wait for prices to close above the trendline
before investing in this island bottom. Since prices do not close above the trend
line (at point C), skip this trade.

bottom (point C) falls well short of the trendline. Trendlines are known for
their ability to support or resist price declines and advances, respectively. In
this case the trendline acts as resistance to the rising trend. It repels the
advance and prices turn lower.
In odier words, die investor should have waited for prices to close above
the trendline before buying the stock. Even if he mistakenly connects points A
and B with a trendline and extends it downward through point C, it should give
him pause. Although prices pierce the trendline, diey fail to close above it. If
they cannot close above it, how will prices rise?
Statistics
Table 25.2 shows general statistics for island reversals. I searched all 500 stocks
for island tops and bottoms and located 245 tops and 278 bottoms over 5 years
of daily price data. The failure rates are good at 13 % and 17%, both below the
20% level that I consider a maximum for reliable formations.
The average decline (21%) and rise (34%) are about what you would
expect for bearish top and bullish bottom reversals. The island bottoms could
be a bit stronger to lift them up with some of die better performing bottoms
(they gain about 40%). However, the real worry is the most likely decline and
rise. For tops die likely decline is 10% or less and for bottoms, the likely gain
is less than 20%. Figures 25.5 and 25.6 illustrate this. Both figures show a fre
quency distribution of losses or gains. I use a frequency distribution to remove
any unwanted skewing of large numbers that affect an average. The tallest col
umn then becomes what I term die most likely decline (for tops) or rise (for
bottoms).
Table 25.2
General Statistics for Island Reversals
Description
Number of formations in 500 stocks from 1 991 to 1 996
Failure rate
Average decline/rise of successful formations
Most likely decline/rise
Number of successful formations showing downward
volume trend
Tops
245
32 or 1 3%
21%
1 0% or less
1 66 or 78%
40 days
1 88 or 77%
Bottoms
278
46 or 1 7%
34%
20%
1 98 or 85%
31 days
205 or 74%
Figure 25.5 Frequency distribution of losses for island tops. Note that the most
likely loss is less than 10%.
Figure 25.6 Frequency distribution of gains for island bottoms. Island bottoms
have a likely loss of 20%.
363

Figure 25.5 shows the column with the highest frequency as having a loss
of up to 10%. You can see that almost 60% of the formations have gains aver
aging less than 20%. That is a sobering thought. The figure warns you not to
expect too much of a decline from an island top in a bull market.
Figure 25.6 shows a graph of gains for island bottoms. The most likely
gain is the highest column at 20% but the 10% column follows closely. In this
case, 42% of the formations (or almost half!) have gains of less than 20%.
Most measure rules use the height of the formation, so I computed the
target price for each successful formation and tabulated whether the pattern hit
its target. Both tops and bottoms do quite well, with 78% and 85%, respec
tively, reaching their predicted prices. I consider values over 80% to be reli
able, so these are close enough for comfort. Of course, if the formation is not
very tall, then it is comparatively easy to make the grade. This is especially true
when just 2 days' worth of prices compose an island (the days have gaps on
either side).
The average formation length is quite short, about a month or so for
both tops and bottoms. Again, the number of small duration islands influences
the duration. Still, when you review all the data, the formations are generally
quick to reverse. About half of all islands (43% for tops and 50% for bottoms)
last 2 weeks or less.
Over the course of the formation, 77% of the tops and 74% of the bot
toms show a receding volume trend. I use the slope of a line found using linear
Table 25.3
Additional Statistics Related to Island Reversals
Description Island Tops Island Bottoms
Pullbacks/throwbacks
Average time to pullback/throwback
completion
ultimate low/high
Percentage of islands occurring near
12month price low (L), center (C),
or high (H)
Percentage change for each 12
Volume for breakout day and next
5 days compared with day before
breakout
Performance for high volume
breakouts versus low volume
breakouts
160 or 65%
8 days
L19%, C37%, H44%
128%, C22%, HI 9%
306%, 187%, 141%,
120%, 126%, 113%
22% versus 16%
195 or 70%
9 days
6 months (178 days)
L34%, C28%, H38%
L31%, C33%, H35%
201 %, 124%, 88%,
84%, 86%, 86%
33% versus 39%
Trading Tactics 365
regression on the volume data to derive this conclusion. In many instances, the
declining volume trend is clearly visible on the chart.
Table 25.3 lists additional statistics. Island tops have pullbacks 65% ofthe
time, whereas bottoms have throwbacks to the gap 70% of die time. In both
cases the numbers are quite high and suggest a trading tactic to make use of
this behavior (that is, wait for the pullback or throwback). I discuss Trading
Tactics in the next section.
The average time to a pullback or throwback is a short 8 or 9 days for tops
and bottoms, respectively. This is slightly shorter than other chart patterns.
For successful formations, the time to die ultimate low or high varies
quite widely between tops and bottoms. For tops, it takes 72 days to reach the
low, but 6 months for bottoms to climb to the ultimate high. Perhaps this
should not be so surprising since it often takes longer to move farther.
I separated the formations into three categories, depending on their
breakout price within the yearly range. Most of the island tops and bottoms
occur in the highest third of the yearly price range. Although that may sound
strange for a bottom, it is not. That is because I define an island bottom to be
when prices decline to, then rise away from, the formation. Tops, on the other
hand, have prices that lead up to, then down from, the formation.
When we substitute performance into the same three categories, we dis
cover that island tops occurring in the lowest third of the yearly price range
perform best with losses averaging 28%. The performance suggests that weak
situations (an island near the yearly low) get weaker (move down even further).
Bottom performance is about evenly split and ranged from 31 % to 35%.
The table shows volume statistics. Clearly, the breakout day, which is the
day prices gap away from the formation, shows very high volume. For tops the
volume remains high through the following week but with bottoms it quickly
calms down and remains quiet.
I did a study to see if high volume breakouts push prices farther. For
island tops, the answer is yes, as those formations with volume 125% of the
prior day show losses averaging 22%, whereas formations with volume 75% of
the prior day have losses of just 16%.
For bottoms, the reverse is true. A low volume breakout scores better
with a 39% rise, whereas those formations with high volume have gains of
33%. In both cases, the 125% (high volume) and 75% (low volume) bench
marks are arbitrary designations.
Trading Tactics
Table 25.4 outlines trading tactics. The first trading tactic is not really a tactic
at all; it is the measure rule, which assists investors in gauging whether a trade
is worth risking. Consider the island bottom shown in Figure 25.7. The high

Table 25.4
Trading Tactics for Island Reversals
Measure rule
Wait for pullback or throwback
Watch trendlines
lowest low from the highest high in the formation.
Add the difference to the highest high for island
bottoms, and subtract the difference from the
lowest low for island tops. The result is the target
to which prices should rise (for bottoms) or fall
(for tops).
Island reversals show a reluctance to continue
moving in the direction of the breakout. Prices
usually reverse direction and quickly fill the gap
before recovering and resuming their original
trend. Wait for the pullback (tops) or throwback
(bottoms) to complete and prices to resume their
original direction before investing.
Trendlines, when pierced, often signal a trend
change. Should an island reversal appear near a
trendline, wait for prices to close beyond the
trendline before investing.
Sep 95 Oct Nov Dec Jan 96 Feb Mar Apr May ]un Jul
Figure 25.7 A failed island bottom. Sometimes the best trade you can make is
none at all.
366
Sample Trade 367
est high in the formation is 24% (which is just below the gap in early February,
not the larger gap in midFebruary) while the low is 17'/2. Add the difference
of 7'/8 to the highest high to get the predicted price target. In this case, the tar
get is 3 !3
/4, a target not reached before the formation fails.
Since pullbacks and throwbacks occur a majority of the time and since
island reversals have a low failure rate but poor performance, it is wise to wait
for the retrace. It usually occurs a week or two after the second gap. If a pull
back or throwback does not occur quickly (in less than a month), then move on
to the next trading situation.
When a pullback or throwback occurs, do not invest immediately. Wait
for the retrace to complete and for prices to turn around and resume their orig
inal direction. Sometimes prices retrace to the formation, then continue mov
ing in the adverse direction.
In die Focus on Failures section of this chapter I discuss the use of the
trendline in detail, so there is not much added here. However, both up and
down trendlines can show a trend change. Wait for prices to close above a down
trendline or below an up trendline before pulling the trigger. Many times
prices will near the trendline and be repulsed, so you want to make sure that
the piercing does, indeed, signal a change in trend.
Sample Trade
Consider the situation faced by Clarence as illustrated in Figure 25.7. He
watched the semiconductor company's stock plummet. During November and
December, the stock formed a type of island consolidation. He knew it was not
a reversal because the two gaps did not line up across from each other. Then
another island formed in January to early February.
Since prices gapped down to the second formation then gapped up away
from it, he knew he was dealing with an island bottom, a better investment
choice for performance than an island top. He used the measure rule to gauge
the likely price to which the stock would climb. The target represented nearly
a 30% rise in price, large enough to risk a trade.
Before he bought the stock, he made a few checks. He saw that the trend
was down, as the stock had fallen from a high of 6ll
/s to the island low of 17!
/2.
Clarence drew the trendline from the highest high downward and saw it go
through the right island gap. This was a good sign as prices had moved above
the line and closed there. It signaled a trend change.
Still, something did not feel right about the stock. It had made a new low
and the semiconductor industry as a whole was soft. Did the island bottom
really mark a turning point or would the stock simply rise up, spin around, and
retest the low? He was unsure, so he decided to wait and see if prices threw

back to the formation. Ifthey threw back then continued higher, he would buy
the stock. Three days after the upside breakout, the stock threw back and
closed the gap. Now, Clarence knew, all the stock had to do was move higher.
It did not. The stock continued moving down and in less than a week had
slipped below the trendline again. He decided to look elsewhere for a more
promising situation.
Looking back at the stock well over a year later, he saw that it reached a
low of 8 and never rose above 27I
/4, the high just after it pierced the trendline.
He realized that sometimes the best trade you can make is none at all.
26
Measured Move Down
RESULTS SNAPSHOT
Appearance
Failure rate
Average decline
First leg decline
Corrective phase rise
Last leg decline
Measure rule
Surprising finding
Synonym
See also
Prices move down, retrace, then move down again.
The two down legs are nearly equal in both price and
time.
Intermediateterm (up to 6 months) bearish reversal
22%
25% in 54 days
16% in 39 days (40% to 60% retrace of prior decline)
27% in 60 days
43% of second legs are longer than first leg price
decline.
The corrective phase should be in proportion to the
first leg decline.
Swing measurement
Flags and Pennants
The measured move down, or swing measurement as it is sometimes called, is
an exciting formation because it vividly tells you how far down it is going.
Unfortunately, with a failure rate of 22%, it is also more risky. I consider
369

formations with failure rates less than 20% to be reliable. That is not to say
that this one is unreliable, especially since it has a 36% average decline, which
is well above the usual 20% decline for bearish reversals. For this formation
even the most likely decline is high but wide at 25% to 40%.
When we examine the two down legs of this formation, we find that they
are nearly the same length, 60 days versus 54 days, with declines of 25% ver
sus 27%. On a percentage basis, the second leg seems longer, but on a price
basis, the second leg is 8% shorter. Remember that as prices fall, they have to
travel less far to make the same percentage decline. This helps explain why
only 43% of the second legs exceed the first leg price decline (in other words,
the measure rule).
A surprising finding that helps with gauging the veracity of the formation
and its performance is the size of the corrective phase. A large first leg price
decline should also show a large corrective phase retrace. If the corrective
phase falls short of the usual 40% to 60% retrace, then beware. Be prepared to
exit the trade well before reaching the target price.
Tour
Figure 26.1 shows what a typical measured move down looks like. The decline
from the high (point A) to the start of the retrace (point C) is called thefirst kg.
The retrace is commonly referred to as the corrective phase and the remaining
Fleetwood Enterprises (Manuf. Housing/Rec. Veh., NYSE, FLE)
Mar 92
Figure 26.1 A typical measured move down. The slope of the trendline is similar
for both legs. Points A, B, and C mark a nested measured move.
decline to the low is called the second leg. The first and second legs are nearly
the same size, but their behavior is described in more detail later in the Statis
tics section.
The corrective phase is simply an upward retrace of the downtrend. It is
a place for the stock to catch its breath and for novice investors to buy into the
situation. They purchase the stock and push it up, believing the decline is at an
end. Do not be fooled; the decline is only half over. That is die beauty of this
formation. Before you buy a stock after a long decline, consider that it might
be making a measured move down and that the decline is not over. Paying
attention might save you some big bucks.
Returning to Figure 26.1, you can see the two legs following a trendline
that has nearly the same slope. This is not always the case, but a surprising
number of formations obey this dictum. Further, a channel—two parallel lines
that follow prices down—can encompass the two legs. Although the example
in Figure 26.1 is weak on die first leg, you can see how the second leg follows
a top trendline, connecting points D and E and extending down.
Lastly, the three points marked A, B, and C mark another measured move
down. This one is more compact and it is not uncommon to find nested for
mations like this. Sometimes, you get one measured move right after another.
Table 26.1 highlights identification characteristics for the measured move
down chart pattern. The first leg occurs as prices reach a new high and a trend
change begins. Prices decline leaving a price peak on the chart. From diere,
Table 26.1
Identification Characteristics of the Measured Move Down
Firstleg Usually begins from a new high. Prices decline rapidly in a
straightline fashion. Avoid declines that curve (they are founding
turns, scallops, or saucers).
Corrective phase Prices can move horizontally but usually rise and recover from
40% to 60% of the prior decline before resuming the downtrend.
If the corrective phase nears or rises above the first leg high, look
elsewhere.
Secondleg The dope of the first leg down trendline often carries onto this
leg. Both legs usually fit inside their own trend channels.
Avoid For cascading measured moves, use the first retrace and not later
ones as they get progressively closer to the end of the trend.
Avoid horizontal, sawtooth consolidation regions and measured
moves that rise from a flat base. Make sure the measure rule does
not predict prices will fall below zero.

prices continue moving lower, usually in a straightline run. Most times you
can draw two parallel lines, one connecting the minor highs and one joining
the minor lows, forming a downsloping trend channel. The corrective phase
stops the decline. Prices can move horizontally but usually retrace a significant
portion of their losses, say, between 40% and 60%.
When the second leg begins, the downturn resumes. Prices usually follow
the slope set by the first leg but this varies from formation to formation. Of
course, the two legs will not share the same trendline since the corrective phase
offsets them. Even so, the second leg usually fits inside its own trend channel
as prices decline in a straightline fashion. The second leg decline approxi
mates the price decline set by the first leg and the time it took to accomplish it.
There are several guidelines that you should follow when searching for
the measured move down. Avoid formations that show a rounded first leg,
where prices move lower but curve around in a sort of rounding turn, scallop,
or saucer. The trend should be a straightline decline.
During the corrective phase, prices should not rebound (or come close) to
the high set by the first leg. If prices near or rise above the first leg high, then
avoid the formation.
Watch for consecutive measured moves in a declining price trend. The
downtrend eventually will end, so it is best to trade on the first or second mea
sured move and avoid the rest.
Occasionally, the prevailing price trend will be horizontal. Prices rise up
and reach a high then begin down in the first leg. When prices return to the
base, they bounce. This bounce, wronglyinterpreted as the corrective phase of
a measured move down, is really a minor high in a consolidation trend. Prices
return to the base and may bounce several more times before beginning a sus
tained move upward. The overall picture looks like a horizontal sawtooth for
mation. Avoid measured moves that spring from a horizontal trend.
The last caveat is to consider the measure rule. I discuss the measure rule
in the Trading Tactics section of this chapter, but the measure rule says the
second leg will approximate the price move of the first leg. If the first leg has a
large decline, you may find that the predicted price is very close to zero or per
haps even negative. Obviously, the stock is not going to go negative and prob
ably will remain far above zero, so you might look elsewhere for a more
promising trade. Examples of these idiosyncrasies follow in the Focus on Fail
ures section.
Figure 26.2 shows two examples of the measured move down formation.
The first one, marked by points A, B, C, and D, begins after a long price rise. The
stock moves up from 43'/s in late November 1994, to 595
/s in earlyJuly. Then,
prices decline following a down trendline and stay within die trend channel
until midAugust, when they reach a low of 51 '/z. The corrective phase begins
on volume that is high but not unusually so. Prices move up and retrace 68%
of the decline before tumbling again. In the second leg, prices move below the
low (point B) and continue lower to point D. Then it is over.
Air Products and Chemicals Inc. (Chemical (Diversified), NYSE, APD)
|un95
Figure 26.2 Two measured moves. Notice how they fit neatly inside a trend
channel. With measured moves, the price decline from C to D nearly matches the
decline from A to B.
The second leg is steeper than the first leg and covers the ground in about
half the time (36 days versus 19 days). In addition, the second leg is slightly
shorter than the first one (the first leg declines by 8'/8, whereas the second
one falls 71
A).
Another measured move occurs in midNovember and ends at about the
same level as the first formation in late January (see points E, F, G, and FI). If
you look closely at Figure 26.2, you can see another measured move that forms
in the first leg from point E to point F. Points El and E2 mark the corrective
phase.
Focus on Failures
What constitutes a failure ofa measured move down? Early in the study of this
chart pattern, I decided that if prices do not dip below the first leg low, then
the formation classifies as a failure. Admittedly, this is subjective and it depends
to a large extent on the size of the corrective phase, but it does weed out the
weaker situations.
There are a number of identification mistakes that I want to point out.
Figure 26.3 shows the first one. The semiconductor maker's stock reaches a
high, along with a host of other chip makers' stocks, in the summer and fall of
1995. The stock forms a headandshoulders top in August and September

Integrated Device Technology (Semiconductor, NASDAQ, IDT1)
Figure 26.3 A headandshoulders top leading to a measured move down. The
headandshoulders top forms the basis for the large decline. The corrective phase
is small in comparison to the large first leg decline. The measure rule for the mea
sured move formation predicts prices will go negative. Think the stock will make it?
before burning out. In a near straightline run, the stock tumbles from a high
of33I
/2toalowof91
/s.
If this decline marks the first leg of a measured move down, how far will
prices fall in the second leg? That depends on how far up the corrective phase
brings prices. The corrective phase rises to a high of 15!
/4.1 discuss the measure
rule later, but it says the second leg approximates the price decline of the first
leg. Ifwe run through the computations, we discover that the predicted decline
is minus 91
/s. Even if the company were to declare bankruptcy, its stock price
would never go negative, there is no way that prices are going to decline that far.
Another key to this failure is the size of the corrective phase. Usually,
prices recover 40% to 60% of the first leg decline, but this one does not come
close (about a 25% retrace). With larger price declines, the corrective phase is
proportionally larger too. However, the formation in Figure 26.3 does not
show such behavior.
Figure 26.4 shows another situation: the flat base problem. Prices are
essentially flat from the start of February. By that I mean the minor lows all
share the same value—about 40. When prices move up in August and reach a
minor high in September, it is nothing unusual. Although the prior minor
highs do not ascend to this height, there is no reason to suspect that a measured
move will follow.
Anadarko Petroleum Corp. (Petroleum (Producing), NYSE, APC)
Apr 95 May Oct Dec
Figure 26.4 A measured move from a horizontal base. These formations rarely
work out as anticipated.
When prices decline to the base at 40 and bounce, a naive investor might
think a measured move down is forming. The corrective phase in late October
and early November sees prices rebound quite a ways up the first leg before
curling over and heading down. If the investor sells short at this point, it will
be a costly mistake. Prices quickly skyrocket to 60 by May from the second leg
low of 447
/s. Why does this formation fail? The strong support level at 40 cur
tails any meaningful decline below that point. In other words, there is nothing
to reverse.
Figure 26.5 shows the last failure, a case of mistaken identity. After a
long, extended rise, a retrace can be expected, maybe even a trend change that
takes prices drastically lower. When prices turn down in early January, a
decline is long overdue. Ifyou connect the minor lows in the uptrend, you dis
cover that prices pierce the up trendline in midDecember. This piercing sup
ports the theory that the trend is changing. It is not conclusive, but it does tilt
the scales in that direction. Throughout the month of December prices essen
tially move horizontally before perking up and making a new high just before
prices plummet. The failure to continue moving higher is another clue to a
trend change. The final clue is when prices descend and drop below the prior
minor high reached in late October. It even tumbles below the support level at
125
/8 before recovering.
The corrective phase of the measured move down sees prices recover
quite far, representing a retrace of 85%. That is well above normal and should

376 Measured Move Down Statistics 377
Champion Enterprises (Manuf. Houslng/Rec. Veh., NYSE, CHB)
Figure 26.5 This measured move down is really the corrective phase of a mea
sured move up.
Table 26.2
General Statistics for the Measured Move Down
from 1991 to 1996
Failure rate
Most likely decline
First leg price decline
Corrective phase price rise
Second leg price decline
Corrective phase percentage retrace
Percentage of formations starting near
12month price low (L), center (C),
or high (H)
Measure rule
622
134 or 22%
36%
25% to 40%
5 months (153 days)
25% in 54 days
16% in 39 days
27% in 60 days
40% to 60%
L0%, C6%, H93%
43% of second legs are longer than first legs
flag a potential problem. With such a large retrace, prices will probably decline
to at most the prior low and stall (because the second leg is slightly smaller than
the first one). It did not even go that far. Prices declined to the prior minor
high at 13'/2 and held steady for a week before moving higher.
Why did this formation fail? The choice of a measured move down for
this situation is poor because of the extent of the corrective phase. If we look
at the larger picture, we discover that the first down leg is nothing more than
the corrective phase of a measured move up!
Statistics
Table 26.2 shows the only statistics for this formation. I uncovered 622 for
mations in 2,500 years of daily price data. Such a large number of formations
means they are common, offering plenty of investment opportunities.
Of the formations I uncovered, the vast majority (88%) are reversals of
the upward trend; the remainder are consolidations. Consolidations typically
occur as part of a long downtrend. Sometimes the formations are small when
compared with the total down move, whereas others are part of two or three
cascading formations in the same bear run.
For measured moves, I define a failure to be when prices do not decline
below the low established in the first leg. I did multiple passes to be sure that I
located as many of the failures as I possibly could. This amounted to 22% of
the formations. The failure rate is quite close to the 20% maximum that I
deem acceptable for reliable formations.
The average decline amounts to 36%, as measured from the highest high
reached in the first leg, through the corrective phase, to the lowest low in the
second leg. Since you probably will not be able to identify a measured move
down until sometime in the second leg, your average decline is likely to be con
siderably less than 36% (probably about 20% since the corrective phase brings
prices up). The most likely decline is between 25% and 40%. Figure 26.6
shows the results of a frequency distribution of declines for all successful for
mations. Notice the center of the graph has four columns of about equal
height. If you ignore the catchall column on the right of the figure, the four
center columns have the highest frequency. As such, I consider them to be rep
resentative of what you can expect to earn from this formation. Again, let me
remind you that the decline measures from the highest high in the first leg to
the lowest low in the second. Once you identify a measured move down, a sig
nificant portion of the decline may already have occurred.
The average formation length is 153 days (about 5 months). The first leg
averages a price decline of25% in 54 days; the corrective phase posts an aver
age gain of 16% in 39 days, and the second leg declines by 27% in 60 days. You
can see in Table 26.2 that the two legs are nearly the same in percentage
decline and time.
How much does the corrective phase recover? On average, about 40% to
60% of the first leg decline. Figure 26.7 shows a frequency distribution of the

30 35 40
Percentage Decline
Figure 26.6 Frequency distribution of declines for the measured move down.
Ignoring the right column, the most likely decline is between 25% and 40%.
45 50 55
Percentage Retrace
Figure 26.7 Frequency distribution of the percentage price retrace. The correc
tive phase typically retraces between 40% and 60% of the prior decline.
Trading Tactics 379
percentage price retrace shown by the corrective phase when compared with
the first leg decline. The highest column, the one on the right, is excluded
because it is a catchall column for large percentage retraces. Ignoring the
right column, the 45% column is tallest but is closely joined by the columns
from 40% to 60%. Thus, I conclude that the corrective phase retrace is most
likely to fall within the 40% to 60% range.
Due to the nature of the measured move down, the chart pattern usually
forms near the yearly high. A frequency distribution of the formation start
(the highest high in the first leg) when compared with the prior yearly price
range shows that 93% ofthe formations begin in the highest third of the yearly
price range. Many of the other formations are ones that occur as prices slide
down a trendline and create several measured moves down in a row.
I counted the number of formations meeting or exceeding the measure
rule and discovered that only 43 % of the formations have second legs that are
equal to or longer than the first legs. I consider values above 80% to be reliable.
This means you should not depend on the target being met. The 80% bench
mark hits when the second leg down move is equal to twothirds ofthe first leg.
Trading Tactics
Table 26.3 outlines trading tactics and the measure rule. Use the measure rule
to help predict how far prices will decline. Refer to Figure 26.8 during the dis
cussion of its computation. In the figure, four points outline the measured
move: A through D. First, tabulate the height of the first leg (shown by points
A and B) by subtracting the lowest low (42) from the highest high (527
/g). This
gives a difference of 107
/8. Subtract the difference from the highest high in the
corrective phase (point C at 473
/4). The result is a target price of 367
/s. Prices
meet the target on December 7.
Once you suspect that a measured move down is forming, probably just
after the corrective phase completes and prices start down, short the stock. Use
the measure rule to predict the price target, but expect prices to come up short.
Place a stoploss order l
/s above the corrective phase high (the corrective phase
is a source of support and resistance). For a more conservative target, follow
the measure rule using twothirds of the first leg height. Prices reach the new
target 80% of the time. For example, twothirds of the first leg height in Fig
ure 26.8 is 7!
/4. Subtracting this value from the corrective phase high gives a
closer target of 40'/2. Cover your short if prices rebound off a support zone or
approach the measure rule target.
After a measured move down completes, the corrective phase often spells
a resistance zone for future moves. Prices pause on the approach to the
378

Table 26.3
Trading Tactics for the Measured Move Down
Measure rule
Short during second
leg, stop loss
Close out
Support/resistance
The second leg is about 10% shorter than the first leg, so expect
the actual price to fall short of the target. Compute the length of
the first leg from the highest high to the lowest low (at the start
of the corrective phase). Subtract the result from the highest
high reached in the corrective phase to get the target price
(which is met 43% of the time). For a more conservative target,
use twothirds of the first leg height. This shortened height
means that prices hit the target 80% of the time.
Short the stock as soon as it becomes clear that a measured
move is in progress. If prices rise above the corrective phase
high, then close out your position. Prices occasionally will rise up
to the corrective phase high a second time before ultimately
declining, so put your stop about % (or more) above the high.
Cover your short when the price drops to a support area and
meets resistance to a further decline, especially if prices near the
measure rule target.
The corrective phase shows future support or resistance.
corrective phase low and at the high. Figure 26.8 has these zones labeled. If
you are nimble, you can anticipate this rise and trade long once the measured
move down completes. Sell if prices run into trouble during the corrective
phase and begin heading lower. In a bull market you can generally expect
prices to eventually push through the corrective phase resistance and move up
to the old high.
Sample Trade
People are nasty; just ask Eddy. He is an airline reservation agent. Between the
company monitoring his phone calls to be sure he peddles a car and hotel
when appropriate and the people screaming at him from the other end of the
phone, it is a tough living. There is nothing he can do about equipment prob
lems or the weather, but people do not seem to care. Even the full moon gets
into the act as that is when the crazies call.
What he would really like to do is invest in the stock market. He does it
now but to a limited extent because ofhis cashflow problem. Fortunately, with
a few clicks of his computer mouse he can flip to the Internet and monitor his
latest stock pick when he is not busy.
That is how he uncovered the situation shown in Figure 26.8. He watched
the stock climb from a low of about 10 in June 1994 to a high of 535
/g in Octo
Sample Trade 381
Oct95
3 Com Corp. (Computers & Peripherals, NASDAQ, COMS)
Figure 26.8 Measured move down followed by corrective phase. Eddy made $5
per share trading this (points AD) measured move down.
ber 1995. Every so often, he would draw trendlines along the bottoms of the
minor lows and notice how the upward trend seemed to be accelerating (the
trendlines grew steeper over time).
Since he knew this could not last, he was ready for a trend change, which
occurred on November 14 when prices pierced the trendline, moving down.
Instead of shorting the stock immediately, he decided to wait for a pull
back. Much to his surprise it never came. Prices moved steadily lower until
they reached a support level at 42. From that point on, prices rose higher
for the next week and a half. Then, they dropped sharply, tumbling $3 in
one session.
When prices fell, they pierced a small up trendline, drawn along the bot
toms of the climb from points B to C. Eddy recognized what was happening
when he drew the trendline on his chart. The chart pattern was making a mea
sured move down, so he shorted the stock that day and received a fill at 42.
He used the measure rule to compute the predicted price move and
placed an order with his broker to cover the short at the predicted price (367
/s).
Just 3 days after he placed the trade, the stock was covered. He made about $5
a share or 12%.
Ifyou look at Figure 26.8, you can see that prices rose to the level of the
corrective phase bottom (point B), then retreated 2 days after Eddy completed
his trade. Later on you can see that prices also stopped rising at the top of the
corrective phase. The corrective phase is a zone of support and resistance.

27
Measured Move Up
Tour 383
Appearance
Failure rate
Average rise
First leg rise
Corrective phase decline
Second leg decline
Percentage meeting
See also
Prices move up, retrace, then move up again.
23%
43% in 87 days
14% in 45 days
37% in 65 days
57%
Flags and Pennants
The measured moved up formation is die reverse of the measured move down.
The measured move up sports a 23 % failure rate, slightly higher than the 20%
maximum I consider reliable formations to possess. However, the average gain
is an astounding 68%, which is misleading. The value represents the price dif
ference between the lowest low in the first leg and the highest high in the sec
ond leg. Obviously, this is a bestcase scenario and you can expect your results
to be a little better than halfthat rate. Why? The reason stems from being able
to identify the formation promptly. When the chart pattern enters the correc
382
tive phase, where prices retrace their gains, only then can one suspect a mea
sured move is forming. Only after prices start moving up again, at the start of
the second up leg, should an investment be placed.
By the Results Snapshot statistics you can see that the second leg averages
a 37% gain. Since an investor probably would not take a position in the stock
until after the second leg begins, the gain will probably be about 30% or so.
Tour
In the preceding paragraphs I mentioned a number of terms. Figure 27.1 out
lines a measured move up formation with the various components labeled. The
first leg is composed of a rise in price that follows a trendline. Many times
a trendline drawn on either side of the minor highs and lows constructs
a channel.
The corrective phase retraces a substantial portion of the rise, usually
35% to 65%, before prices resume rising. In Figure 27.1, the corrective phase
begins in late January and extends through most of February. Prices retrace
55% of the first leg price move. Once the correction completes, prices climb
even more rapidly during the second leg. You can see that prices bow upward
instead of touching the trendline in a sort of roundingover maneuver before
Allegheny Ludlum Corp. (Steel (General), NYSE, ALS)
Nov 92 Dec |an 93 Feb Mar Apr May jun
Figure 27.1 A measured move up. The second leg gain nearly matches the gain
posted by the first leg.

384 Measured Move Up
topping out inJune. The rise constitutes what is commonly called the second
leg. It is the rise from the corrective phase to the end of the formation.
Once the formation completes, prices sometimes drop back to the level of
the corrective phase. In this example, you can see that prices dropped to just
below the top of the corrective phase (in July) before recovering. Of course,
sometimes prices do not stop there at all. A trend change occurs and prices
simply tumble and return to the base of the formation or, worse, continue
moving down.
Table 27.1 lists the identification characteristics for this formation. The for
mation usually, but not always, begins when prices bottom out after a down
trend. The declining price trend can range all over the scale. In Figure 27.1, for
example, prices reach the November lows after shooting up in a bull run that
starts in December 1991 and ends 2 months later. Then prices meander—
essentially moving horizontally with a slight downward bend. Six months later,
they decline from a high of about 18'/4 to a low of 143
/4.
Consider Figure 27.2 where the price decline is short—barely a month
long. Over the longer term (not shown in the figure), prices are rising. They
Table 27.1
Identification Characteristics of the Measured Move Up
Downward trend A downward price trend that lasts from a few weeks to over a
year usually precedes the start of the formation. The formation
begins a trend change that usually starts from near the yearly
low.
First leg Most times prices follow a trend channel upward before entering
the corrective phase.
Corrective phase Prices decline, usually between 35% to 65% of the first leg move,
before heading upward again. The retrace is usually proportional
to the first leg rise: Large retraces follow large rises. Sometimes
the corrective phase resembles a sawtooth pattern (a few sharp
rises and declines in a row) before prices break away and zoom
upward. This sawtooth pattern usually associates with a long
price climb leading to the formation.
Second leg Prices rise, loosely following the slope of the trendline set by the
first leg. Prices commonly fit inside a channel as they rise, but this
is not a prerequisite.
Avoid Avoid formations where the retrace travels too far down the first
leg. Anything beyond an 80% retrace is probably too far and too
risky to invest in.
Applied Power (Machinery (const./mining), NYSE, APW)
May 95 |un |ul Aug Sep Oct Nov Dec
Figure 27.2 A falling wedge marks the corrective phase in this measured move
up formation. Note the receding volume trend of the wedge.
begin climbing in November 1993 at a price of 14'/2 and reach a preformation
high of 27 in May 1995—a near double. The figure shows that the first leg has
a slight bow to it in the early part of the rise. However, you can extend the up
trendline and draw a parallel one connecting the minor highs and see that the
first leg fits inside a trend channel. The second leg does even better. The bot
tom trendline touches several places and a parallel top trendline (not shown for
clarity) also intersects the minor highs nicely.
In this case, a falling wedge composes the corrective phase. This forma
tion makes trading the measured move easy since it predicts a price rise. Once
prices break out from the wedge, buy into the stock and ride the upward move.
If you bought the stock following this procedure, you would make somewhere
between 15% and 20%, depending on when you traded the stock. That is not
a bad return for a hold time of about 6 weeks. Also note the very distinctive
downsloping volume trend for the falling wedge.
The slope of the two trendlines, along the bottoms of the legs, are nearly
the same. It is somewhat surprising how often this holds true. However, just
because there is a wide variation in the trend slope is no reason to eliminate a
formation from consideration.
In this example, the corrective phase sees prices retrace their prior gains
by 40%, within the usual 35% to 65% range for measured move up forma
tions. Sometimes when the rise leading to the start of the formation is exten
sive, the corrective phase becomes long and choppy, resembling a sawtooth
formation marked by quick rises and sharp declines. In such a case, it might be

prudent to wait for prices to rise above the high established during the first leg
before investing. That way you can avoid the most common measured move up
failure.
Focus on Failures
What exactly is a measured move up failure? I define a failure of a measured
move up chart pattern to be when prices do not rise above the prior leg high.
The definition is a subjective measure. With deep corrective phases, even a rise
to the old high can be a substantial move. Still, I feel the benchmark is a good
one, so that is the one used in the statistics.
Figure 27.3 shows the most common type of measured move up failure.
The stock forms a double top that kills the second leg rise. The failure is clear
since prices rise to 37s
/s, just I
/B below the prior high before heading lower. Cer
tainly the second leg does not near the price move of the first leg as do most
wellbehaved measured moves. Why does this particular formation fail? The
figure shows a choppy, horizontal sawtooth pattern leading to the first leg
rise. The first leg soars above the two tops of the sawtooth and moves up
smartly. Then prices round over and start correcting. The figure, at this point,
reminds me of a mini bumpandran reversal. However, the bump phase just
does not meet the twotoone height ratio of die leadin phase. Still, it does
give you pause about investing in this situation.
Jan 93 Feb Mar Apr May Jun Jul
Figure 27.3 A measured move up that fails after turning into a double top.
Statistics 387
The second leg starts rising with no significant change in volume. This is
a warning sign. There is a common Street axiom that says rising prices need
high volume but falling prices can decline oftheir own weight. This formation
appears to be an example of that axiom. Since there is little upward buying
pressure to push prices higher, they fade out just below the prior top then tum
ble. The resulting decline sees prices fall below the start of the first leg.
Before we move on to statistics, I want to alert you to some identification
problems. You want to avoid formations that have corrective phases that
descend too near the first leg start. I do not have a set amount for this, but I
would probably steer away from formations that retrace more than 80% or so
of the prior upward move.
In addition, if the first leg does not follow a straight course upward or if
it fails to stay within a welldefined trend channel (as does the first leg in Fig
ure 27.3), you might want to look elsewhere for a more promising situation.
Sometimes when a chart formation does notfeel right or look right, then it is
giving you a warning to stay away. Since this is a common chart pattern, you
can easily find another opportunity.
When prices rise steadily for a long time, say over a year or more, then
begin a measured move up, the corrective phase might be excessively choppy.
I mentioned this behavior in the Identification Guidelines section and in Table
27.1, but it is something to keep in mind.
Also, do not be too quick to buy into the situation. Remember that the
corrective phase should be proportional to the first leg rise. By that I mean
prices should fall anywhere from 35% to 65% of the first leg move before
beginning the second leg. If prices only fall 15 % before turning up, then it
might be a false breakout.
Sometimes prior peaks are a key to how far prices retrace. These minor
highs are often places of support. When prices decline to that level, they pause
and move horizontally for a time before continuing down or rebounding. Vol
ume is often a key to the level of support you can expect from these types of sit
uations. A prior peak with high turnover will give more support to a stock on
its way down. That is not to say that the stock will not burn through the sup
port, just that it might take more of a push to fall off the cliff.
Statistics
Table 27.2 shows comparatively few statistics for measured move up forma
tions. I located 501 formations and ofthese formations, 69% act as reversals of
the prevailing trend. This means prices are heading steadily down before the
start of the first up leg. The remaining 156 formations occur as part of the
upward trend.
The failure rate at 23% is higher than the 20% maximum I like to see.
The implication is that about one out of four formations fail to rise above the

Table 27.2
General Statistics for the Measured Move Up
from 1991 to 1996
Failure rate
Most likely rise
First leg price rise
Corrective phase price decline
Second leg price rise
Corrective phase percentage retrace
Percentage of formations occuring
near 12month price low (L), center
(C), or high (H)?
Measure rule
501
114 or 23%
68%
30% to 60%
43% in 87 days
14% in 45 days
37% in 65 days
35% to 65%
L79%, C21%,
57% of second legs are longer than first legs
first leg high before trending down. As mentioned in the Focus on Failures sec
tion of this chapter, many failures occur while prices are advancing and form
ing a second top. They stall out and turn lower, leaving a double top formation.
For those formations that work as expected, the average gain is 68%. As
explained earlier, this may sound like an outsized gain and it is. Since you
probably will not be able to (or want to, for that matter!) buy into a measured
move before the end of the corrective phase, the second leg (37% gain) better
represents the performance of the formation.
Figure 27.4 shows the results of a frequency distribution of gains for suc
cessful formations in this study. I define the most likely gain as being repre
sented by the highest columns. Since the column on the right is a catchall
column, it is excluded from the analysis. The remainder of the chart indicates
that the most likely gain is between 30% and 60%, with a tendency to gravi
tate toward the upper end of the range. Since over a quarter of the formations
have gains over 100%, you might find a situation in which your gains are well
above the most likely gain. Remember, I compute the gains from the lowest
low in the first leg to the highest high in the second leg. If you invest near the
start of the second leg, you should reduce your expectations accordingly.
For a better gauge of the expected gains, consider the performance of the
various components. The first leg has a 43 % average gain in about 3 months.
The corrective phase brings prices down by 14% in 45 days, whereas the last
leg sends prices higher by 37% in about 2 months. Your likely gain, depend
Statistics 389
60 70
Percentage Gain
100 >100
Figure 27.4 Frequency distribution of gains for the measured move up. Exclud
ing the tallest column, the most likely rise is between 30% and 60%.
ing on when you take a position in the stock, will probably be near the second
leg gain.
I took a close look at the corrective phase retrace by doing a frequency
distribution of the percentage retrace for all successful formations. I graphed
the results using a 1% interval ranging from 20 to 80.1 wanted to see if there
was any truth to the beliefthat the retrace would be one ofthe Fibonacci ratios
on a percentage basis. The Fibonacci ratio is the ratio between any two suc
cessive numbers in a Fibonacci sequence. The Fibonacci sequence begins with
die values ofzero and one and successive numbers are the sum ofthe prior two
(as in 0, 1, 2, 3, 5, 8, 13 and so on). Once the sequence develops, two impor
tant ratios between pairs and the inverse of alternate pairs become clear: 0.618
and 0.382. The theory says that, on a percentage basis, support at the 62% and
38% retracement levels is common.
If there is any truth to die average retrace being a member ofthese values,
then a frequency distribution would illustrate it. I do not show the graph
because there are too many columns to be clearly presented, but the results are
easily described. The overall graph resembles a bell curve with noticeable
peaks. The first peak is at 35% and it towers above the surrounding ones. The
next few peaks, which are the tallest on die graph, fall at 47%, 50%, and a
somewhat smaller one at 52%. As we advance up the scale, we find anodier
cluster in die 62% to 64% range, widi 62% being die highest. Other peaks
toward die outer ends of die graph are at 28% and 79%. If we boil down die

results, we find that the most common retraces during the corrective phase
occur at 35%, 50%, and 62%, close to the predicted values of 38% and 62%.
A frequency distribution of the start of each successful formation in the
yearly price range shows that most of the formations begin life within a third
oftheir yearly low. Only 21 % ofthe formations begin in the center third ofthe
yearly price range. The significance ofthis should be obvious but I state it here
anyway: Ifyou think you have a measured move up occurring within onethird
of the yearly high, then you are probably wrong. Avoid those measured moves
that do not occur near the yearly low.
Slightly over half (57%) of the formations fulfill the measure rule. The
measure rule conveniently leads us into the Trading Tactics section.
Trading Tactics
Table 27.3 shows trading tactics for measured move up formations. The mea
sure rule predicts the level to which prices will rise. To estimate the target
price, compute the height of the first leg. I use the measured move up forma
tion shown in Figure 27.5 as an example. Locate the highest high in the first
leg. Usually this is somewhere near the beginning of the corrective phase, and
point A indicates this in the figure. From this value (21H), subtract the lowest
low (H'/s) in the first leg, shown as point B. The difference (7) is the height of
the first leg. Add it to point C (183
/s)—or the lowest low in the corrective
phase—to arrive at the target price. In this case, the target price is 253
/8. Prices
reach the target just 10 trading days after the corrective phase ends.
On a. price basis, the second leg is about 10% longer than the first leg, on
average, even though it represents a 37% price change versus 43 % for the first
leg. (This anomaly is due to rising prices. If the percentages are the same, the
price move will not be.) However, just 57% of the formations have second legs
Trading Tactic
Table 27.3
Trading Tactics for the Measured Move Up
Explanation
Measure rule Calculate the height of the first leg from highest high to lowest
low. Add the difference to the lowest low in the correct phase.
The result is the expected target price. Decide if the predicted
move is worth the risk of a trade.
Yearly low Choose formations that start (have their lowest low during the
first leg) in the lowest third of the yearly price range. Avoid
those forming near the yearly high.
Buy Take a position in the stock sometime after the corrective phase
completes and prices rise during the second leg.
Support/resistance The corrective phase shows future support or resistance.
Trading Tactics 391
Pacific Scientific (Precision Instrument, NY5E, PSX)
Jun95 ]ul Aug Sep Oct Nov Dec |an 96 Feb Mar Apr
Figure 27.5 Measured move up formation with symmetrical triangle. Michelle
rode this measured move up in a stock she owned. She sold when the breakaway
gap closed. A symmetrical triangle shows a typical volume trend.
that are equal to or longer than the first legs. In other words, slightly over half
the successful formations fulfill the measure rule. I consider values above 80%
to be reliable, so the measured move up formation falls well short of the mark.
Once you calculate the target price using the measure rule, ask yourself if
the gain is large enough to justify a trade. If the answer is yes, then look at the
chart again. Are there areas of resistance on the way to the target price where
the stock might get hung up? If so, you might need to lower your target. Ifyou
are lucky and significant resistance is above your target, you can move your
price upward to just below the resistance zone. In all likelihood the stock will
shoot into the resistance zone, so you will have ample opportunity to close out
your position.
When selecting a formation to trade, it is best to choose those that start
near the yearly low. In my analysis of these chart patterns, only two appear
within the upper third of the yearly price range. The vast majority (79%) are
near the yearly low. After prices leave the corrective phase, then buy the stock.
To gauge the breakout point, draw a downsloping trendline along the minor
highs in the corrective phase. Once prices dose above the trendline, then buy
the stock. As the stock approaches the target price, do not be too quick to sell.
If prices are on a roll, go with the flow and wait for prices to start declining.
Obviously, ifprices pause near but below the target price, then it might be wise
to sell.

Once prices begin moving down, they sometimes return all the way down
to the corrective phase before meeting any meaningful support. They may
pause at the top of the corrective phase or rebound at the bottom of it. Some
times, prices just sail right on through. Whatever the case, be aware that ifyou
do not sell near the target price and decide to hold on, you might lose all your
gains.
Sample Trade
Michelle is an engineer. Over the years, she has developed a thick skin to take
the ribbing from her college colleagues in a maledominated profession. Even
when she ventured into the professional environment after college, the ribbing
continued. Make no mistake: She is pretty and they just wanted her attention.
I saw this firsthand when I stopped by her office with a question. She was not
there at the time, but her desk blotter had the scribblings of love notes from
dozens of men. Of course, I added my own. But I digress.
If you were to give a Rubic's cube to Michelle, she would not necessarily
solve it. First, she would want to take it apart to see how it is constructed. This
inquisitiveness coupled with her ability to solve tough problems in a unique
way makes her special even among engineers. She is also an investor with the
same qualities.
Michelle had a unique way to take advantage of the situation shown in
Figure 27.5. She already owned the stock but believed it was running out of
steam. During the prior November to February period (not shown), she saw
the stock form a double top. Prices declined from a high of 24'/s to the low of
14'/s at the start of the measured move.
Unfortunately, as a novice investor, she was unable to pull the trigger and
sell it after prices confirmed the double top. She rode the decline down to the
low and saw 41 % ofher gains evaporate. When the stock began moving up, she
breathed easier. Still, she vowed to do better the next time. As the stock started
its climb, she saw the increase in volume. The increase meant that the run
would be an extended one, as there seemed to be enough enthusiasm to send
prices higher.
All bull runs must pause now and again and this situation was no differ
ent. Michelle saw the stock pause and consolidate for nearly a month during
July. She looked back at the chart and noticed that the stock had reached a zone
of resistance where there were several old highs that stalled prices near the 21
level. Volume picked up and when prices shot upward, she immediately recog
nized the measured move formation. Did she sell? No. She hung on for the
ride. Michelle calculated that the stock would rise to 253
/8, a new yearly high.
She suspected that the stock might find resistance at the old highs of 24, and
that is exactly what happened. A symmetrical triangle formed in the stock. The
formation obeyed the rules for symmetrical triangles, lower highs and higher
Sample Trade 393
lows with a receding volume pattern, and she was confident that she had cor
recdy identified it.
Since there was no way to tell which direction the stock would break out
of the triangle, she sat tight. Then prices gapped out the top. Was the gap a
breakaway gap or an exhaustion gap? She reasoned that since the gap appeared
just after a region ofconsolidation, it was most likely a breakaway gap, so prices
would continue rising, but how far? She hoped the triangle represented die
halfway mark of an up move. She knew that symmetrical triangle formations
sometimes act like halfmast formations, so she expected a climb to 28 (see the
measure rule for Symmetrical Triangles). To her it sounded like a long shot,
but one worth waiting for. Her calculated price target of 253
/s was met the day
prices jumped out of the triangle.
About a week after prices left the triangle, they reached a new high then
fell back. When prices closed the gap in the first part of September, she
decided to sell her holdings. Fortunately, the next day prices zoomed upward
and she was able to sell her shares at 24'/2, near die daily high of 251
/8. As she
watched, the stock tumbled back to the middle ofdie corrective phase, right in
die center ofthe support zone. Then, the stock recovered. As die stock climbed
and posted a new high, she wondered if she had sold too soon. She felt better
after reviewing the chart 6 months later and seeing prices hovering in die $ 15
range.

28
OneDay Reversals
RESULTS SNAPSHOT
Tops
Appearance
Failure rate
Average decline
Volume trend
Fullbacks
Synonyns
Bottoms
Appearance
Failure rate
Average rise
Volume trend
Throwbacks
Synonyms
394
A large 1day upward price spike with prices closing
near the low
24%
Heavy volume on reversal day but recedes quicldy
71%
Climax Day, Selling Climax
A large 1day downward price spike with prices
closing near the high
17%
Heavy volume on reversal day but recedes quickly
61%
Climax Day, Selling Climax
* Tour 395
Oneday reversals (ODRs): The only surprise in the Results Snapshots is that
the formations perform so well. Sure, the 24% failure rate for tops is above the
20% maximum I arbitrarily assign to reliable formations, but it could be worse.
The average loss for tops at 19% is quite close to the average of 20% for all
bearish formations. However, bottom reversals, with gains averaging 26%,
are well short of the usual 40% for bullish patterns. The most likely rise is
10% to 15 %, about what you would expect. I suspect that by the time you place
a trade after this formation, it will be too late and your return will be un
remarkable.
Fullbacks are prevalent (happening in 71% of the ODR tops) and suggest
you might wait for them before investing in an ODR top. Throwbacks are a bit
weak, however, so do not depend on seeing them in ODR bottoms.
Tour
The ODR top shown in Figure 28.1 marks the crest of an unbalanced head
andshoulders top. It appears after a long climb that begins in January. The
ODR has a closing price near the daily low and the spike itself is longer than
many of the spikes over the preceding year.
Figure 28.2 shows what the bottom version of the ODR looks like. It
appears after a downtrend that in this case is not very long. The downtrend
Figure 28.1 A oneday reversal top appears as a large spike with a closing price
near the daily low. It marks the head of an unbalanced headandshoulders top.

396 OneDay Reversals
Thomas Industries Inc. (Building Materials, NYSE, Til)
Aug 93 Sep Oct Nov Dec Jan 94 Feb Mar
Figure 28.2 A oneday reversal bottom appears after a downtrend. Note the
large downward price spike with a closing price near the daily high.
begins just 2 months before the reversal. The ODR bottom has a closing price
near the daily high and accompanies high volume, and the spike is longer than
other spikes throughout the prior year.
Why do ODRs form? When I see a large price spike, I think ofone thing:
stop running. Imagine a series of stoploss orders placed consecutively below
the current price. It does not take much of a downward move to trigger the
first stop. If the number of shares being sold is large enough, coupled with the
assistance ofshort sellers, traders, and market makers, the price may drop even
further. When that happens, the next stop triggers and so on. The price tum
bles until enough buying demand comes online to stop the cascade. This stop
running (or gunning the stops) is intentional.
Once it is over all the people wanting to sell their shares have already sold,
so there is overwhelming buying demand. The price recovers to near where it
started the day. At the close, a long thin line appears on the bar chart. It is an
ODR, where prices start the day about where they end, but in the interim,
prices plummet several points. The selling pressure abates leaving only buying
demand to carry prices higher over the coming days.
The same scenario applies to ODR tops. Prices quickly rise but finish the
day near the lows. Afterward, overwhelming selling pressure takes prices lower.
How do you identify ODR tops and bottoms? Table 28.1 highlights some
characteristics to assist you in your selection process. First, start with a price
trend. In many cases, the ODR occurs at the end of the trend, so the trend is
usually a long drawn out affair (over several months). Some ODRs occur after
prices move for only a few weeks, so it varies from formation to formation.
For ODR tops look for a rising price trend; for ODR bottoms, the trend
should be declining. Do not make the mistake of picking an ODR top in a
declining price trend or an ODR bottom in a rising price trend. Although
these spikes might act as consolidations, it is best to limit your selections to
reversals and depend on a trend change.
Figures 28.1 and 28.2 show an ODR top and an ODR bottom. In Figure
28.1 you can see part ofthe upward trend that begins at a price of $5.70 nearly
a year before the reversal. Figure 28.2 shows a much shorter and sharper
decline. A zoomout of the figure shows an upward trend that begins in
November 1992 and peaks in August. Then, the stock moves horizontally and
tumbles for 2 weeks down to the ODR.
Both figures show large price spikes that seem to poke well beyond their
peers. The daily range exceeds anything on the charts up to that date. The
spikes seem to stand alone with good visibility to the left.
Another key identification guideline is where prices close. For tops the
closing price should be at or near the daily low. Bottom reversals show just die
opposite: The closing price should be near the daily high. The closing price
near the end of the range suggests prices are likely to continue in the new
direction. For tops the direction is down and for bottoms it is up. Although die
next day's trading range may cover some of the ODR, prices commonly close,
and the range moves, in the predicted direction.
Table 28.1
Identification Characteristics of OneDay Reversals
Pricetrend For ODR tops, look for a price uptrend over several weeks or
months. For ODR bottoms, the price trend should be down.
Largespike Both tops and bottoms sport tall price spikes. The spikes should be
twice as large, or more, as the average spike over the prior year.
Close near end For tops the closing price should be near the low of the day, and for
bottoms it should be near the high. By near, I mean within one
third of the daily range.
Volume Volume is usually higher on the reversal day than the prior day and
can be unusually strong.

The last guideline is volume. Usually, an ODR, either a top or bottom,
shows high volume, certainly volume that is above the prior day. This is clear
in Figure 28.2, where the high volume lasts for several days as prices climb, but
is less outstanding in Figure 28.1. Volume seems to be higher on bottoms than
on tops, supporting the belief that rising prices need a push while declining
prices can fall of their own weight.
Focus on Failures
Unfortunately, failures occur too often with these formations. Failures such as
those shown in Figures 28.3 and 28.4 are typical. Figure 28.3 shows an ODR
top failure. The price trend is up, the ODR spike towers above the surround
ing prices, the closing price is near the daily low, and there is high volume—
all the ingredients needed for a successful ODR. It works. Two days later,
prices drop from an ODR high of 39 to a low of 35!
/2.
If investors were to try to take advantage of this situation, they may not
recognize an ODR until the next day or even later. Therefore, I compute per
formance using the lowest price in the ODR to the ultimate low. In this case,
the ODR low is 37'/8, rendering a price decline of 4%. I consider declines of
less than 5% as failures (the socalled 5% failure).
Figure 28.4 shows a similar situation with an ODR bottom. Here, die
downward spike is obvious. Volume attending the spike is high and prices close
Circus Circus Enterprises, Inc. (Hotel/Gaming, NYSE, CIR)
|ar>96 Feb Mar Apr
Figure 28.3 A 5% failure of a oneday reversal top. Prices drop to a support level
then recover.
Statistics 399
American Brands, Inc. (Tobacco, NYSE, AMB)
Jul 92 Aug
Figure 28.4 A oneday reversal bottom fails after prices rise by less than 5%.
near the high for the day. The situation suggests prices will move higher and
they do for nearly 2 months, then they begin moving lower again. From the
daily high at the ODR to point A, the ultimate high for this formation, prices
climb by less than 4%. This performance qualifies as a 5% failure.
The reason for failure in each case is similar. Can you guess what it is?
They both stumble across support or resistance barriers that they cannot pen
etrate. Figure 28.3 has significant support at the 351
/2 level, whereas Figure 28.4
has resistance in the 43 to 44 range. The movement stalls at the zones and
reverses, setting up for the eventual failure. The lesson to be learned from
these two failures is that you should always look to see how far you expect the
move to go. Consider the areas of support and resistance before making a
trade.
Statistics
Table 28.2 shows statistics for ODR tops and bottoms. ODRs are not as plen
tiful as some other formations, but that is not to say they are scarce. On the
contrary, I uncovered 235 tops and 331 bottoms in the stocks I examined.
Almost a quarter (24%) of the ODR tops and 17% of the ODR bottoms
fail. Prices continue moving in the same direction after the formation as before
(in other words, 5 % failures, where prices briefly move in the expected direc
tion then reverse). When they do work, ODR tops have average declines of
19%, whereas ODR bottoms rise by 26%.

400 OneDay Reversals Statistics 401
Table 28.2
General Statistics for OneDay Reversals
Description Tops Bottoms
Failure rate
Average decline/rise of
successfulformations
Most likely decline/rise
235
56 or 24%
19%
10%
331
5 7 or 17%
26%
1 0% to 1 5%
Percentage of pullbacks/
throwbacks
Average time to pullback/
throwback completion
Days to ultimate low/high
Percentage of ODRs occurring
near 12month price low (L),
Percentage gain for each
Volume for ODR day and next
5 days versus day before ODR
Number of ODR days with
volume above prior day
Number of days with twice
prior day's volume
71% or 165
13 days
2 months (66 days)
16%,C17%,H77%
L20%,C22%, HI 7%
182%, 110%, 77%,
62%,69%,72%
170 or 72%
9 3 or 40%
61 % or 202
16 days
3 months (96 days)
L.35%, C34%, H31%
L30%, C23%, H28%
214%, 107%, 86%,
75%, 74%, 74%
285 or 86%
176 or 53%
Figure 28.5 illustrates the results of a frequency distribution of losses or
gains for ODR tops and bottoms, respectively. By far, the tallest columns are
those situated near the 10% and 15% categories. These categories signify the
most likely returns for ODRs. Almost half(49%) ofthe tops decline by 15% or
less, an alarming result if you short a stock expecting a long decline. Bottoms
perform marginally better, but that is scant comfort when the study period
encompasses a raging bull market.
Pullbacks at 71% and throwbacks at61% are reassuringlyhigh (especially
pullbacks). However, they take somewhat longer than usual to complete at 13
and 16 days, respectively (normal completion time is about 10 to 12 days). If
you are considering shorting a stock showing an ODR top, you would proba
bly be wise to wait for a throwback before investing. After the throwback, wait
for prices to head down again before selling short.
The time it takes to reach the ultimate low or high is brief, at 66 and 96
days for ODR tops and bottoms, respectively. This coincides with the small
average losses and gains for the two formation types. Small returns take a cor
respondingly shorter time to reach their price targets than do larger gains.
15 20 25 30 35 40
Percentage Cains or Losses
• Tops Q Bottoms
45 50 >50
Figure 28.5 Frequency distribution of returns for oneday reversals. The chart
suggests that the most likely return is less than 15%.
Ifwe divide the prior year's price range into three parts, we discover that
ODR tops predominantly form within a third of the yearly high. This is no
surprise, really, because the chart pattern forms after an extended up move.
ODR bottoms form near the yearly low but not decisively so. You can see
that the range splits evenly around 30% and change. This surprised me. I
expected bottoms to appear near the yearly low, not spread evenly across the
range. The results suggest that ODR bottoms happen after a short price
retrace, not an extended downtrend.
Ifwe substitute the performance into the yearly range, we discover that the
best performing tops occur in the center of the yearly price range (with declines
averaging 22%), whereas bottoms outperform near the low (a 30% rise). The
results are close enough to each other that nothing meaningful results, but you
can use the information to select better ODR candidates. Choose ODR bot
toms near the yearly low and ODR tops in the center third of the yearly price
range for best average performance.
Table 28.2 shows volume numbers for the two formation types. Both
show heavy volume on the reversal day that quickly diminishes. Since this is an
average, I counted die number of ODR days with volume above the prior day.
For ODR tops 72% are higher, whereas 86% of ODR bottoms have higher
volume. If you set the benchmark to twice the prior day's volume, the scores
drop to 40% for tops and 53% for bottoms, meaning that high volume usually
accompanies ODRs—well, higher than the day before the ODR at any rate.

Trading Tactics
I do not recommend trading ODRs. Although their average return is accept
able, the most likely return is just too small at less than 15%. I consider these
values just too skimpy to risk a trade. Coupled with a high failure rate, for
ODR tops anyway, one should look for a more promising formation. However,
that is not to say that these formations are not useful. They are. Ifyou are con
sidering buying a stock and see a large upward spike on high volume, beware.
Prices will likely head lower. The same applies to downward spikes, which sig
nal a bullish reversal. Again, that is worth knowing if you see one of these
spikes in a stock you own. A large downward spike may cause you concern, but
it is really a bullish event. Prices generally move higher, especially if the close
is near the daily high.
Sample Trade
Consider howJim used an ODR top in a stock he owned (see Figure 28.6). Jim
likes to ski. When he left for his ski trip in midFebruary, the stock was mak
ing new highs. He was having so much fun that he forgot to check in with his
broker and was unaware of the 22% decline (from 25 to 195
/s). When he
returned from his trip and got back into his daily routine, the news awaiting
him was shocking.
Kaufman and Broad (Homebuilding, NYSE, KBH)
Jan 92
Figure 28.6 Oneday reversal top formation following a pipe and a broadening
formation. Jim used the oneday reversal top as the final sell signal. Highlighted is
a pipe top (pretend it is on the weekly scale) and a broadening bottom.
Sample Trade 403
On a weekly scale, the stock made a pipe top suggesting prices would
tumble and they did. As Jim followed the stock each day, it appeared to be
making a sort of broadening formation. His experience told him that a break
out from a broadening bottom (a bottom since prices were trending down to
the formation) could occur in either direction, so he was sure to stay close to
his charts. The day after he saw the ODR top appear on high volume, he
decided that the price was the best he could do. The pipe, broadening forma
tion, and ODR were all clues pointing to the same conclusion: The stock was
going down. He sold that day at 223
/s, well above his purchase price of 105
/8.
InJim's case, he was not trading the formation itself. Instead, he used the
information to protect profits in a stock he already owned. In August the stock
reached a low of 113
/8, about half the price at which he sold.

29
Outside Days
RESULTS SNAPSHOT
Upside Breakouts
Appearance
Failure rate
Average rise
Surprising findings
Downside Breakouts
Appearance
Failure rate
Average decline
Surprise findings
A daily price range wider than the prior day;
breakout is upward
25%
32%, with most likely rise less than 10%
Many. See Table 29.3.
A daily price range wider than the prior day;
breakout is downward
Shortterm (up to 3 months) bearish consolidation
42%
17%, with most likely decline less than 10%
Many. See Table 29.3
Outside days are mirror images of inside days. Outside days are characterized
by a wider price range, one that is both higher and lower than the prior day.
The formation comes in two varieties: those with upside breakouts and those
with downside ones.
Upside breakouts, as the Results Snapshot outlines, sport a failure rate of
25%, just above the maximum 20% rate that I consider acceptable. The aver
age rise is a subpar 32% with a likely rise of less than 10%. Downside break
outs perform substantially worse in some respects. Their failure rate is
exceedingly high at 42% with an average decline of 17%. The average decline
is not bad for a bearish formation but the most likely loss is less than 10%.
With likely gains or losses ofless than 10%, you have to consider whether this
formation is worth trading. Meager gains coupled with a high failure rate sug
gest dial this is probably one formation to avoid.
Table 29.1 lists identification characteristics for outside days. This formation
is easy to identify and it is also common. Figure 29.1 shows what I mean. I have
identified over a dozen outside days in the chart (each dot highlights an outside
day). An outside day is just as it sounds: The daily price range is outside the
prior day. By that I mean the high is higher than the prior high and the low is
below the prior low. In essence, an outside day looks like a 2day broadening
formation; the price range broadens out.
The combination of the wide outside day with the comparatively narrow
day before implies one thing: a price move. Traders, once they spot an outside
day, wait for prices to move above the high or below the low. That becomes
the direction they trade. For example, the first outside day shown in Figure
29.1 begins with a 3
A point difference between the high and low of the day
before (46 to 463
/4). The outside day widens the range to 467
/s and 457
/s. The
following day, the day after the outside day, prices make a new high but the
stock closes unchanged. A trader would be wise to wait for a definitive break
out—a close either above or below the outside day's high or low.
Two days later a breakout results when prices move decidedly lower and
close at 45'/2, comfortably below the low of the outside day. A downward
Table 29.1
Identification Characteristics of Outside Days
Higher high The daily high must be above the prior daily high. No ties allowed.
Lower low The daily low must be below the prior daily low. No ties allowed.
Daily range The day before the outside day must have a daily range. In other
words, the daily high cannot be the same as the daily low.
404

406 Outside Days
Allied Signal (Diversified Co., NYSE, ALD)
)ul 95 Aug Dec Jan 96 Feb Mar
Figure 29.1 Numerous outside days highlighted by black dots. Point A suggests
prices will move higher and they do, but only briefly. Point B is a more timely buy
signal.
breakout appears, telling astute traders that prices should continue moving
down. And that is just what happens. The stock reaches a low of 41'/s in late
October.
Before you get too excited about diis formation, look at point A on the
chart in Figure 29.1. It highlights an outside day too. A day later prices close
above the high posted by the outside day and it appears the stock is moving up.
The breakout signal is correct, too, as prices close even higher the next day.
But that is all. From that point on, prices tumble.
Let us say you buy after the breakout (2 days after the outside day; the
first day after die outside day suggests an upward price move so you buy the
next day) and happen to receive a fill at 46. During September you have a
handful of days to sell the stock at a small profit or breakeven. Your first real
chance to make a tidy sum is during the holidays in late November. On May 1
the stock climbs to over 60. That is not a bad move from 46, but during
the interim you would need the fortitude to wait out a decline to 41 (a 10%
paper loss).
By now you might be asking yourself if this formation really works. Take
a look at point B. This outside day calls the turning point exactly. Imagine get
ting in at the low, 41'/s, and riding it up to 60. That is almost a 50% move. Of
course, it also requires perfect timing, a quality that eludes most of us.
Scan the various figures in this chapter and decide for yourself whether
this formation is trustworthy. Perhaps there are lessons to learn ifwe focus on
the failures.
Focus on Failures
Since this formation has no definitive breakout direction, one cannot assign a
failure to a particular chart pattern simply because it breaks out in the wrong
direction. Instead, I wait for what I call a 5% failure. If a chart pattern breaks
out upward, for example, moves up by less than 5% before turning around and
tumbling, then the formation is a 5% failure. This also applies to downside
breakouts. If a stock breaks out downward but travels down less than 5%
before heading up substantially, it is a 5% failure.
Figure 29.2 shows an example ofa 5% failure. After a sharp run up begin
ning in late April, the stock takes a breather in midMay and essentially moves
horizontally for just over a week before continuing up. It is during this pause
that the stock forms an outside day. The stock trades within a very narrow
range, 26'/s to 26. The following day, the outside day, the stock has a price
range of 263
/8 to 257
/g. The wider range, both above and below the previous
day's range, classifies the new day as an outside day.
Figure 29.2 The outside day suggests a downside breakout but prices move
down by less than 5% before recovering. The result is a 5% failure. Later, a dead
cat bounce drops prices drastically before they bounce.

408 Outside Days
The next day the stock closes lower, suggesting a downside breakout.
However, the stock, although posting a lower low the following day, closes
higher. If investors sell their holdings on the belief that a downward trend
change is at hand, they are disappointed in the short term. The stock moves up,
in the process creating three outside days in a row, peaking in early June at
293
/8. The outside day is a 5% failure because the stock only drops about Vg
before recovering.
Look what happens 3 months later. The stock falls off a cliff and splashes
$9 lower, a massive, 1day decline of 38%. The stock executes a deadcat
bounce and ends up closing even lower 4 months later (typical for a deadcat
bounce).
During the rounding over of the stock, there are a total of five outside
days, three of them predicting a decline. The last outside day with a downside
breakout (suggesting lower prices) is just over a week before the tumble.
We discuss statistics in a moment, but let me reference the failure rate.
For upside breakouts, the formation is above the 20% maximum permissible
rate with 25% being failures. For downside breakouts, the formation performs
even worse: 42% are failures. Of the two, the upside breakout is more reliable
than the downside one, but still below par.
I searched through the formations in the 51 stocks that I catalogued, and
I could find no consistent guide to determining whether a formation is going
to fail or not. With a bull market from 1991 to 1996, the 5year period under
review, it is no wonder that upside breakouts fail less frequently than down
side ones.
At first I thought these formations occurred at shortterm turning points
but that was rejected when I highlighted all formations in a stock (not just 10
per stock as I have in the study). Outside days seem to appear all over the
place. I am inclined to think that these chart patterns are just random events
with no real value. With that in mind, let us look at the statistics.
Statistics
Table 29.2 shows general statistics for outside days. I programmed my com
puter to identify every outside day and it soon became apparent diat there
were too many to log. Not only did they clog my screen but they would have
overloaded my spreadsheet as well, so, just like inside days, I limited the num
ber of outside days I would review.
The computer counts the number of formations in a stock and skips as
many as necessary so that I end up with 10 widely space formations per stock.
For example, if a stock has 150 formations (which is not unusual, let me assure
you) over a 5year span, the computer selects the first 1, skips the next 14, then
selects another and so on until it has 10 identified. Selecting the formations in
this manner spreads them out over time and using only 10 per stock diversifies
Statistics 409
Table 29.2
General Statistics for Outside Days
Description
Upside
Breakout
Downside
Breakout
510 formations in 51 stocks from
1991 to 1996, limited to 10
formations per stock
Failure rate
formations
For successful formations, days
to ultimate high/low
Percentage of outside days
occurring near 12month low
(L), center (C), or high (H)
Volume day before to day after
O.Sx 25day moving average
268
148 consolidations,
120 reversals
67 or 25%
32%
10%
4 months (120 days)
L21%, C25%, 1154%
L49%, C32%, H32%
86%, 112%, 115%
37%
29%
242
133 consolidations,
109 reversals
101 or 42%
17%
10%
L33%, C29%, H38%
LI4%, C20%, HI 7%
86%, 112%, 115%
18%
20%
them over a number of different stocks and industries as well. I believe that
the 510 formations I ended up using are a representative sample of the entire
database.
Since chart patterns with upside breakouts might perform differently than
downside ones, I analyzed them separately. The pattern splits almost evenly
between upside breakouts, with 268 formations identified, and downside
breakouts, at 242 formations. In both cases, they act as consolidations of the
shortterm price trend.
The failure rate, at 25% for upside breakouts and 42% for downside ones,
is above the maximum 20% that reliable formations possess. Failure is mea
sured by first observing the direction ofthe breakout. When prices start a new
trend, the breakout direction is easy to ascertain. When the breakout direction
is less clear, the closing price is the key along with a higher high or lower low.
Eventually, the stock will close above or below the outside day and that
becomes the breakout direction. Then the ultimate high or low is found by
determining when the trend changes, typically with a significant price move in

410 Outside Days
an adverse direction. If prices fail to move higher than 5% before reversing
direction and dropping through the outside day's low (for upside breakouts) or
high (for downside ones), then the formation is a failure. In short, prices must
move more than 5% in the breakout direction to classify as successful.
The average rise or decline is 32% for upside breakouts and 17% for
downside ones. Both figures are a bit shy of the benchmark; a wellperforming
bullish formation typically scores about 40%, whereas bearish formations
return about 20%. So, not only is the failure rate higher than it should be, the
performance is subpar too.
A frequency distribution of the gains and losses gives a better perspective
on what an average investor can expect to earn. Most of the formations have
gains of just 5% to 10%. A few larger gains pull the overall average upward.
Figure 29.3 shows the results of the frequency distribution for both types
of breakouts. The tallest column represents the gain or loss with the highest
frequency. This turns out to be the first column (up to a 10% return) for both
types of breakouts. Adding the columns with similar breakout directions
together, we see that 56% of the formations with downside breakouts have
losses less than 15%. Upside breakouts perform marginally better with 52% of
the formations having returns of 20% or less. In essence, your chances of hav
ing a poorly performing formation is comparatively high.
30 35
Percentage Gain/Loss
• Cains D Losses
Figure 29.3 Frequency distribution of gains and losses for outside days. The
tallest column shows the most likely return, the one with the highest frequency,
and it is 10% for both gains and losses.
Statistics 411
For successful formations, it takes 4 months (120 days) to reach the ulti
mate high and 49 days to reach the ultimate low, on average, for upside and
downside breakouts. That sounds about right. It takes upside breakouts about
twice as long to travel twice as far.
Where in the yearly price range do outside days occur? Those with upside
breakouts usually occur within a third of the yearly high. This is also true for
downside breakouts, but the range is closer to a third for each. Mapping per
formance over the three ranges, we discover the best performing outside days
are those with breakouts within a third of the yearly low. They have gains
averaging 49%. Downside breakouts are more evenly split, but the center third
of the yearly price range performs best with losses averaging 20%.
A review ofthe volume statistics turns up no surprises. The day before the
outside day scores 86% ofthe 25day volume moving average. The outside day
shows slightly higher volume, 112 % ofthe average, and the following day reg
isters 115%. I did not separate the volume statistics according to breakout
direction.
Do outside days showing a large number of shares traded mean a more
powerful move? Yes, and no. Formations with upside breakouts having volume
over 1.5 times the 25day moving average (that is, at least 50% above normal)
show gains of 37%, above the 32% registered for all formations with upside
breakouts. Downside breakouts are similar, with losses of 18% versus 17%
after ignoring volume.
In comparison, low volume shows a different trend. Upside breakouts
perform worse (with a 29% gain) but downside breakouts perform better,
with losses of 20%. For this test, formations with volume levels 50% below
average qualify.
Table 29.3 shows some surprising results for outside days. Does the clos
ing price the day before the outside day predict the eventual breakout direc
tion? Yes, and no. I divided the daily price range into three segments, the
upper and lower segments representing 2 5 % of the price range with the cen
ter section representing 50%. Then I compared the closing price in the upper
or lower segments with the breakout direction. When the stock closes within
2 5% of the daily high the day before the outside day, the stock shows an upside
breakout 46% of the time—less than correctly selecting the side of a coin flip.
However, when the price closes within 2 5 % of the daily low, an upside break
out occurs 61 % ofthe time. I admit that this is a strange result, but that is what
the statistics show. By the way, this is for all formations, even those with 5%
failures, not just those with successful breakouts.
Does the closing price of the outside day predict the breakout direction?
Using the same methodology, an upside breakout when the outside day's close
is within 25% of the daily high predicts correctly 66% of the time. Similarly, a
downside breakout when prices are within 25% ofthe daily low is correct 62%
of die time.

412 Outside Days
Table 29.3
Surprising Results for Outside Days
Description Results
Does closing price of day before outside day predict
breakout direction?
Does the closing price of the outside day predict the
breakout direction?
Does closing price of outside day versus prior day
predict breakout direction?
Do outside days result in larger price ranges the
next day?
The smaller the daily pricerange ratio between
the day before and the outside day, the larger
the rise or decline.
Shorter formations are more powerful.
Maybe
Yes
Yes
17% have larger price ranges
Result is random
True, for upside breakouts only
Does the closing price of the outside day versus the prior day predict the
breakout direction? This is a bit confusing so let me explain. I divided the day
before the outside day into three sections: 25%, 50%, and 25% of the daily
price range. Then I compared the closing price of the outside day with those
three sections. When the outside day's closing price is higher than 25% from
the daily high of the prior day, then an upside breakout correctly signals 65%
of the time. Similarly, a downside breakout happens 62% of the time. I do not
attach any cosmic importance to this finding simply because the outside day is
wider than the prior day, so a close 25% from the prior day's high (low) or
higher (lower) is comparatively easy to reach.
Of the three findings, I place the most importance on the belief that the
close ofthe outside day predicts the breakout direction. If the close of the outside day
is within 25% of the daily high or low, then prices are likely to move higher
(upside breakout) or lower (downside breakout), respectively.
Do outside days result in larger price ranges the next day? No. Only 17%
of the formations show a wider price range the day after an outside day. This
is really no surprise since an outside day by definition is a wide animal, so it is
only natural that the daily price range narrows somewhat the next day.
With inside days, there appears to be a relationship between the daily price
range of the 2day formation with the ultimate gain or loss. Smaller daily price
range ratios perform better than larger ones. With outside days there appears
to be no such relationship. A scatter plot of the results shows the relationship
to be random.
The last surprising finding is that smaller formations appear to be more
powerful. Figure 29.4 shows the relationship. I computed the height (the dif
ference between the intraday high and low) for the outside day and drew a scat
Trading Tactics 413
Figure 29.4 Scatter plot of daily price range versus percentage gain. Shorter for
mations perform better by scoring larger gains. This applies to upside breakouts
only.
ter plot of the range with the resulting percentage gain (for upside breakouts
only). The chart shows that shorter formations have larger price gains. For
example, the highest gain from a formation with a daily price range over $2 is
less than 75%, while the best gain for formations under $2 is over 350%. Many
comparatively short formations have gains over 100%.
As you look at the figure, you can count the number of dots over 100%.
There are 20. There are 185 short formations (meaning a daily range of less
than $2) with upside breakouts, so the chances of any given short formation
scoring an outsized gain is small—about 11 % (that is, 20/185). Ifyour outside
day has a price range of just $0.50, then the likelihood of showing a gain over
50% is about one in three (33%). That is not bad, but the bottom line is: Don't
hold your breath. The relationship for formations with downside breakouts is
random.
Trading Tactics
There are few trading tactics for outside days, so I do not present them in table
form. The first real question you need to answer is ifyou want to trade this for
mation at all. Almost half the formations (42%) with downside breakouts fail.
Upside breakouts perform better (25% failure rate), but they still fall short of

414 Outside Days V ;
the benchmark 20% maximum for reliable formations. Do you really want to
trade this one?
If the answer is still yes, then stick to outside days with upside breakouts.
Remember that a closing price within the top 25% ofthe daily price range dur
ing the outside day correctly predicts an upside breakout two out ofthree times
(but does not address the failure rate, so be careful). If the outside day happens
to have a narrow daily price range, say less than $0.50, that is also an advantage.
Shorter formations with upside breakouts perform better. Justin's trade in the
Sample Trade section also poses additional ideas.
Sample Trade
Justin is a successful doctor. When he is not seeing patients, he toys with new
investment techniques both on paper and in realtime. After reviewing the
outside day formation, he derived five rules that the formation had to meet
before he would invest in it.
1. The formation must have an upside breakout. This is due primarily to
the poor showing of downside breakouts coupled with the larger
gains possible with upside breakouts (32% gain versus 17%).
2. Theformation must be near the yearly low. Outside days within a third
of the yearly low handily outperform (49% average gain versus 32%)
the other two ranges (center third and highest third).
3. The outside day must have volume 50% above the moving average. High
volume outside days score larger gains (37% versus 29%) than low
volume outside days.
4. The daily close must be within 25% of the daily high. This suggests an
upside breakout that, in turn, suggests prices will climb.
5. The daily price range must be less than or equal tofifty cents. Outside days
with a narrow price range perform better than those with a very wide
daily range.
Justin knew that it was only a matter of time before the formation
appeared in a stock he was familiar with and that met his five rules. He was not
going to chase the formation and search for it in stocks he did not know. In
early November, the formation finally appeared in a stock he followed (Figure
29.5).
Working the five rules backward, the outside day has a daily price range
of $0.32 (17.13 16.81), well outside the prior day's high (17) and low (16.94).
This satisfies rule five. The stock closed at the daily high, meeting condition
four. Number three passes as well since the 25day volume moving average (up
Sample Trade 415
Oct 92 Nov Dec Jan 93 Feb Mar Apr May Jun jul
Figure 29.5 Outside day with upside breakout Justin bought the stock after the
outside day appeared and broke out upward. He more than doubled his money.
The day he bought was an inside day.
to, but not including the outside day), is about 95,400 shares, whereas the vol
ume during the outside day is 273,600 shares. This is well above the 50% min
imum (that is, 50% above the moving average or 143,100 shares). The
formation must be near the yearly low, rule two says. The yearly high up to this
point is 21.31 and the low is 15'/8. The close, at 17.13, just slips beneath the
17.19 level that marks the lowest third of the yearly price range. The first rule
says that the formation must have an upside breakout.
All Justin could do at this point was wait. While waiting, he plotted the
stock quotation each day and 2 days later, another outside day appeared. This
did not seem to contradict his analysis so he ignored it and waited for a clear
breakout signal. He received the signal when prices moved above the outside
day's high and closed there. The following day, he bought the stock at I7l
/i, the
low for the day. When he checked in with his broker, he recognized the new
chart pattern as an inside day and hoped that this meant the stock would con
tinue higher. Eventually, that is just what happened.
SinceJustin is a longterm investor and believed in the fundamentals ofthe
company, he saw no reason to take profits anytime soon. There were bumps
along the way, sure, but as a buyandholder, he was unconcerned. On the
weekly chart, he drew upsloping trendlines, then had to redraw them as the
stock climbed even faster in 1995. Then prices pierced the upward trendline,

416 Outside Days V'
moving down. ToJustin, this was a big negative and coupled with some changes
at the company of which he disapproved, he decided to sell. In late January
1996, he sold the stock at 41, below the daily high of 42 but well up from the
daily low of l&A. He more that doubled his money on the trade even though it
tookjust over 3 years to do it.
Justin was early. If he had held onto the stock for 4 more months, he
could have sold at 66. Of course, a month after that the stock was down to 45.
30
Pipe Bottoms
RESULTS SNAPSHOT
Appearance
Failure rate
Average rise
Surprising findings
See also
Two adjacent, downward price spikes on the weekly
chart
12%
SeeTable 30.3
Horn Bottoms
After researching the performance ofhorn tops and bottoms, the natural thing
to do is to remove the intervening week and test the pattern again. That is
where the pipe formation comes from. Pipes bottoms are an exciting discovery
with a low failure rate (12%) and a high average rise (47%).
I conducted an indepth study of pipe bottoms on daily price charts and
came up disappointed. The statistics show that daily pipes have a failure rate of
18% with an average gain of 33%. Both numbers are respectable but what
really bothers me is the most likely gain, which is just 10%. Almost halfthe for
mations (45%) have gains less than 20%. However, there are a number oflarge
gains; almost a quarter of the formations (23%) have gains over 50%.
I began to believe that an investor trading daily pipes would either pick a
formation that fails or one that has such a small gain as to be unprofitable, so I
417

418 Pipe Bottoms v
'
discarded the research and looked at the weekly chart. As you can see from the
preceding numbers, the performance is quite good. Even the most likely rise
holds up well, being 20% (but it can range from 10% to 40% almost equally).
Almost 40% of the formations have gains over 50%. Not only is this formation
worth exploring further, it just might be an outstanding performer worth
adding to your technical toolbox.
Tour
Figure 30.1 shows a pipe formation and the price appreciation that results. The
chart is on the weekly scale and you can see that prices begin dropping in mid
October, 1993, down to the pipe pattern. Prices make a straightline run
downward from a high of 2 9 to the left pipe low of 2 0l
/i. Volume picks up dur
ing the left pipe spike and is even higher the following week. The two down
ward price spikes, of almost equal length and overlapping, mark a turning
point, a signal that the decline is over.
From the low, prices move up smartly and reach a new high of 447
/g in
early November, which is a climb of almost 120% in just 9 months. For this
formation, such large gains are not unusual. Almost 20% of the pipes have
gains over 90%.
Ann Taylor (Retail (Special Lines), NYSE, ANN)
O N
Figure 30.1 A pipe bottom on the weekly chart. Pipes commonly form after a
retrace in an uptrend or at the bottom of a prolonged downtrend.
How do you correctly identify pipe bottoms? Table 30.1 outlines the identifi
cation characteristics. Although there are a number ofguidelines to consider,
they are really quite obvious. Consider the pipe shown in Figure 30.2. The first
guideline suggests that you use weekly charts. Although pipes appear on daily
charts, they do not perform as well as pipes on weekly charts. Two, adjacent
downward price spikes compose the pipe bottom, and it looks like two parallel
lines on the chart. The price difference from the left low to the right low is
minimal. A significant number of formations (414 or 94%) have lowtolow
price differences of $.25 or less. Figure 30.2 shows no difference in the two
pipe lows.
The pipe spikes should appear as a large price drop and wide price range
for 2 weeks in a row. The week before and after the pipe should have low prices
near the top of the pipe. This makes the pipe stand out on the price chart as an
easily recognizable formation. For example, the pipe shown in Figure 30.2 has
a prior week low of 1 !5
/g, somewhat near the left pipe high of 129
/i6 (certainly
Table 30.1
Identification Characteristics of Pipe Bottoms
Weekly chart
Two downward adjacent spikes
Lowtolow price variation
Large spikes
Large overlap
High volume
Obvious pipe
Pipe bottoms on the daily price chart exist, but pipes
on the weekly charts perform better. Use the weekly
chart.
Locate two downward price spikes that are next to
each other.
The price difference between the two lows of the pipe
is small, about $0.25 or less, but you should allow
more variation for higher priced stocks.
Prices should spike down unusually far during the 2
weeks, more than most downward spikes during the
year. The pipe stands alone as the prior week and the
following week have low prices that are near the pipe
highs.
The 2 weeks composing the pipe should have a large
price overlap between them.
Not a prerequisite, but most pipes show above
average volume on at least one or both spikes.
The pipe should be obvious on the chart. If it does
not stick out like a sore thumb, then you should look
elsewhere. The best performing pipes appear at the
end of downtrends.

420 Pipe Bottoms
Conseco Inc. (Insurance (Life), NYSE, CNC)
O N D 92 F N D 93 F M A M
Figure 30.2 Another example of a pipe bottom on the weekly chart. Point A is
another pipe bottom with less spectacular results.
well above the low of 105
/i<;). The right side does even better. It sports a low of
12'/4, near the right pipe high of IZH.
The spike decline should be unusual. The length should be well above the
average downward spike length over the past year. It must appear as a large
decline on the price chart, not just another 2week blip in a sea of long down
ward price spikes. The pipe has a large price overlap. This is clear in Figure
30.2 as the left side of the pipe is just slightly taller than the right side. As a
selection guideline, what you do not want to see is a large left side and a short
right side.
The volume for the week of each pipe spike is usually above average but
need not be. Pipes with above average left volume and below average right vol
ume perform better than all other combinations. However, I would not
exclude a pipe simply because it does not obey the volume characteristics.
The last guideline is perhaps a summation of all others. The pipe must be
unusual enough to jump out of the price chart. Usually, this is because pipes
form at the end of a decline and mark the turning point, such as that shown in
Figures 30.1 and 30.2. Less frequently, pipes act as a consolidation of the
upward trend. They spike downward for 2 weeks, then prices continue rising.
Ifyou look closely at Figure 30.2, you should be able to see another pipe.
I have made it easy for you by marking it as point A. The pipe is not quite as
well defined as the other pipe and the price appreciation is certainly not as
spectacular. Prices rise from the right pipe low of 15;
/s to a high of 18J
/8 before
prices resume their downward trend. This particular formation just clears the
5% failure cut with a gain of 5.5% (as measured from the average of the high
and low price the week following the right pipe to the ultimate high, which
happens to be the same day in this case).
Focus on Failures
Pipes do not have a breakout point, thus there are no upside breakout failures.
With pipe bottoms, there is only the 5% failure. A 5% failure is when prices do
not move higher by more than 5% before reversing the trend. Figure 30.3
shows two such failures. The two pipes in April and May show good definition.
They look like pipes, but they do not act like pipes. After the pipes complete,
prices should move up smartly, but they head lower instead. Why? Volume on
the left spike of both pipes is below the 25day moving average of the volume.
However, the right side shows higher than average volume in both formations.
Still, the volume pattern is unconvincing, as it usually appears most brisk at
pipe bottoms. As a contrast, look at die pipe on the far left side of the chart.
Both spikes show volume that is well above average.
Perhaps the best clue to the failures lies buried in die guidelines outlined
in Table 30.1. Prices should drop unusually far during the 2 weeks, more than
Atlantic Richfield Co. (Petroleum (Integrated), NYSE, ARC)
A S O N D 9 3 F M A M J J
Figure 30.3 Pipe bottom failures. Clues to these two pipe failures are in the spike
lengths and volume trend. The best performing pipes form when prices are trend
ing down.

422 Pipe Bottoms Statistics 423
most downwardprice thrusts during theyear. As you look at the chart, you can see
several downward, 1week spikes (December and February, for example) that
nearly equal the length of die two pipe formations. The entire chart seems
filled with ragged price spikes of varying lengths. For this reason, you should
be skeptical of investing hi these two pipes.
Incidentally, before we leave Figure 30.3, it is a good time to illustrate a
somewhat common feature of pipes. Since pipes often appear at the end of a
downward price trend, prices sometimes rise up, loop around, and retest the
low. The pipe in early December 1991 is an example. Prices top out at 124H
(not shown) in late October then plunge downward. In just 6 weeks they reach
the pipe low of 99'/s. Prices bounce upward, round over, and fall back on them
selves, forming a new low at 98'/s. The retest of the original low completes in
late March.
Pipe bottoms seem to exhibit support at their lows. Rarely do prices drop
more than Vi point or so below the pipe low before recovering and beginning
an extended upward trend. The !
/2 point is not an absolute rule as it depends on
the price of the stock (the one shown in the chart is a $100 stock and it drops
$1 below the prior low).
So even though you may buy into a stock above the top of the pipe and
watch prices fall, hold on. Do not sell until prices drift below the pipe low. I
discuss trading tactics later in this chapter.
You might ask why the March 1992 low is not a pipe formation. The rea
son is the same as the one cited earlier. The downward price spikes do not
descend far enough to differentiate them from normal price behavior. The
two downward price thrusts are not obvious enough to qualify them as a pipe
(and the volume is weak too).
Statistics
Table 30.2 shows general statistics for pipe bottoms. These formations are so
plentiful (442) that I stopped searching at 200 stocks. Most of the formations
are reversals (60%), whereas the remainder are consolidations ofthe prevailing
trend. The failure rate at 12% is quite good. I consider values below 20% to be
reliable. The 52 formations that fail are of the 5% failure variety, that is, prices
fail to continue rising by more than 5% before the trend reverses.
The average rise at 47% is exceptionally high. As a conservative measure,
I averaged the high and low price the week after the right pipe spike and used
it as the base in computations to the ultimate high. This assumes an investor
buys sometime during the week after the pipe and receives a fill in the middle
of the weekly price range.
I use a frequency distribution of gains to find the most likely rise (see Fig
ure 30.4). Ifyou ignore the rightmost column for a moment, the figure shows
that the first four columns are almost the same. The 20% column has the
Table 30.2
General Statistics for Pipe Bottoms
1991 to 1996
Failure rate
Most likely rise
Left pipe volume versus 25day moving average
Right pipe volume versus 25day moving average
442
52 or 12%
47%
20%
130%
117%
highest frequency, followed by 30% and 40%. Thus, you might believe that
the 20% column represents the most likely gain. You can see from the chart
how the tall right column, with gains over 90%, pulls the average upward. If
you sum the values over 50%, you discover that 39% of the formations have
gains over 50%. That is quite good.
The left and right pipe volume numbers are 130% and 117%, respec
tively, ofthe 25dayvolume moving average. These values emphasize thatvol
ume is higher on the left spike and diminished on the right spike but still above
average.
90 >90
Figure 30.4 Frequency distribution of gains for pipe bottoms. The most likely
gain is 20%, but it may be pulled higher by the tall right column.

424 Pipe Bottoms
The time from the right pipe spike to the ultimate high is long, almost 10
months. This follows what we have seen in other formations. Small gains com
plete quickly but an average rise of 47% takes time.
Table 30.3 shows a number ofunusual results for weekly pipe bottoms. I
did not use any ofthe ideas presented in the table while searching for pipe bot
toms. The reason is that I did not want to fit the results to the data. So, I
located the formations then found what parameters worked best (I verified the
results in outofsample tests).
The benchmark is a 47% gain, measured from the average ofthe high and
low prices the week after a weekly pipe bottom to the ultimate high. The gain
assumes you buy the stock sometime during the week after seeing a pipe and
sell just before a major trend reversal. Changing the starting point to the right
pipe low boosts the average gain to an amazing 60% and expands the most
likely gain to a range between 20% and 50%.
Most pipe bottoms, by choice, have little or no difference between the
low prices composing each spike of the pipe. I divided the results into those
that have a difference between the two lows and those that do not. Pipe bot
toms with a price difference have a 49% gain, whereas those with no difference
show gains averaging 43%.
Table 30.3 __
Surprising Results for Pipe Bottoms
Description
Average Rise Failure Rate
Benchmark
Rise for pipes with price difference between lows
Rise for pipes with no price difference between lows
Percentage that lower left (L) pipes work better than
lower right (R) pipes
Right pipe as inside week
Percentage of high volume left (L) pipes, right (R)
pipes, both (B) pipes that score better
Percentage of low volume left (L) pipes, right (R)
pipes, both (B) pipes that score better
High left pipe volume, low right pipe volume
High right pipe volume, low left pipe volume
Linear regression price trend: 3 months up
Linear regression price trend: 3 months down
Large price spike (5x average)
High volume left spike, low volume right spike,
downsloping 3month price trends
47
49
43
L50,R47
50
LSI, R47, B51
L42, R47, B44
53
40
41
51
58, 33 samples
55, 62 samples
12
28
12
11
52, 28 samples 25
Statistics 425
Since a pipe with an uneven low price works better, which works best, a
lower left or right side? It turns out that pipes with a lower left spike show
superior gains at 50% versus 47%. If a lower left spike works best, what of the
high price? I did not do a thorough analysis ofthe various combinations but did
examine how an inside week performs. An inside week is one in which the high
is below the prior high and the low is above the prior low. In essence, the price
range is inside the prior week's range. When the pipe bottom is an inside week,
it scores a gain of 50%.
How does volume play into the performance? I looked at volume on each
of three categories: volume on the left side of the pipe, on the right side, and
on both sides. High volume on the left side and both sides perform best,
accompanying pipes with gains averaging 51%. Low volume on the three cat
egories show that the performance is best on pipes with low volume on the
right side, scoring a 47% rise.
Knowing that the left side with high volume and the right side with low
volume score best, how does die combination perform? The combination
scores an average gain of 53%, well above the 47% benchmark. If you are
unfortunate to locate a pipe with the opposite volume combination (low left
volume and high right), then the average performance sinks to just 40%.
When searching for this formation, I noticed it often forms near the end
of a downward price retrace, so I examined the data to be sure that this is
indeed the case. I measured the price trend by using linear regression on clos
ing prices for the 3 months leading to the pipe. When the price trend is
upward, the average gain is a paltry 41%. Downward price trends do much bet
ter with gains averaging 51%. These results imply that the best gains occur
after prices have been moving down for quite some time. It is as if the forma
tion is signaling a climactic end to the decline. After the pipe, prices move
upward and score outsized gains.
Lastly, I observed that large gains follow large downward spikes. I mea
sured the average spike length over the course of a year by looking at every
adjacent 3week period. If the center week is the lowest, I subtract the differ
ence from the low to the lowest low of the adjacent weeks. This is like mea
suring the smallest difference in length between your middle three fingers. I
added the difference to all the other spike differences and computed the aver
age. Then, I compared this average spike length with pipe spike length. When
the pipe spike is at least five times the average spike length, the resulting gain
is 58%, but there are only 33 formations that make the cut. Dropping the mul
tiple to four times the average lowers the average gain to 55%, but 62 forma
tions make the grade.
Lest you try to read too much into these statistics, let me warn you that
while the average percentage gain may rise, the failure rate usually rises too.
Table 30.3 shows some of the failure rates (those left blank were not mea
sured). For example, ifyou combine the best of the volume characteristics with

426 Pipe Bottoms Sample Trade 427
a downward price trend over 3 months, you discover an average gain of 51%
for pipes showing those characteristics (only 28 formations qualify) but the fail
ure rate zooms to 25%, above the 20% maximum failure rate for which I con
sider a formation reliable.
Trading Tactics
Table 30.4 lists pipe trading tactics. Perhaps the most critical feature of a pipe
bottom is what happens in the third week. While you can easily spot two adja
cent downward price spikes, toss the formation aside if prices do not rise the
third week. The third week, the week following the second pipe spike, should
leave a welldefined dual spike visible on the price chart. The 4week pair
(which includes the weeks before and after the pipe) is Vshaped and is even
more clear when combined with a downward price trend.
As mentioned in Table 30.4, a downward price trend is usuallywhere you
will see these formations, at least the best performing ones. Prices move down,
reach the pipe bottom, then turn around and start climbing. Figures 30.1 and
30.2 are good examples of this behavior.
Once you have identified a pipe bottom on the weekly scale, buy the
stock. Since a stock will often retest the most recent low before starting on a
sustained journey upward, be prepared for it. Place a stoploss order H point
(to allow room for the retest to drift below the pipe low) below the lowest pipe.
If hit, then prices are probably going to continue down. In such a case, close
out your position and send a letter home asking for more money!
Trading Tactic
Table 30.4
Trading Tactics for Pipe Bottoms
Explanation
Downward plunge
Buy
Stop loss
Watch for throwback
Many of the best performing pipes show a downward price
trend leading to the formation. Pipes often occur at the bottom
of a retrace in an upward price rise or mark the end of an
extended price decline.
After a pipe bottom passes the identification characteristics
shown in Table 30.1, buy the stock.
Pipes act as support zones but prices sometimes dip up to '/2
point below the pipe low, so use that as your stoploss point.
Raise your stop as prices climb.
While not a throwback per se (because the weekly scale is being
used and throwbacks happen in one month or less), be aware
that prices sometimes retest their lows and drop slightly below
the pipe low before beginning a sustained upward climb.
Sample Trade
One way to learn how to trade pipe bottoms is to review what Peter did. Peter
is one of the more intelligent software engineers I know. Not only is he smart,
but he is personable as well. He is very helpful and friendly unless management
turns the screws and demands that work actually be done on time. Then the
pressure seeps in and tempers flare. When the pressure gets too intense and
Peter feels the need to take a break, he does not take a walk as most other peo
ple do. Instead, he invests in the stock market. Since he has an Internet con
nection in his office, he is online in just seconds. The situation shown in
Figure 30.5 intrigued him. Prices had been moving horizontally since April
1992, forming an extended base on which an upside breakout of significant
proportions would evolve, he hoped.
Over the shorter term prices began trending down in midJanuary 1994.
They reached a low the week ofApril 4, 1994, accompanied by above average
volume. Had this downward price spike been alone, it might have signaled a
oneday reversal (oneweek reversal really, since we are on the weekly scale).
However, another downward spike appeared the following week. Prices did
not drop to the low of the prior week (383
/8) but they came close at 385
/8. The
dual spikes were long enough to set them apart from the surrounding price
BankAmerica Corp. (Bank, NYSE, BAG)
9 3 N D 9 4 F M A M ) ) A S O N D 9 5 F M A M | | A S O N D 9 6 F M A M |
Figure 30.5 Pipe bottom with preceding brief price dip and following low retest.
Peter bought this stock the week after the pipe completed and sold it for a 71%
gain 2 years later.

428 Pipe Bottoms
action, certainly longer than the brief, 1week, dip in midMarch. The follow
ing week prices moved up smartly, leaving a clear pipe bottom visible on the
chart. At the end ofthat week, Peter bought the stock and received a fill at 42.
He set a stop loss at 377
/8, H below a whole number. Peter placed the stop there
because he knew that whole numbers are sometimes support areas. Placing the
stop just below 38 would give the stock every opportunity to turn around and
move higher.
As Peter watched the stock, he was pleased that it was working out so
well. The real test, he knew, would be when the stock approached the top ofits
trading range. Over the prior 2 years, it had reached a high of 55'/2 and a low
of 401
/2. Ifyou exclude 3 months when prices shot higher then fell back down,
the range was tighter with a high about 49. Peter knew that 49 and 55H were
the keys. If prices pierced those levels, then they would probably continue
moving higher.
Peter watched the stock and when it reached a high of 50'/4 and fell back,
he knew this run was not the one that would send the stock higher. He saw
prices crumble again and hoped that it was only a retest ofthe low and not the
start of a new downside breakout.
During late November, prices reached a low of 385
/8, tying one of the pipe
lows. Then prices moved modestly higher. Peter decided to double his position
and bought more stock. In early February, prices broke out of their congestion
zone and zoomed higher. From that point on, there was no looking back.
Prices continued rising in an almost straightline bead until April 1996.
Then, after setting a new high (803
/8), prices backtracked. Expecting a retrace
in an uptrend, Peter held onto his shares. He watched the shares sink and
when they reached 72, he gave up and sold. Prices dipped to 693
/4 before begin
ning upward again. Peter sold too soon (as it continued moving substantially
higher). Still, he made $30 a share or 71% in about 2 years.
31
Pipe Tops
R E S U L T S S N A P S H O T
Appearance
Failure rate
Average decline
Surprising result
See also
Two adjacent upward price spikes on the weekly
chart
18%
A 3month downtrend (minimum) leading to the
formation results in the best performance, with losses
averaging 24% and an 11% failure rate.
Horn Tops
Ifyou compare the above statistics with horn tops, you will not find much dif
ference because the two formations are similar, even in appearance. Still, the
18% failure rate for pipe tops is quite good, below the 20% maximum for for
mations that I consider reliable. The average decline is 21 % with a likely loss
between 10% and 20%. This is an improvement from horn tops, which show
the most likely loss to be just 10%. The best performing pipe tops are those
with price trends at least 3 months long leading down to the formation. The
pipes appear as part of a retrace in a downtrend and perform much better than
other combinations. They score average losses of 24% with a smaller failure
rate of 11%.
429

430 Pipe Tops Identification Guidelines 431
Tour
Figure 31.1 shows what pipes look like and how they perform. There are three
pipe formations shown in the figure, all of them tops, all warning of an
impending trend change. The pipe on the left occurs while prices are still ris
ing and acts as part of the consolidation of the trend. I consider it a failure
because prices climb significantly above die pipe top before tumbling.
The center pipe really marks the turning point. It towers above the sur
rounding hillside and prices on either side of it fall away. The resulting forma
tion looks like an upside down V.
The pipe on the right is the last one before prices really begin tumbling.
It flags die last chance to exit your holdings or place a short at a good price.
From the high at 39'/4, prices tumble to below 13 by the end of this study—a
67% loss.
Table 31.1 outlines the guidelines for correctlyidentifying pipe tops. First, use
weekly charts as they make pipes easy to spot. You should see two adjacent
upward price spikes. The twin spikes should be unusual in that they should be
well above the surrounding prices and taller than most other price spikes
Pipe
Figure 31.1 Three pipe tops. The first one is a failure because prices rise signifi
cantly above the pipe top.
Table 31.1
Identification Characteristics of Pipe Tops
Weekly chart, upward spikes
Large overlap
Upward retrace in downtrend
Use the weekly chart and locate two adjacent upward
price spikes. The two spikes should be longer than
similar spikes over the prior year and tower above the
surrounding prices.
The two spikes should have a large price overlap be
tween them. Do not pair a tall spike with a short one.
The price difference between the two highs is small,
usually B
/B or less, but can vary up to $1 or so for
higherpriced stocks.
The best performing formations appear at the top of a
retrace in a prolonged downtrend.
throughout the year. Also, the two spikes should show a large overlap between
them; one spike should not be significantly shorter than the other and they
should have many prices in common.
The peaktopeak spike price difference should be small, usually less than
3
/8. In the study, only a handful of formations have tip price differences of $1.
The vast majority (88%) have differences of3
/s or less.
Ifyou look at enough pipe tops, you will discover that many form as part
of an upward retrace during a downtrend such as that shown in Figure 3 1 .2.
Figure 31.2 This pipe top appears during a retrace in a longterm downtrend, as
do some of the best performing pipes.

432 Pipe Tops
You can see that prices peak in midMarch at a price of 397
/8, then move down
for just over a month before recovering. The second top does not last long
before prices start moving down again. Prices take a sharp 1day drop from
3 5'/2 to 29 and move down even further before recovering. The upward retrace
sees prices bounce to a high of 307
/8 from a low of 261
/4. That is when the
pipe forms.
The twin highs of the pipe are just I
/B apart and are well above the sur
rounding high prices. Except for the spike in midMarch, the price spikes are
unusually tall when compared with other spikes throughout the prior year.
Since the pipe spikes both share the same low price, the two spikes exhibit a
large price overlap.
The pipe signals a resumption of the downtrend. In less than 2 months,
prices drop to a low of 18 before recovering slightly.
Focus on Failures
The vast majority of failures occur when prices decline by less than 5% before
resuming their upward trend. These 5% failures do not happen that often, only
18% ofthe time for pipes, but their occurrence is significant enough to warrant
a review of the situation.
Many failures occur when prices are trending up. The uptrend ranges
from several months to over a year and the pipes seem to signal a coming trend
change. Sometimes they do and prices drop, but by less than 5%. At other
times, the drop is more severe, but it is in the future, from 2 to 5 or more
months ahead. In between the pipe and the drop there are higher prices.
Sprinkled among the uptrend failures are those related to downtrends.
Pipes usually appear at the end of a long downtrend or shortly after the trend
changes and begins moving up. Instead of an upward retrace in a downtrend,
the pipe marks the turning point for higher prices.
Consider Figure 31.3, a pipe top in a stock that has been moving sideways
for about a year. Upward breakouts from these long, flat consolidation areas
typically mark the beginning of a long rise, as in this case. For many chart for
mations, even those that have a measure rule to predict what the eventual price
will be, there is still one overriding rule: There must be something to reverse.
You can see that the consolidation region narrows over time, reminiscent
of a long, symmetrical triangle. Even the volume pattern supports the forma
tion by receding most of the way along the chart pattern. Since the boundaries
of a symmetrical triangle mark lines of support and resistance, the possible
decline from the pipe base to the triangle boundary is just 5 %, not a very com
pelling investment. In essence, there just is not much of a climb to reverse.
However, in all fairness, ifthe pipe correctly predicted a downside break
out from the triangle, I would be telling you a different story. Since it is diffi
cult or impossible to predict the breakout direction from a symmetrical
Statistics 433
Figure 31.3 This pipe top forms near the end of a long symmetrical triangle,
investor should wait for a downside breakout before trading this pipe.
An
triangle, it is best to wait for the actual breakout. If investors waited for prices
to pierce the bottom trendline after the pipe appeared, they could have saved
themselves from a loss. As it is, the pipe shown in Figure 31.3 is a failure of
prices to decline. Prices reached a high of 603
/4 in September, more than dou
ble the price where the pipe forms.
More often than not, a pipe acts as a consolidation of the trend, so you
should not depend on it acting as a reversal. Although a reversal does happen
46% of the time, you should invest believing that prices will continue in the
direction of the prevailing trend. If the trend is upward, then prices will con
tinue moving higher. Prepare for the worst; hope for the best.
Statistics
Table 31.2 shows general statistics for pipe tops. I uncovered 443 formations
in just 150 stocks before I decided that my spreadsheet was full enough. Of
these formations, 54% act as consolidations of the prevailing trend, whereas
the remainder act as reversals. The failure rate, at 18%, is just under the 20%
threshold I consider the maximum allowed for reliable formations. Most of
these failures are of the 5 % variety, in which prices begin heading lower but
soon turn around and climb significantly higher (after falling by less than 5%).
The average decline is 21 %, about what you would expect from a bearish chart
pattern.

434 Pipe Tops
Table 31.2
General Statistics for Pipe Tops
Number of formations in 150 stocks from 1991 to 1996 443
Reversal or consolidation 238 consolidations,
205 reversals
Failure rate 80 or 18%
Average decline of successful formations 21 %
Most likely decline 10% to 20%
Left pipe volume versus 25day moving average 123%
Right pipe volume versus 25day moving average 123%
For successful formations, days to ultimate low 4 months (123 days)
Figure 31.4 shows a frequency distribution of losses. The chart suggests
that the most likely decline is in the 10% to 20% range. You can see that the
first three columns are nearly equal in height. Together they represent 55% of
the formations. In essence, over half of the formations have declines of 20% or
less on average. In a statistical anomaly, the volume for both weeks of the pipe
average 123% ofthe 25day volume moving average. In a moment, we will see
that volume—either high or low—plays an insignificant part in whether the
Figure 31.4 Frequency distribution of losses for pipe tops. The graph emphasizes
that the most likely decline is in the 10% to 20% range.
decline is large or small. For those formations that decline by more than 5%,
it takes about 4 months to reach the ultimate low.
Table 31.3 explores some surprising facets of pipe tops. The benchmark,
which comprises all formations that work as expected, has an average decline
of21% with an 18% failure rate. When we separate pipes into those with price
differences between the highs from those that reach the same high price, we dis
cover that the average loss is 23% for those pipes with no price difference. The
average loss from pipe tops with different high prices is 21 %.
Volume, as mentioned in the preceding section, does not appear to be a
significant determinant to the performance of pipes. With high volume on the
left and right side of the pipe spikes, the performance ranges from 19% to 22%,
respectively. Low volume on the spikes results in similar, although slightly
better, performance, at 21% to 23%. The different combinations, however, do
make a difference. For those formations with high volume on the left spike and
low volume on the right, the loss is 18% while the failure rate drops margin
ally to 16%. When you flip the volume around, that is, low left volume and
high right spike volume, the average loss amounts to 24% but the failure rate
also climbs, to 22%.
I used linear regression to assess the 3month price trend leading to the
formation. When the price trend is up, the chart pattern scores an average
decline of 19%. However, when the trend is down, the decline averages 24%
Statistics 435
Table 31.3
Surprising Results for
Description
Benchmark
Decline for pipes with price difference between
highs
Decline for pipes with no price difference
between highs
Percentage of high volume left (L) pipes, right
(R) pipes, both (B) pipes that score better
Percentage of low volume left (L) pipes, right
(R) pipes, both (B) pipes that score better
High left pipe volume, low right pipe volume
Low left pipe volume, high right pipe volume
Linear regression price trend: 3 months up
Linear regression price trend: 3 months down
Pipe Tops
Average Decline
(%)
21
21
23
LI 9, R22, B21
L23, R21, B23
18
24
19
24
22
Failure Rate
(%)
18
18
16
22
11

436 Pipe Tops r
with only an 11 % failure rate. When contemplating pipe tops for shorting
consider looking for a declining price trend.
I also examined pipes for their height, as measured from the lowest high
of the two spikes to the highest high of the two adjacent weeks. Spikes that are
four times or more higher than the average spike (computed over the course of
a year) result in a 22% decline, admittedly not very exciting, especially since
the benchmark is 21 %.
Trading Tactics
Table 31.4 outlines trading tactics for pipe tops. The performance of pipes
depends on the prevailing trend. For larger percentage losses, look for pipes
that appear in downtrends. Pipes will usually appear in an upward retrace of a
longterm decline. Try to find pipes where the decline is evident but just start
ing. What you do not want to do is invest near the end of a downtrend. Of
course, trying to determine when a trend will end is something of an art. How
ever, if the stock has been trending down for many months (such as a year or
more), then you should probably look elsewhere.
In longterm uptrends, the pipe might signal the end of a trend. Some
times it is premature by a few months, so do not be in too much of a rush to sell
the stock short. At other times, a review of the surrounding price patterns
might be rewarding. Double or triple tops sometimes show pipes on one ofthe
tops, calling the turn exactly.
For many uptrends, the pipes represent periods of shortterm weakness.
Prices will move down for a month or two (sometimes more) before resuming
the uptrend. The decline might be 10% to 20% but seldom represents a sig
nificant percentage change. Still, they can be profitable ifyou are careful (and
lucky).
Table 31.4
Trading Tactics for Pipe Tops
Downtrends
Watch for trend end
Longterm uptrends
Uptrend retrace
The best performing pipes occur during downtrends. Prices
bounce upward, form a pipe, then resume their downward
trend.
Do not invest if the pipe appears after a long downtrend. The
pipe may signal the end of the trend.
Look at longterm uptrends. If a pipe appears in an uptrend of a
year or more, then the pipe might signal a trend reversal. Be
careful as the pipes sometimes are premature by 2 to 5 months.
Pipes often appear in uptrends. They mark shortterm weakness
where the trend reverses and moves down. These can be
profitable shortterm moves.
1
Sample Trade 437
Sample Trade
Johnny is a civil servant working in one ofthe state offices. He handles most of
the paperwork for companies just getting started. Most are sole proprietorships
that go bust in less than a year but there are exceptions. He tries to use his daily
contacts to further his investment acumen. Discussions with customers have
helped him spot profitable trends and have helped him avoid costly mistakes.
His interest turned to the steel industry when he learned that the federal
government was thinking of punishing foreign producers for dumping steel in
the United States. He learned about the trend from comments made about
how prices for steel products were dropping rapidly. Companies using the
cheap steel thought the decline was great but die steel companies did not agree.
That is why they started jumping up and down on their favorite politicians.
When Johnny saw the situation depicted in Figure 31.5, he formed a
unique plan to profit from the pipe top. He measured the percentage gain
from the base (point A in the figure) to point B, the first minor high. The rise
was 34% (that is, (193
/4 143
/4)/143
/4). Then he calculated the amount of the
retrace from points B to C, which turned out to be 14%.
As he watched the price climb from point C to the pipe, he whipped out
his calculator and discovered that the percentage change was 36%, quite near
the 34% gain of the first push. He suspected and hoped that the pipe top
marked the start of a downward retrace that would take prices lower, probably
around 15% lower. So, he sold the stock short and received a fill at 22'A. He put
Allegheny Ludlum Corp. (Steel (General), NYSE, ALS)
Pipe
Figure 31.5 This pipe top appeared at the end of a riseretrace pattern that saw
prices climb by 35% and fall by 15%.

438 Pipe Tops
an order to close out his position should the stock decline by 15% to 19. On
the other side, he placed a mental stoploss order at 23 '/4, slightly above the
right pipe high at 233
/i6.
The stock moved horizontally for several weeks then tumbled. When it
reached 19, his short was covered and he made about $3 a share in 5 weeks.
Meanwhile, the stock bottomed out at 183
/4, just below his target and an
amount similar in size to the earlier retrace.
Lest you get too excited about this riseretrace type trade, let me caution
you. Although I have used this maneuver profitably, many times things do not
turn out quite so neatly. Be careful and make use of stoploss orders, especially
if you are shorting a stock. Search for support and resistance zones to help
gauge the ultimate decline.
32
Rectangle Bottoms
RESULTS SNAPSHOT
Upside Breakouts
Appearance
Failure rate
Average rise
Volume trend
Premature breakout
Throwbacks
Percentage meeting
Surprising finding
See also
Downside Breakouts
Appearance
Prices trend down to the formation then oscillate
between two horizontal trendlines before breaking
out upward.
0%
46%, with the most likely rise about 20%
Upward
12%
61%
93%
Actual breakout is opposite that shown by premature
breakouts 75% of time.
RectangleTops
Prices trend down to the formation then oscillate
out downward.
439

Reversal or consolidation Shortterm (up to 3 months) bearish consolidation
Failure rate
Average decline
Volume trend
Premature breakout
Fullbacks
Percentage meeting
Surprising findings
See also
4%
Downward
17%
70%
65%
Actual breakout is opposite that shown by premature
breakouts 75% oftime.
RectangleTops
Like other formations without a classic definition ofa top or bottom, I decided
to give rectangles a definition by separating them based on the price trend
approaching the chart pattern. If the trend is downward, then the formation is
a bottom. If the trend is up, then the formation classifies as a top. The distinc
tion is perhaps arbitrary, but it does help with searching for reasons to a given
breakout direction.
This chapter concerns itself with rectangle bottoms. Bottoms have two
breakout directions: up and down (no surprise, right?). The Results Snapshot
lists the more important statistics. The failure rate for rectangles with upside
breakouts is 0%. Do not get too excited as the sample size is small (41 forma
tions). I do not believe that upside breakouts never fail. They do, it just did not
happen on my shift. One thing is clear, though: They are reliable. The average
rise is 46%, above the usual 40% gain for bullish formations. The most likely
rise at 20% is about what you would expect. With such a large gain, it is no sur
prise that the formation meets its price target 93% of the time. I consider any
thing above 80% to be reliable.
Rectangles with downside breakouts perform almost as well. They have a
4% failure rate. That is very good as I consider reliable anything below 20%.
The average loss is 19%, just shy of the usual 20% for bearish formations. Pull
backs score well, appearing 70% of the time. That score is high enough on
which to base a trading tactic. After a downside breakout, short the stock then
add to the position once a pullback returns to the formation and begins mov
ing down again (always wait for the downside move because prices might con
tinue climbing).
A surprising finding is that a premature breakout is opposite the genuine
breakout direction 75% of the time. We explore this further in the Statistics
section.
Tour 441
Tour
Figure 32.1 shows an example of a rectangle bottom. The shortterm price
trend is upward (for 3 days anyway) leading to the formation, but I discard it.
I look at the intermediateterm trend, which is down, and view the decline a
few days before the formation start as an overshoot. This commonly happens
just before prices oscillate between the support and resistance zones.
The chart pattern forms after prices loop around during the October
1994 to January 1995 period and retrace some of their gains. Prices drop
quickly from the support zone at 57 to 52 H, then they bump up against the top
of the rectangle resistance zone at 56 and slink back to find support at 54.
Prices bounce off the two zones like a PingPong ball ricocheting off
players paddles. Up and down, up and down prices boomerang on volume that
is rising. Soon, one player sneezes and the ball shoots past him. Prices move up,
pausing only a day before moving higher on heavy volume. Prices quickly
climb and enter another congestion zone; this time it is a descending triangle
with an upside breakout.
Kellogg Co. (Food Processing, NYSE, K)
Nov 94 Dec Jan 95 Feb Mar Apr May
Figure 32.1 A rectangle bottom shows an intermediateterm downtrend leading
to the formation with an upside breakout. After the breakout, prices move into a
descending triangle and burst upward out of this formation as well.

Table 32.1 shows a few identification guidelines for rectangle bottoms. While
reviewing the guidelines, consider how they apply to the rectangle shown in
Figure 32.2.
The price trend leading to the rectangle bottom is downward, which is
what separates rectangle bottoms from their top brothers. I usually use the
intermediateterm price trend when considering whether prices are trending
up or down. This is subjective and varies from formation to formation. As
shown in Figure 32.1, I ignore the few days just before the formation starts,
choosing to use the prevailing longer trend instead. Prices bounce between two
levels, setting up a support zone at the bottom and a line of resistance at the
top. Ifyou connect the minor highs with a trendline, it should be horizontal or
nearly so. A similar line drawn below the bottoms forms a parallel trendline.
The two trendlines bound the price action. Occasionally, one of the lines will
not be exactly horizontal or will break near the end, which is fine as long as the
slope is not too steep to disturb the overall picture.
At least two touches of each trendline are required for a valid rectangle.
Figure 32.2 shows three alternating touches. Touches need not alternate, but
you should have at least two clearly defined minor highs and two minor lows
coming close to or touching the trendlines. Except for the briefpunch through
the top in early December, prices stay within the two boundary lines until
breaking out downward on light volume in midJanuary.
The volume pattern is somewhat random but it does seem to track the
breakout direction. For Figure 32.1, the volume pattern is up and so is the
breakout. Figure 32.2 shows high volume at the start of the formation and light
volume just before the downside breakout. I caution against trying to predict
the breakout direction by looking at the volume trend. Although it does track
Table 32.1
Identification Characteristics of Rectangles Bottoms
Downward price trend
Horizontal trendlines
Touches
Volume
The short or intermediateterm price trend leading to the
formation is down.
Two horizontal, or nearly so, trendlines bound prices along
the top and bottom of the formation.
There should be at least two touches of each trendline (at
least four touches total).
Volume follows the breakout direction. For upside breakouts,
the volume trend is upward; downside breakouts show a
receding volume trend.
Figure 32.2 A rectangle bottom appears in a downtrend. Prices fall out the bot
tom then retrace to the rectangle top before moving lower.
the breakout direction more often than not, the chances of it going with the
breakout direction is just above picking the correct side of a coin toss.
Focus on Failures
Figure 32.3 shows a rectangle bottom in a downtrend. Since most rectangle
bottoms act as consolidations of the prevailing trend, the breakout is expected
downward. Prices drop away from the support line at 54% and 2 days later pull
back to the formation. Prices move horizontally for several days before ulti
mately climbing above the top of the rectangle. When prices pierce the top
rectangle trendline and close above it, I consider the downtrend to be over.
Since the initial decline moves just 3% down, the formation is a failure.
Formations that do not move in the breakout direction by more than 5% are
what I call 5% failures. Although this formation ultimately moves lower, it
does so only after closing above the formation top. If you sold this formation
short expecting a price decline, you may be stopped out for a loss. Certainly,
your worry would climb along with prices.
Failures do occur, as Figure 32.3 illustrates, but rectangles are reliable.
Out of 95 formations, just two are failures. That observation conveniently
brings us to the Statistics section.

444 Rectangle Bottoms Statistics 445
|an92
Figure 32.3 Rectangle bottom in a downtrend. This is one of a handful of rec
tangle failures. Prices break out downward and move less than 5% before closing
above the formation top.
Statistics
Table 32.2 shows general statistics for rectangle bottoms. With prices leading
down to the formation qualifying the chart pattern as a bottom, only the break
out direction is uncertain. I divided the statistics according to the breakout
type: upside and downside breakouts.
Of the 95 bottoms studied, I found slightly more rectangles acting as
consolidations (54) ofthe prevailing price trend than reversals (41). Expect the
breakout direction to follow the trend leading to the formation. If, upon entry,
prices are moving down, expect them to continue moving lower after the
breakout 55% of the time. That result is not much above a coin toss but it is
useful.
As mentioned in the Focus on Failures section, just two formations fail,
placing the failure rate at 4% for rectangles with downside breakouts or 2 % for
rectangle bottoms overall. Let me emphasize that because I did not find any
failures of rectangle patterns with upside breakouts does not mean there are
none. I just did not find any in the stocks I was looking at. However, as a rule,
rectangle bottom formations rarely fail, but they do have premature breakouts,
so be careful.
The average gain for upside breakouts is 46% and downside breakouts
lose 19%. The first number is above the usual 40% gain, whereas the second
is below the 20% posted by other bearish formations.
Table 32.2
General Statistics for Rectangle Bottoms
Description
Failure rate
(start to breakout)
Number showing downward
volume trend
Performance when volume is
trending down or up
next 5 days compared with
day before breakout
Upside Breakout
41
41 reversals
0%
46%
20%
38 or 93%
3 months (84 days)
1 9 or 46%
Down 43%, up 49%
163%, 167%, 106%,
91%, 91%, 133%
Downside Breakout
54
54 consolidations
2 or 4%
19%
10% to 15%
34 or 65%
3 months (87 days)
29 or 54%
Down 1 7%, up 22%
171%, 201%, 141%,
124%, 121%, 133%
I use a frequency distribution ofgains or losses to compute the most likely
rise or decline. Figure 32.4 shows the most likely gain for rectangle bottoms
with upside breakouts. You can see that the graph looks quite irregular over
all. Even though the tallest column is 60%, that cannot represent the most
likely gain (it is just too high and is above the average, too). I consider the sec
ond tallest column, 20%, to be representative ofwhat an investor can expect to
earn. The reason for the irregular looking graph is the sample count. With just
41 samples to spread over 10 categories, the numbers can throw off the fre
quency distribution. Oddly enough, the tallest column has the highest number
of samples—10. The 20% column has 8 samples and the others have 4 or less.
Figure 32.5 shows a similar situation for rectangle bottoms with downside
breakouts. The two tallest columns, 10% and 15%, represent the most likely
losses. The first two columns have 12 and 13 samples, respectively, whereas the
other columns have 6 samples or less. Again, due to the small sample size, the
figure may or may not be accurate. However, the results from both Figures
32.4 and 32.5 are about the same as other bullish and bearish formations.
A discussion of the measure rule occurs in the Trading Tactics section,
but it involves adding or subtracting the rectangle height from the breakout
direction. The result is the minimum expected target price. For upside break

25 n
40 50 60 70
Percentage Gain
Figure 32.4 Frequency distribution of gains for rectangle bottoms with upside
breakouts. Due to the small sample size, I consider the 20% column to be the most
likely gain an investor can expect to receive.
25 30 35
Percentage Loss
Figure 32.5 Frequency distribution of losses for rectangle bottoms with down
side breakouts. The two tallest columns represent the most likely loss.
446
Statistics 447
cuts, almost all (93%) ofthe rectangles meet or exceed their targets. For down
side breakouts, only 65% reach their targets. I consider values over 80% to be
reliable, so rectangles with downside breakouts fall short.
The average formation length is the same for either breakout direction,
about 3 months. Those formations with an upside breakout have a receding
volume trend 46% of the time, whereas those with downside breakouts show
receding volume 54% of the time.
If we separate the formations by their volume trend, we discover that
upside breakouts having an upward volume trend perform better (a 49% gain)
than those with a receding volume trend (gains of 43%). The same is true for
rectangle bottoms with downside breakouts. Those with an upward volume
trend score losses of22%, whereas receding volume trends have losses of 17%.
Thus, if your formation has an upward volume trend, expect a slightly better
push to the ultimate high or low.
Shown in Table 32.2 is the volume trend for the day prices move above
or below the rectangle trendline boundary and the next 5 days. For upside
breakouts, the volume is high for a few days then returns to about normal.
Downside breakouts show unusually high volume for the following week. I
doublechecked the statistics and could not find anything wrong with them, so
I guess that is just the way the 54 chart patterns act. Your results, of course,
may vary.
Table 32.3 shows premature breakout statistics for rectangle bottoms
with both upside and downside breakouts. Premature breakouts are rare,
occurring only 5 times (12%) in upside breakouts and 9 times (17%) in down
side ones. The volume pattern is not significantly different from a genuine
Table 32.3
Premature Breakout Statistics for Rectangle Bottoms
Description
Number of premature breakouts
occurring in genuine breakouts
Volume at premature breakouts
Number of formations with premature
breakouts having both upside and
downside premature breakouts
Number of premature breakouts in
same direction as genuine breakout
different direction from genuine breakout
Upside
Breakout
5 or 1 2%
Down 224%
0
0
5 or 100%
Downside
Breakout
9 or 1 7%
Up 21 7%, down 121%
2 or 1 8%
4 or 36%
7 or 64%

breakout to help with identification. Volume, incidentally is compared with the
25day moving average. Premature downside breakouts have more than dou
ble (224%) the average volume, whereas premature breakouts in formations
with downside breakouts are above average (217% for upside premature break
outs and 121% for downside premature breakouts) too. Just two formations
have premature breakouts in both directions. These just happen to occur when
the real breakout is downward.
The last set of statistics in the table suggests premature breakouts are on
opposite sides to the real breakout. A premature breakout is when prices close
outside the formation boundaries but quickly rejoin the formation. The real
breakout comes later as prices soar away from the formation, usually without
returning.
Table 32.3 shows that most genuine breakouts are on opposite sides of
the premature breakout. Only 36% of downside breakouts have breakouts in
the same direction, whereas most of the others are on opposite sides. Combin
ing the statistics indicates that 75% (12/16) ofthe genuine breakouts are on the
side opposite the premature breakout.
Table 32.4 shows breakout statistics. Rectangle bottoms with downside
breakouts show the most likelihood of prices returning to the formation (70%).
Upside breakouts have throwbacks occurring 61% of the time. Both numbers
are good. The average time to complete a throwback is 11 days and for a pull
back it is 9 days, which is about the same as other formation types. The time to
reach the ultimate high is a long 9 months for upside breakouts but a short 2'/i
months for downside breakouts. This makes sense as a large gain takes more
time to travel further.
Where in the yearly price range do most breakouts occur? Most upside
breakouts happen in the center third ofthe yearly price range. Downside break
outs occur in the lowest third of the yearly price range. I expected upside
breakouts to show the highest population in the upper third of the price range
Table 32.4
Breakout Statistics for Rectangle Bottoms
Throwback/pullback
Average time to throwback/
pullback completion
Days to ultimate high/low
near 12month price low (L), center
(C), or high (H)
lookback period
25 or 61%
11 days
9 months (275 days)
LI 9%, C50%, H31%
L45%, C49%, H36%
38 or 70%
9 days
L56%, C32%, H12%
L21%, C16%, HI6%
Trading Tactics 449
since the breakout is at the top of the formation. Mapping the performance
over the yearly price range shows that those with upside breakouts do best
when they break out in the center third ofthe yearly range, scoring an average
return of 49%. It is also odd that those near the yearly high do worst, with
gains averaging 36%. Usually, the momentum players grab the stock near the
yearly high and bid it up.
The best performing downside breakouts are those that occur within a
third of the yearly low, with losses of 21 %. It seems that a stock having a tough
time only sees things get worse after a downside breakout from a rectangle.
This follows the belief that you should short stocks making the new low list
and not those making the new high list.
Trading Tactics
Table 32.5 lists trading tactics for rectangle bottoms. The first tactic is to
determine the predicted price target using the measure rule. The rule first finds
die height of the formation by subtracting the lowest low from the highest
Table 32.5
Trading Tactics for Rectangle Bottoms
Measure rule
Wait for breakout
Tall rectangle scalp
Throwbacks, pullbacks
Other
Measure the height of the rectangle by subtracting the value
of the trendlines from each other. For upside breakouts, add
the height to the top trendline; for downside breakouts,
subtract the value from the bottom trendline. The result is the
expected minimum price move. For a maximum price target,
measure the length of the rectangle and extend it vertically
above the top trendline (for upside breakouts) or below the
bottom one (downside breakouts). The price then becomes
the maximum expected move.
Since you cannot be sure in which direction a rectangle will
break out, wait for prices to dose outside the trendline before
trading in the direction of the breakout.
If the rectangle is tall enough, sell or sell short near the top
trendline and buy or cover near the bottom trendline.
If you have a downside breakout, watch for a pullback and
short the stock or add to your short position once prices begin
descending again. Use the same technique for an upside
breakout: Wait for the throwback then initiate or add to your
position when prices rise.
Watch for rectangles forming as the corrective phase of a
measured move formation and adjust the target price
accordingly. Rectangle reversals sometimes appear as flat
bottom formations

450 Rectangle Bottoms J
high. In essence, just subtract the value of the two trendlines from each other.
Figure 32.6 shows an example of this. The top trendline is at a value of
127
/i6 and the bottom one is at 11. The difference, 17
/16, is the formation
height. Add the height to the value ofthe top trendline to get the upside break
out target (137
/g) and subtract it from the value of the lower trendline to get the
downside breakout target (99
/i6). Some analysts suggest measuring the length
(not the height) of the rectangle, flipping it vertically, and adding or subtract
ing it from the top or bottom trendline to get the maximum price move (for
upside and downside breakouts, respectively). This sounds a bit farfetched,
but it is a handy guideline. Use it with caution as I have not verified how well
it works.
Since you cannot predict the breakout direction with complete accuracy,
wait for the breakout before investing. Place the trade after prices close outside
the rectangle trendlines, then trade with the trend. If the formation is tall
enough, consider placing an intraformation trade near the two trendlines.
Short at the top when prices begin descending and cover when they rebound
off the bottom trendline (do not cover too soon as prices may continue mov
ing down). Go long at the bottom and sell at the top trendline when prices
begin falling. Again, wait for a direction change as prices may stage an upside
breakout.
Throwbacks and pullbacks allow investors another opportunity to place a
trade, add to their position, or get out with a smaller loss. Take advantage ofit
but wait for prices to complete their throwback or pullback before placing a
trade or adding to a position. The reason is that prices may continue in the
adverse direction instead of returning to the trendline and rebounding.
Sometimes, rectangle bottoms form as the corrective phase of a measured
move formation. See Measured Move Down for information on how to take
advantage of the situation. Occasionally a rectangle will mark the end ofa sub
stantial decline and appear like a flat bottom before prices rise.
Sample Trade
Figure 32.6 shows a paper trade I made recently. The rectangle bottom
appeared after prices dropped from a high of 205
/g in October 1997. The drop
was a painful one but it did not occur all at once. Prices dropped quite rapidly
to 15 where they moved horizontally for 8 months. Then the second halfofthe
decline took over and prices reached a low of about 11.
Prices bounced off the low several times, like a boy taking his first steps
on a trampoline. They were tentative, shaky, with not much enthusiasm. Then
in midOctober 1998, prices touched the bottom trendline and moved quickly
across the formation to tie the September high at 123
/s. A few more oscillations
and the two trendline boundaries became apparent.
Sample Trade 451
Shelby Williams Industries, Inc. (Furn/Home Furnishings, NYSE, SY)
Apr 98 May Jun Jul Aug Sep Oct Nov Dec )an 99 Feb
Figure 32.6 Rectangle bottom followed by upside breakout. The measure rule
applied to this rectangle bottom computes the formation height as the difference
between the trendlines. Adding the difference to the value of the top trendline
gives an upside breakout target of
If you look at the overall picture, you might think that prices would con
tinue down—a downside breakout (following the downward trend). I could not
tell which direction prices would go, so I decided to wait for the breakout. If
the formation acted as a consolidation, then the breakout would be downward.
However, with a twostep downtrend from the high at 205
/8, this reminded me
of a measured move down with a really long corrective phase. I thought it
might break out upward. If the rectangle was taller, I would try an intraforma
tion trade (buy at 11'/s, sell at 123
/8, then reverse).
In early December, prices pierced the top trendline and closed above it;
the rectangle staged an upside breakout. I noticed the breakout the day after it
happened and called in my trade. I received a fill at 13, midrange for the day.
I estimated that a support zone had formed at 11 %, so I placed a stop at
1 l5
/s. Prices had stopped at this level just before the chart pattern formed and
again just before the December breakout. A better stop would have been just
below the lower rectangle trendline because both trendlines act as support or
resistance zones. However, I did not want to take such a large loss (15%+).
Even paper trades go wrong and that is what happened here. A day after
buying the stock, prices returned to the rectangle formation to do more work.
Prices slowly, agonizingly, moved lower until hitting my stop in late Decem
ber. I took a paper loss of 11%.

It turns out that I placed a trade on a premature breakout. The statistics
suggest that after an upside premature breakout, the genuine breakout direc
tion should be down. Well, that did not work out either. Prices shot out the top
of the formation. As I write this, they are hovering about 13, my purchase
price last time. I suspect prices might throw back to the formation again. Ifthat
happens and they start moving up again, perhaps I will buy the stock for real.
33
Rectangle Tops
RESULTS SNAPSHOT
Upside Breakouts
Appearance
Failure rate
Average rise
Volume trend
Premature breakout
Throwbacks
Percentage meeting
Surprising finding
See also
Downside Breakouts
Appearance
Prices trend up to the formation then oscillate
out upward.
2%
52%, with the most likely rise between 20% and 30%
Downward
11%
53%
91%
The actual breakout is opposite that shown by
premature breakouts 65% of the time.
Rectangle Bottoms
Prices trend up to the formation then oscillate
out downward.
453

454 Rectangle Tops
Failure rate
Average rise
Volume trend
Premature breakout
Fullbacks
Percentage meeting
Surprising findings
See also
0%
Downward
23%
55%
77%
The actual breakout is opposite that shown by
premature breakouts 65% of the time.
Rectangle Bottoms
Rectangle tops with upside breakouts fail 2% of the time. The average rise is
an astounding 52% with a likely gain between 20% and 30%. These numbers
are excellent. With such a large gain, it is no surprise that the measure rule
works out well with 91% of the formations hitting their price targets.
Downside breakouts tell a similar tale, but one that is not quite as rosy.
Downside breakouts have a 0% failure rate not because they are excellent per
formers, but because I did not find any that failed. I am sure there are rectan
gle tops with downside breakouts that fail to move down by less than 5%. The
average loss is 20% with a strong likely decline of20%, too. This is unusual but
it suggests there are few large declines to skew the average. A glance at the fre
quency distribution shows that the assumption is correct. The percentage of
formations with downside breakouts meeting or exceeding the price target is
77%, just below the 80% threshold that I consider reliable.
The only surprising finding deals with premature breakouts. When a pre
mature breakout occurs in a rectangle top, the probabilities suggest that the
genuine breakout occurs on the opposite side. This relationship is not as strong
as for rectangle bottoms but it comes close.
Tour
Figure 33.1 shows an example of a rectangle top. Prices begin their upward
trek in June 1992 at 14 and reach the rectangle in May of the following year.
Then prices consolidate for over a month, bouncing between overhead resis
tance at 245
/s and support at 235
/s. A trendline drawn across the minor highs is
horizontal as is the one connecting the minor lows. There are a number of
touches of both trendlines suggesting a reliable formation. At the start, prices
overshoot both up and down by peeking outside the two trendlines. This is not
Tour 455
Williams Companies Inc. (Natural Gas (Diversified), NYSE, WMB)
Jan 93 Oct
Figure 33.1 Rectangle top with an upside breakout performs well in this uptrend.
a problem because it occurs too early in the chart pattern before it can be rec
ognized as a rectangle.
The volume pattern begins in the typical manner — receding. However,
about twothirds of the way to the breakout the pattern changes. Volume gets
heavier as if building pressure for the upcoming release. Then, mysteriously,
volume subsides as prices move horizontally just below the top trendline for
over a week. When prices pierce the top trendline, volume picks up but not
remarkably so. Volume just builds on the expanding trend that is developing
since prices began sliding along the trendline top.
Prices climb away cleanly. There is a slight, 3day dip in late June when
it looks as if prices are trying to throw back to the formation top, but the buy
ing pressure is just too strong. The retrace stops and prices turn around and
continue moving up.
Why do rectangles form? A rectangle chart pattern is a struggle between
the haves and the havenots. Those that own the stock but want to sell have
identified a price at which they are willing to part with their shares. When the
price reaches that level, they sell, forcing the price down. When prices fall,
they quit dumping the stock. On the other side is another group of investors
who want to acquire the stock. They place buy orders at what they perceive to
be the fair value. When prices fall to their target price, the buy orders over
whelm supply and the price rises. If this struggle goes on long enough, prices
bounce between one extreme and the other. Over time, you can draw a hori
zontal trendline along the peaks and another along the valleys as a rectangle

456 Rectangle Tops
formation takes shape. Eventually, one of the sides runs out of ammunition. If
the people selling their shares run out first, buying demand overwhelms sup
ply and the price pierces the top trendline. If the buyers spend all their money
and back away from the table, prices drop through the bottom of the rectan
gle. In either case, the shares continue in the breakout direction because of
growing demand (the price moves upward) or increasing supply (the price
tumbles).
Table 33.1 shows identification guidelines for rectangle tops. Over the short to
intermediate term, the price trend should be leading up to the formation. This
upward trend is what distinguishes the formation from rectangle bottoms. The
distinction is arbitrary; I wanted to see if there is any difference in the way the
two perform.
As a rectangle forms, prices rise to a resistance level and fall back to a
support area for another try. If this pattern continues, the minor highs can
be joined with a trendline drawn along the top of the formation, and another
trendline can be drawn below the minor lows. The two trendlines are hori
zontal or nearly so. If there is a slight tilt to the trendline, do not worry. A
slight tilt is fine as long as it does not disturb the overall appearance of a con
gestion region.
To qualify as a rectangle, prices must touch each trendline at least twice.
The touches need not alternate from one trendline to the other, but the minor
highs and lows must be distinct. You do not want to see two touches along the
top as part of the same minor high. Instead, look for two distinct hills and two
valleys at a minimum.
The volume trend varies from formation to formation but usually
recedes. Many of the charts accompanying this chapter show such a trend.
Figure 33.2 shows what a rectangle top looks like. Prices are trending up lead
Table 33.1
Identification Characteristics of Rectangle Tops
Rising price trend
Horizontal trendlines
Touches
Volume
The short to intermediateterm price trend leading to the
formation should be up.
Two horizontal (or nearly so) trendlines outline the price action,
one above the minor highs and one below the minor lows.
There should be at least two touches of each trendline (at least
four touchs total).
Volume usually recedes until the breakout.
May 95 Jun |ul Aug Sep Oct Nov Dec Jan 96 Feb
Figure 33.2 A rectangle top with receding volume trend. Although most rectan
gles exhibit receding volume, do not automatically exclude those with rising vol
ume trends. Three profitable trading opportunities are marked where prices cross
from one side to the other.
ing to the rectangle. Then they bounce between support at 54 and overhead
resistance at 59'/2. The wide, tall rectangle has plenty of trendline touches. If
you are lucky, you might be able to get three or four trades from this forma
tion (as marked by the numbers on the figure). Each sidetoside pass repre
sents a price change of about $5, plenty of profit opportunity to be of interest
to the more adventurous trader.
The volume pattern trends downward over the formation. Near the end,
the volume spurts upward propelling prices higher until they break out and
zoom to new highs. The statistics reviewed later in this chapter suggest that the
majority of rectangles have receding volume trends. That is true but it may be
difficult to see. I would not exclude a rectangle formation simply because the
volume trend is not rising.
Focus on Failures
Of the nearly 300 rectangles I reviewed, just 5 fail. After separating rectangles
into tops and bottoms, the top variety has just three failures. Figure 33.3 shows
an example of a failure. Prices break out of the formation at 35s
/8 and move
upward to a new high of 37. However, they stall in midApril before turning

458 Rectangle Tops
Figure 33.3 A 5% failure of a rectangle top. Prices follow the existing trend
upward but only for a little gain before heading back into the rectangle and shoot
ing out the other side.
around and throwing back to the formation. Once prices choose a new direc
tion, they head down at a good clip. The brief climb represents a 4% price
change. I consider anything less than a 5% move in the breakout direction to
be a failure. The failure confirms when prices close beyond the side opposite
the breakout. It is what I call a 5% failure.
I flag 5% failures because I want a method to catalog poorly performing
chart patterns. Look at this another way: Had you bought this stock when it
left the rectangle top, you would be upset when it throws back to the formation
and continues lower. You might even take a loss if you are not quick on the
trigger. Fortunately, rectangle failures are rare. I uncovered no failures in rec
tangle tops with downside breakouts. That does not mean they will never
occur. It just means I did not spot any in the stocks under review.
Statistics
Table 33.2 shows a number of statistics related to rectangle tops. When com
bined, I uncovered just over 200 rectangles in 500 stocks over 5 years' worth of
daily price data. Most of the formations are consolidations of the prevailing
trend, especially for downside breakouts that have 140 consolidations.
Upside breakouts have a 2% failure rate, an exceedingly low value. Down
side breakouts do even better with no formations failing. I consider zero fail
Statistics 459
Table 33.2
General Statistics for Rectangles Tops
Failure rate
formations
(start to breakout)
Number showing downward
volume trend
Performance when volume is
trending down and up
next 5 days compared with
day before breakout
140
140 consolidations
3 or 2%
52%
20% to 30%
62
62 reversals
0
20%
20%
125 or 91 % 34 or 65%
3 months (97 days) 3 months (85 days)
89 or 64% 44 or 71 %
Down 55%, up 48% Down 20%, up 19%
164%,164%,119%, 168%, 158%, 115%,
105%, 112%, 96% 112%, 107%, 93%
ures to be a statistical anomaly. I am sure that ifI looked at more stocks I could
come up with a few failures. However, for both breakout directions, the low
failure rates underscore that rectangles are reliable formations.
For upside breakouts, the average rise is 52%. This is well above the usual
40% gain. For downside breakouts, the decline averages 20%. The most likely
gains and losses range from 20% to 30%. These values are also strong num
bers, especially for upside breakouts.
Figure 3 3.4 shows a frequency distribution of gains for rectangles with
upside breakouts. I exclude the catchall category ">90%" and conclude that
the tallest columns (20% and 30%) represent the most likely gains. The right
most column illustrates why the overall average is 52% (because a large num
ber of outsized gains pull up the average). When you consider all the columns,
you may conclude that your gains will probably be better than 20% or 30%.
For gains up to 30%, only about a third of the formations (38%) fall in that
range. The remainder have gains over 30%, a mouthwatering return indeed.
Figure 33.5 shows a frequency distribution for rectangles with downside
breakouts. This figures shows that 66% ofthe formations have losses amount
ing to 20% or less. That is a warning, suggesting that your return from rec
tangles with downside breakouts may be less than you hope. The figure also
shows why the most likely rise is the same as the average rise. There are no tall
columns in the upper range to distort the average.

Figure 33.4 Frequency distribution of gains for rectangle tops with upside break
outs. The most likely gain is in the range of 20% to 30%, but several large gains
(over 90%) overshadow the results.
Figure 33.5 Frequency distribution of losses for rectangle tops with downside
breakouts. Two out of three formations have losses less than 20%.
460
A discussion of the measure rule occurs in the Trading Tactics section of
this chapter, but it involves computing the height of the rectangle and adding
or subtracting it from the breakout point. The result is the expected target
price. For upside breakouts, 91% of the formations meet or exceed their price
targets, whereas only 65% of rectangles with downside breakouts make theirs.
I consider values above 80% to be reliable.
The average formation length is about 3 months for both breakout types.
Over 3 months' time 64% ofupside breakouts and 71% of downside breakouts
show receding volume trends. Some analysts suggest that a receding volume
trend is key to a sound rectangle. When I examined this hypothesis, I discovered
that rectangles with upside breakouts and a receding volume trend outperform
their rising volume counterparts with gains of 55% versus 48%. Downside
breakouts are more evenly matched with losses at 20% versus 19%.
Table 33.2 shows the results of a comparison of breakout volume with the
day before the breakout. I used the day before the breakout instead of the aver
age volume because ofthe receding volume trend. The day before the breakout
may be the lowest value in the series and an average volume would probably
involve a higher unfair comparison. Even so, volume is high on the breakout
day and remains high through the following week for both breakout types.
Table 33.3 shows statistics related to premature breakouts. I define a pre
mature breakout to be when prices close outside the rectangle trendlines and
yet return to the formation. For upside breakouts, there are 15 formations
with premature breakouts and 14 associated with formations that eventually
breakout downward.
The volume during a premature breakout is higher than the 25day mov
ing average of the volume, as you would expect. This means it is difficult or
impossible to distinguish a premature breakout from a genuine one. Table
33.3 shows that premature upside breakouts in formations with a genuine
Statistics 461
Table 33.3
Premature Breakout Statistics
Description
Number showing premature breakouts
Volume at premature breakout versus
Number of formations with premature
breakouts having both upside and
downside premature breakouts
same direction as genuine breakout
different direction from genuine breakout
for Rectangle Tops
Upside
Breakout
1 5 or 1 1 %
Up 1 84%,
down 1 30%
2 or 1 2%
5 or 29%
12 or 71%
Downside
Breakout
14 or 23%
Up 146%,
down 132%
0
6 or 43%
8 or 57%

462 Rectangle Tops
upside breakout have volume that is 184% ofthe average. Downside premature
breakouts in the same formation type have volume that is 130% above the
average. Only two chart patterns have up and down premature breakouts in the
same formation.
Does a premature breakout signal the direction of the actual breakout?
Both breakout types suggest a premature breakout is opposite the genuine
breakout direction. A genuine upside breakout follows a downside premature
breakout 71% of the time. Downside breakouts are not so reliable: Only 57%
of the formations with premature upside breakouts later show genuine down
side breakouts. Ifwe combine the premature breakout statistics and compare it
with the total, we find 65% of the premature breakouts are on the opposite side
of the actual breakout.
Table 33.4 shows statistics related to genuine breakouts. Throwbacks to
the formation top and pullbacks to the bottom occur in 53% and 55% of the
formations, respectively. The numbers are not high enough to suggest a trad
ing tactic to take advantage of them.
Both breakout directions complete their throwbacks and pullbacks in 11
days, which is about average for these types of retraces. No formation has a
throwback or pullback more than 30 days after the breakout. Any price retrace
after a month I consider due to normal price action, not a throwback or
pullback.
For upside breakouts, it takes about 8 months to reach the ultimate high
and almost 3 months for downside breakouts to bottom. Ifyou are a longterm
investor of a rectangle top with an upside breakout, be prepared to wait for the
long term before cashing out. For downside breakouts, a shortterm play is
in order.
Where in the yearly price range do most formations occur? Both break
out types occur in the highest third of the yearly range. Mapping performance
over the range shows that the best performing upside breakouts have gains of
Table 33.4
Breakout Statistics for Rectangle Tops
Throwback/pullback
Average time to throwback/
pullback completion
price low (L), center (C),
or high (H)
Percentage gain for each
74 or 53%
11 days
8 months (244 days)
L3%, C21%, H76%
L22%, C51%, H56%
34 or 55%
11 days
3 months (82 days)
LI8%, C36%, H46%
L20%, C21%, H19%
Trading Tactics 463
56% when they break out from formations near the yearly high. Downside
breakouts are about evenly split across the three categories.
Trading Tactics
Table 33.5 explains trading tactics for rectangle tops. The measure rule pre
dicts the minimum target price. First, compute the height of the rectangle by
subtracting die value of the lower trendline from the upper one. Add the dif
ference to the top trendline for upside breakouts and subtract it from the bot
tom trendline for downside breakouts. The result is the target price.
For an example of the measure rule and how it applies to rectangles, con
sider the rectangle top pictured in Figure 33.6. The top trendline has a value
of 383
/4, whereas the bottom one perches at 333
/4. The difference of 5 is the
height of the rectangle. If this rectangle were to break out downward, then the
target price would be 283
/4 or the lower trendline value minus the formation
height. Since the breakout is upward, add the height to the top trendline, giv
ing a target price of 433
/4. Prices reach the target about a month after the
breakout.
I have read that to compute the maximum price move, one physically
measures the length of the rectangle and applies it to the top trendline for
upside breakouts or subtracts it from the bottom trendline for downward
Table 33.5
Trading Tactics for Rectangle Tops
Measure rule
Consolidation
Wait for breakout
Tall rectangle scalp
Other
Measure the height of the rectangle from trendline to trendline.
For upside breakouts, add the height to the top trendline; for
downside breakouts, subtract it from the bottom trendline. The
result is the minimum expected move. For a maximum price
target, measure the length of the rectangle and extend it
vertically above the top trendline (for upside breakouts) or
below the bottom one (downside breakouts). The price
becomes the maximum expected move.
More than two out of three rectangles act as consolidations of
the prevailing trend. Expect the breakout to continue the trend.
Since you cannot be sure in which direction a rectangle will
break out, wait for prices to close outside the trendline before
trading in the direction of the breakout.
If the rectangle is tall enough, sell or sell short near the top
trendline and buy or cover near the bottom one.
Watch for rectangles forming as the corrective phase of a
measured move up formation and adjust the target price
accordingly. Rectangle reversals sometimes appear as flat top
formations

464 Rectangle Tops
Feb 95
Figure 33.6 Rectangle top with breakaway gap and exhaustion gap. Dave traded
this formation after buying it once the throwback completed.
breakouts. When using my computer, the technique comes close to the ulti
mate high. On paper, the results are less accurate. I have not tested this method
extensively and cannot vouch for its accuracy. However, one has to wonder
how measuring a formation (in inches) can accurately translate into a price
move, but who knows, the system might work or at least prove helpful.
Returning to Table 3 3.5, the breakout direction is usually in the direction
of the prior trend. For Figure 33.6 the direction is upward and that is the
direction in which the breakout occurs. Once prices close outside a formation,
then a breakout (or premature breakout) occurs. If the breakout is upward, go
long or cover your short. If the breakout is downward, then short the stock or
sell your position. If the rectangle is tall enough and providing you discover it
quickly enough, you can trade the formation as it swings from trendline to
trendline. Short or sell at the top trendline and cover or buy at die bottom
trendline. Keep an eye on the price trend leading to the formation in case a
breakout occurs. Ifthe stock moves outside the rectangle trendline and you are
losing money, close out your position. You might also want to get on the band
wagon and trade in the direction of the new trend.
Ifthe breakout turns into a premature breakout when prices return to the
rectangle proper, do not panic. There is still a chance that prices will resume
their original breakout direction. Again, ifthe trade goes against you by shoot
ing out the other side of the rectangle, then close out your position and do it
quickly. Ifyou hesitate, you may have another opportunity to add to your posi
tion or close it out if the formation pulls back or throws back. Take advantage
Sample Trade 465
ofit especially ifyou are losing money. Often, prices will return to the forma
tion boundary then turn away. If you do not get out during the pullback or
throwback, then it is likely your losses will grow. Do not pass up the second
chance and do not hope that prices will continue recovering. They will not!
Before placing a trade in a rectangle formation, see if the chart pattern is
part of a larger pattern. Sometimes, the rectangle is the horizontal part, called
the corrective phase, of a measured move up formation. Knowing that a rec
tangle is a subpart of a measured move allows you to get a better gauge on the
expected price move. When the rectangle top is a reversal of the prevailing
price trend, the resulting formation resembles a flat top. Suspect that a rever
sal might be under way if the price trend leading up to the rectangle is unusu
ally steep.
Sample Trade
Dave is an artist. It is tough making a living and he wants to move to the com
puter world and become a graphics artist. He has been playing around with
some hardware and software that duplicate the feel of a brush on various tex
tures but wants to get the latest versions.
Recognizing chart patterns comes easily to him. With his keen eye, he has
been on the prowl for a lucrative stock play. That is one reason he stumbled
across the rectangle shown in Figure 33.6, but he did not spot the rectangle in
a timely fashion. The only reason he noticed it is because of the throwback.
Throwbacks and pullbacks are peculiar enough with their hooking retrace that
they are easy to spot. One has only to look back and identify the associated for
mation.
Dave computed the formation height and applied it to the top of the rec
tangle to get the expected minimum price move. Did he pull the trigger when
prices threw back to the formation? No, he waited. He followed the stock
closely and when it gapped up (a breakaway gap), he bought and received a fill
at 40. Each day the stock moved higher and in 3 days it had reached the target
price of 433
/4. The next day the stock gapped again (exhaustion gap) signaling
an impending end to the rise. The next day prices faltered, and that is when he
sold and closed out his position at 47'/2. He netted over $7 a share or 18% in
less than a week.
He took half his profits out of the market and upgraded his tools. With
the remainder, he kept trolling the markets looking for another opportunity.

34
Rounding Bottoms
RESULTS SNAPSHOT
Appearance A long, rounded upward turn in prices
Longterm (over 6 months) consolidation
38%
Failure rate
for upside breakout
Average rise
Percentage meeting
Synonyms
See also
5%
36%
Rounding Turns, Saucers
BumpandRun Reversal Bottom; Cup with Handle;
HeadandShoulders Bottom, Complex; Scallops,
Ascending and Descending
Rounding bottoms, rounding turns, and saucers are synonyms for the same
formation. The pattern differs from the cupwithhandle and scallop forma
tions in subtle ways, so be sure to study those formations if you are unsure
about identification.
If you consider that this formation acts as a consolidation of the prevail
ing trend and anything acting contrary to that is a failure, then the failure rate
for this pattern is very high at 38%. However, ifyou consider upside breakouts
Tour 467
only (a breakout being when prices move above the left saucer lip), then the
failure rate drops to just 5%. I consider failure rates above 20% to be alarming.
The chart pattern sports a 54% average gain with a likely rise of20%. Almost
a quarter of the formations have gains over 90% and that pulls the average
upward. The large gain masks the performance of the most likely rise. With a
deep cup formed over many months, it makes sense that the measure rule
would have trouble predicting accurate price targets. Prices fulfill the measure
rule just 36% of the time for successful formations with upside breakouts.
Tour
Figure 34.1 shows an example of a rounding bottom on a daily scale. I would
not call it a good example because the bottom is too irregular. In midMay
there is an outofpattern downward price decline that ends with prices quickly
rebounding. In late June prices jump up then fade back down. The June rise is
not uncommon so do not get too excited when it happens in a stock you own.
Prices should return to near the base of the rounding bottom before continu
ing to rise. The volume trend takes on the appearance of being rounded ifyou
ignore the four annoying spikes in the center.
A rounding bottom marks a struggle between buying demand and selling
pressure that is nearly equal. Through the first part of the formation, the
Canandaigua Brands Inc. A (Beverage (Alcoholic), NASDAQ, CBRNA)
Jan94 Feb Mar Apr May Jun ul Aug Sep Oct Nov
Figure 34.1 A rounding bottom on a daily scale. The bottom takes a brief dip in
midMay and a quick rise in late June.
466

468 Rounding Bottoms
sellers have the upper hand as they drive prices lower. Eventually, the forces
come into balance and the stock bottoms out and moves horizontally. Later
still, buying demand picks up and the stock inches upward. The climb is not
always a smooth one. Sometimes, a large upward demand spike occurs send
ing the price skyrocketing, but in a month or so prices head back down and
plane out slightly above where they left off. Then they resume their climb.
When the stock reaches the old high, selling pressure usually drives prices
lower, forming a handle. Prices recover and break through the old high and
push higher still.
As chart patterns go, rounding bottoms are easy to identify. Table 34.1 lists
guidelines for their identification. Since rounding bottoms are often quite long
(in this study, the longest is over 2l
/i years), I usually use the weekly scale to
make identification easy. I search for a price pattern that looks like a bowl or
saucer. Once I discover the pattern, a quick glance backward finds prices trend
ing upward for quite a while. The rounding bottom is usually a gentle retrace
of some of the gains.
Consider Figure 34.2. The most recent up leg of the climb to the forma
tion begins in late December 1991 on very high volume. Prices climb from a
weekly low of about 4 to a high of 95
/ie, a gain of 235% in about 3 months.
Then the stock eases over. Prices move lower and retrace much of their gains.
The decline is not a quick straightdown affair. Rather, the stock moves lower
on its way to 4% by curving around and flattening out.
Once prices reach the low, they move hesitantly higher by traveling hor
izontally for several weeks before beginning an accelerated climb. Prices reach
the level of the left saucer lip and do not pause. They keep climbing until they
reach 13 and then 16 before backing down to 11.
Table 34.1
Identification Characteristics of Rounding Bottoms
Weekly scale
Rounded bowl shape
Curving volume trend
I use the weekly scale to identify these behemoths, although
you can use the daily scale if you wish.
The price trend curves gently usually over many months and
usually after an upward price trend. Connect the weekly low
prices to visually construct a saucer or bowl shape in your
mind.
The volume trend sometimes mimics the price trend by
appearing as a bowl.
91 A S O N D 9 2 F M A M | | A S O N D 9 3 F M A M | J A S O
Figure 34.2 This is a good example of a rounding bottom on the weekly scale.
Notice the bowlshaped volume trend.
A rounding bottom does not require a handle, which is a price consolida
tion area that commonly forms immediately after the right saucer lip, but most
times you will see one. After reaching the lip of the saucer, prices usually drop
and consolidate before heading higher. A handle is typical behavior when
prices reach an old high. The rise falters as tepid demand or excessive selling
push prices lower; then, the two highs act as a resistance zone. Sometimes,
prices make several attempts before pushing through the resistance and mov
ing higher; sometimes, prices just give up and roll back downhill.
The volume trend usually echoes the price trend by rounding upward too.
You can see this in Figure 34.2 although it is not as pronounced as it some
times is.
Focus on Failures
To be fair, the preponderance of failures are probably due to semantics. When
studying this formation, I discovered that most act as consolidations ofthe pre
vailing trend. After moving upward, prices round down, swing around, and
head back up. I label those chart formations that do not fit this template as fail
ures. Figure 34.3 shows an example of such a failure; it acts as a reversal of the
primary trend. Prices trend down beginning on the left side of the chart to the
base of the rounding bottom. Then, prices swing around and begin a new
trend upward.

470 Rounding Bottoms Statistics 471
American Home Products (Drug, NYSE, AHP) Table 34.2
General Statistics for Rounding Bottoms
, M | | A S O N D 9 3 F M A M | | A S O N D94F M A M J | A S ON D95 F M
Figure 34.3 Rounding bottom that acts as a reversal of the trend.
The formation is a founding bottom, but it does not act like a consolida
tion. Ifyou had shorted this formation with the expectation that prices would
continue down, you probably would have wound up with a loss. Thus, I con
sider this formation and others like it as failures.
Another type of failure, called a 5% failure, is when prices move in the
intended direction after the formation completes but fail to travel more than
5% before turning around and heading substantially in the opposite direction.
Only 11 of the formations with upside breakouts are 5% failures. This means
once a formation breaks out upward, there is a good chance it will continue
moving up.
Statistics
Table 34.2 shows general statistics related to rounding bottoms. I uncovered
243 formations in 2,500 years of daily price data. Of the formations I discov
ered, about 67% are consolidations of the prevailing trend and the rest are
reversals.
The failure rate is exceedingly high at 38%. As explained in the Focus on
Failures section, I label every formation that is not a consolidation ofthe trend
a failure and include 5% failures too. Ifwe consider only formations that break
out upward, the failure rate drops to a more palatable 5%, which is very good.
to 1996
Failure rate
Failure rate if waited for upside breakouts
Most likely rise
243
93 or 38%
11 or 5%
54%
20%
50 or 36%
Obviously, the statistic suggests a trading tactic: Wait for prices to climb above
the right saucer lip before buying the stock (watch out for handles too; buy
after the handle forms and prices move up).
The average rise of successful formations is 54% compared with a usual
gain of 40% in other bullish chart patterns. However, the most likely rise is
20%. Figure 34.4 shows a frequency distribution ofgains for successful round
40 50 60
Percentage Gain
Figure 34.4 Frequency distribution of gains for rounding bottoms. Almost a
quarter of the rounding bottoms have gains over 90%.

Figure 34.5 A treacherous example of a rounding bottom that has no left saucer
lip. The rounding turn forms after disappointing earnings send the stock into a
deadcatbounce.
climb above the handle high (or right cup lip). When it did (point A), or so I
thought, I paper traded the stock and bought in, just as it crested (point B).
That turned out to be a minor high. Prices dropped the next day then slowly
recovered making another handle. A good place to sell is when prices pierce
the up trendline in early December.
Sample Trade
How do you use the trading tactics to improve your investment performance?
Consider what Glen did with the situation shown in Figure 34.6. It was his
dream to become a day trader but he had neither the trading capital nor the
necessary skills for the job. He decided to get there one trade at a time.
In December, as he was flipping through his charts, he came across what
appeared to be a mild double bottom. On the daily chart the two bottoms in
August and November were barely discernible. Was it a valid formation and
should he buy the stock now? Glen decided that the retrace between the two
bottoms was not high enough and the two bottoms not clear enough to be
worth considering. He justified his action by thinking that if he was having a
hard time spotting the formation, then others would have the same trouble. If
no one spots the formation, then prices will not rise.
Sample Trade 475
Alco Standard Corp. (Office Equipment & Supplies, NYSE, ASM)
91 S O N D 9 2 F M A M J I A S O N D 9 3 F M A M J | A S O N D 9 4 F M A M | | A S O N D
Figure 34.6 The double bottom formation is barely discernible within the round
ing bottom on the weekly scale.
When he flipped to the weekly chart, it changed the characterization of
what he was seeing. On his screen was an obvious rounding bottom. The vol
ume pattern supported the conclusion: receding as prices declined and round
ing up as prices rose. So, he decided to wait for the rounding bottom to stop
near the prior saucer lip at about 2 !3
/g. When it paused for 2 weeks in Febru
ary, he knew die formation was primed. The question then became, what was
it going to do next? The only way to find that out was to wait.
The following week prices dropped. He waited until prices closed above
the right saucer lip and headed higher. He knew that to buy earlier risked a
downturn in the stock from which it might not recover for a long time. If the
stock ventured above the right saucer lip, then the probabilities suggested a
continuing push higher.
When prices hit 22, he bought. He looked back at his chart and decided
to put a stoploss order H below the saucer lip, just below a support level. He
decided that if the stock hit his stop, in all likelihood it was going down. Con
tent with his investment decision and trading plan, he was confident that his
career change to day trading was a simple step away. He was even more confi
dent as the stock climbed. He began looking through brochures from several
companies that offered seminars on day trading. Then the stock declined and
closed below the up trendline. It was a warning sign that anyone could have
missed. Glen certainly did.

The following week when he received a call from his broker saying prices
had hit his stoploss order, he was shocked. Glen booked a loss ofabout a buck
a share. As he watched the stock, he became even more upset. It turns out the
stock sold at the low for the week.
Three years later, after day trading was over for the day, Glen happened
to review this trade. He decided to pull up the chart and gasped at what he
found. The stock peaked at 66, exactly triple his purchase price.
35
Rounding Tops
RESULTS SNAPSHOT
Appearance
Failure rate
Failure rate if waited for
upside breakout
Average rise
Percentage meeting
Surprising finding
Synonyms
See also
As prices move up, they curve around then breakout
upward.
19%
6%
41 %, with most likely rise between 20% and 40%
69%
Most of the time prices rise after a founding top
completes.
Domes, Rounding Turn
BumpandRun Reversal, Top; Head-and-Shoulders
Tops, Complex
When is a top not a top? When it is a rounding top or dome and prices break
out upward. That is the real surprise with this formation as most of the round
ing tops have upside breakouts. The failure rate is 19%, a sliver below the max
imum rate of 20% that I consider reliable formations to possess. Ifyou wait for
an upside breakout, then the failure rate drops to 6%. At first, I set the break
477

478 Rounding Tops
out point as being onethird up the formation. I considered that a 30% retrace
after the dome rounds over as a sign of an upside breakout. Prices pierce many
downsloping trendlines by that rise and it also gives other investors time to
recognize the trend change (so they help push up the price). However, the
results were just too unreal to use (a 0% failure rate and average climb of 58%).
The computation neglected other formations that began moving up but ulti
mately ended lower. I reworked the figures using the dome top as the break
out point. The results are more in line with other formations. Rounding tops
have an average gain of 41 %, with a likely rise between 20% and 40%.
Tour
Figure 35.1 shows a rounding top on the daily time scale that appears irregu
lar with its many price spikes. The stock begins its upward trek in early Febru
ary. When prices summit in midMay and start rounding over, the climb
amounts to a gain of22%. In a series ofsteps prices decline and reach their low
before turning up in September. Once prices make the turn, they rapidly climb
above the old high and soar to over 71 by the time this study ends.
The reason a rounding turn occurs is not difficult to explain. Prices move
up on bullish enthusiasm confirmed by high volume at the start. Knowing that
prices are climbing, sellers hold on to their shares a bit longer. This forces
demand to climb along with the share price. However, as prices rise, buying
demand tapers offand eventually catches up to supply. Prices round over at the
™ Identification Guidelines 479
top. Since the shares are fetching a premium to intrinsic value, more sellers
appear. The smart money starts selling, too, and the price drops.
Once investors discover the upward price momentum has turned, selling
pressure increases, forcing the price down. Volume may pick up as more
traders try to dump their shares as prices decline. Eventually, the decline ends
when nervous novices toss in the towel and sell their holdings. When all those
who considered selling their shares have sold, the smart money jumps in and
buys the stock.
Table 35.1 outlines the characteristics that rounding tops possess. Using either
the daily or weekly time scale, prices start moving up from the base of the
dome formation. As they move up, they bend over and round off at the top,
then continue their rounding turn until they head down and retrace much of
the prior rise. Buying demand often cuts the decline short before prices return
to where they started.
Like the rounding bottom formation, rounding tops have the same vol
ume pattern, but with rounding tops they are opposite the dome shape. By that
I mean volume is lowest at the center ofthe formation and higher at either end.
This is just a guideline, not an inviolable rule. Many times you will see an irreg
ular volume trend over the life of the formation. Pay it no heed; it is still a
rounding top. What is important is that prices round over and a bowlshaped
volume trend just adds evidence to the veracity of the chart pattern.
Figure 35.2 shows two examples of a rounding top on the weekly time
scale. The first one begins at a low price of 12'/i6 in late December 1991. Prices
rise at a gentle rate then the weekly highs begin rounding over near the top. By
late April prices are heading down again, retracing some of their gains. Prices
only decline to 14 before heading up on their way to the second formation.
Table 35.1
Identification Characteristics of Rounding Tops
|an 95 Feb Mar Apr May |un Jul Aug Sep Oct Nov Dec
Figure 35.1 A rounding top on the daily time scale.
Daily or weekly scale
Rounded halfmoon shape
Curving volume trend
These formations are often long enough to appear on the
weekly charts as well as the dailies.
The price trend curves beginning from the lower left
upward to the top of the dome then rounds over and
moves down again. At the formation end, prices sometimes
bottom somewhat higher than where they started.
The volume trend sometimes appears rounded too, but
inverted from the dome. In other words, volume is
occasionally higher on either end and shallow in the center.

480 Rounding Tops
Sears Roebuck (Retail Store, NYSE, S)
91 D 92 F M A M A S O N D 9 3 F M A M | | A S O N D 9 4 F M A M
Figure 35.2 Two rounding tops on a weekly time scale. The gentle rounding over
gives way to a rising trend most of the time. Note the roundingappearing volume
trend on the right formation.
The second dome is not as rounded as the first, but the volume pattern is
characteristic of rounding tops. Volume is higher at either end of the forma
tion than at the center. Overall, the volume pattern looks like a wide bowl on
the chart.
The price pattern of the second dome finishes higher than where it
begins, just as did the first one. From the low of 2013
/i6, prices climb to a high
of 53 /s a few weeks before the end ofthis study inJuly 1996. That is a gain of
about 260%. The first rounding top has a rise of 210% as measured from the
low inJune 1992 to the high reached during the second rounding top in Octo
ber 1993.
Focus on Failures
As I was searching for rounding tops in the stock charts, it quickly became
obvious that they act as consolidations of the prevailing trend. The statistics
confirmed my hunch. Thus, I consider everything that is not a consolidation of
the prevailing trend to be a failure. The reason for this is the same one posited
for the bottom variety. If you buy into a situation believing that prices will
leave the formation in the same trend as they enter, you will probably be upset
when prices reverse. Reversals are rare, happening less than 20% ofthe time,
but they do occur.
Figure 35.3 shows an example of a rounding top failure. For 5 months
prices rise on their way to the start ofthe formation. Then prices dip for a few
weeks just before the rounding top begins. A person investing in this formation
would expect prices to continue moving higher, resuming the intermediate
term uptrend. However, they would be wrong. Prices reverse and head lower.
From the high reached in the rounding top to the low just 3 months later
prices decline by 60%.
I could find few technical clues as to why this formation did not perform
as expected. The 2week decline leading to the start of the formation is signif
icant as it drops below the intermediateterm trendline, part ofwhich is shown
in the figure. Prices recover and move along the trendline during midOctober
through early November, then rise above it smartly until tumbling below it in
midJanuary. The first trendline piercing during late September serves as a
warning that the end is approaching. Investors were getting nervous about
holding the stock. Eventually, this nervousness translated into higher volume
aspricesplummeted.
I consider an upside breakout to occur when prices rise above the dome
top. Usually prices continue moving higher by more than 5%. When they do
not, it is called a 5% failure. Just six formations (6%) fail to move higher than
5% before plunging. This suggests that once prices break out upward, they
continue moving up.
Amgen Inc. (Drug, NASDAQ, AMGN)
|un 92
Figure 35.3 A rounding top failure. The piercing of the up trendline (partially
shown) before the formation began is a clue to the failure of this formation to con
tinue moving up.

482 Rounding Tops
Statistics
Table 35.2 contains general statistics for rounding tops. Rounding tops are
somewhat rare, occurring only 165 times in 500 stocks over 5 years. Most of
the formations act as consolidations of the prevailing trend; the remainder are
reversals. As explained in the Focus on Failures section, I consider reversals to
be failures of the formation to move in the intended direction. As such, they
comprise most of the failures. If you wait for an upside breakout or for prices
to rise above the dome high, then the failure rate drops to 6%. That is well
below the 20% maximum that I consider reliable formations to have.
The average rise of successful formations is 41% as measured from the
dome top to the ultimate high (when the primary trend changes). This is about
average for bullish formations.
Figure 35.4 shows that the most likely rise ranges from 20% to 40%. The
graph shows the results from a frequency distribution of gains. The tallest
columns are the ones with the highest frequency. I consider them to represent
the most likely rise. Usually the chart looks more like a bellshaped curve, but
this one appears irregular with spikes at 80% and 90%. I think the irregular
pattern is due to the scarcity of formations. If there were more successful
rounding tops, the pattern would appear smoother.
Although rounding tops do not have a measure rule, that did not stop me
from checking to see if an acceptable gauge is available. I calculated the for
mation height from the dome high to the right side low and added the result
to the dome high. Only 69% ofthe formations have prices that move above the
calculated target price. I consider values above 80% to be reliable, so this chart
pattern falls short. The Trading Tactics section discusses use of the measure
rule in more detail.
Table 35.2
General Statistics for Rounding Tops
to1996
Failure rate
Failures after upside breakout
Most likely rise
Statistics 483
165
31 or 19%
6 or 6%
41%
20% to 40%
59 or 69%
6 months (182 days)
30 40 SO 60 70 80 90 >90
Figure 35.4 Frequency distribution of gains for rounding tops. The most likely
rise ranges between 20% and 40%.
The average formation length is 6 months, rather long as far as forma
tions go. This length reinforces the belief that rounding top formations read
ily appear on weekly time charts.
Table 35.3 shows statistics related to breakouts. I consider prices to break
out of the formation when they rise above the dome high (upside breakouts).
When you filter the breakouts into the two types, you find that there are 101
upside breakouts and 64 downside ones.
I did not consider a rounding top breakout to be that significant except
for the direction. The majority of breakouts are upward, contrary to the pop
ular belief that domes represent tops; prices should fall after a rounding top
occurs. That is not what I found with over 100 formations supporting the con
clusion. I feel safe basing the performance statistics on those formations with
upside breakouts.
Table 35.3
Breakout Statistics for Rounding Tops
Upside breakout
Downside breakout
101 or 61%
64 or 39%
1 year (371 days)
L0%, C5%, H95%

484 Rounding Tops
For successful formations with upside breakouts, it takes about a year
(371 days) to reach the ultimate high. That is a long time for a 41% average
gain. We have seen with other formations that large percentage gains take
longer to reach the ultimate high. Rounding tops are no exception but the rise
is perhaps more gentle.
If you split the yearly price range into thirds to see where the breakout
resides, you discover that nearly all break out within a third of the yearly high.
Just four formations (5%) break out in the center third of the yearly price
range. With most formations having breakouts in the highest category, there
is no need to map performance over the three ranges. The formations meet the
average, gaining 41%.
Trading Tactics
Table 35.4 shows trading tactics for rounding tops. The measure rule estimates
the minimum expected price target. Although a measure rule is not supposed
to exist, I found that it works 69% of the time. I view values over 80% to be
reliable, so in that regard, it comes up short of the mark. When using the rule,
be sure to look for resistance areas. They are the areas where the price rise is
likely to stall, forcing the measure rule to underperform. To use the measure
rule, subtract the lowest low from the highest high in the formation, which
gives the formation height. In Figure 35.5, point A shows the lowest low at
45s
/8, whereas point B depicts the highest high at 497
/8. Add the difference, 4'A
(the formation height), to the highest high (point B) to get the target price. In
this case, the target is 54'/s, met in earlyJuly.
There are several ways to profit from rounding tops. The suggested
method is to wait for the breakout, prices to climb above the dome high. Since
prices are already climbing, they continue moving up 94% ofthe time. That is
a reassuring number but no guarantee of success. Ifyou like to take more risk,
Table 35.4
Trading Tactics for Rounding Tops
Measure rule
Buy on breakout
Buy above 30% retrace
Right low support
Compute the formation height by subtracting the right dome
low from the formation high. Add the difference to the high to
get the target price.
Buy when prices close above the dome high.
For a more risky but profitable trade, buy when prices rise
above the right dome low by at least 30% of the formation
height.
The right dome low shows support. If prices throw back to this
level and continue down, sell.
Sample Trade 485
Tandy Corporation (Retail (Special Lines), NYSE, TAN)
Aug Sep
Figure 35.5 A rounding top with a rising wedge. This formation turned into a
profitable opportunity for Sharon. She bought into the situation and sold after the
rising wedge breakout.
buy at a lower price (onethird of the formation height, above the right dome
low). I use the 30% retrace amount since a rise of that magnitude usually
breaks a downsloping trendline that sometimes forms as prices decline during
the rounding turn. A breakthrough ofa trendline or even a 30% retrace is usu
ally strong enough to command attention from other investors (they jump on
the uptrend) and minimizes the chance of a downside breakout.
Ifyou purchase a stock after a rounding turn completes and see prices rise
for a month or so, curl around, and fall below the right dome low, sell the
stock. Most likely it is going to continue down. Watch for a bounce at the right
dome low as that area sometimes acts as a support zone. As always, look for
other areas of support to gauge how far the decline may go.
Sample Trade
Sharon is a highenergy player. She is the one you see careening out ofcontrol
when skiing down the expert slope. She is the one you see night after night
relaxing in a bar after work, surrounded by men. In other words, she is fun to
be with, the life of the party.
Her investment style mirrors her lifestyle. When she spotted the round
ing top pictured in Figure 35.5, she waited for just the right moment to buy. At
first, she thought it might be a headandshoulders top but the two shoulders

486 Rounding Tops
and head were at about the same price level and the volume pattern was
all wrong.
In midJune, when prices began heading up and pierced the down
sloping trendline, she bought the stock and received a fill at 47. Then she held
on and watched the stock daily. As prices rose, she noticed that the oscillations
from minor high to minor low seemed to be narrowing. To her, these oscilla
tions indicated that a rising wedge was forming, but the volume pattern was
abnormal. With a rising wedge, the volume pattern tends to recede over time.
In early September, Sharon grew alarmed because the volume trend
began to decline drastically. Her studies showed a tendency for a severe drop
off in volume just before a rising wedge breakout, so the day after prices
pierced the lower wedge trendline, she sold the stock at 62.
Her analysis was perfect. After she sold the stock, prices pulled back to the
lower wedge trendline and hung on for 2 more days before tumbling. At the
start of the new year, the stock reached a low of 341
/s.
36
Scallops, Ascending
and Descending
RESULTS SNAPSHOT
Ascending
Appearance
Failure rate
Average rise
Volume trend
Percentage meeting
Surprising finding
See also
Descending
Appearance
Prices peak, retrace, and curve around then form a
higher peak. The price pattern looks like the letter J.
25%
33%, with the most likely rise between 20% and 30%
Upward but looks like the letter U—highest at the
ends of the formation
71%
Consecutive scallops in a trend get shorter and
narrower.
Cup with Handle; HeadandShoulders Bottom,
Complex; Rounding Bottoms
Prices peak, curve downward and around, then form a
lower peak. The price pattern looks like die letter J
reversed.
487

Reversal or consolidation Shortterm (up to 3 months) bearish consolidation
Failure rate
Average decline
Volume trend
Percentage meeting
See also
3%
24%, with the most likely decline about 20%
No discernible volume trend
52%
HeadandShoulders Bottoms, Complex; Rounding
Bottoms
The classic definition of scallops refers to the ascending variety only, where
you find repeated saucershaped formations in a rising price trend. I reasoned
that if there is an ascending variety, there probably is a descending variety. I
decided to find out, but before I discuss the two types of formations in detail,
I think it is worth reviewing the major findings.
For ascending scallops, the failure rate at 25% is above the 20% maxi
mum that reliable formations possess. Descending scallops have a failure rate
of just 3%. Why the big difference in failure rates? The difference is because I
use the highest high on the right side of the formation to calculate the per
centage change to the ultimate high or low. For ascending scallops, this penal
izes performance since the right edge is much higher than the left. For
descending scallops, it helps performance because the right side is well above
the formation low and the ultimate low.
The percentage rise for ascending scallops is 33%, respectable but
mediocre when compared to other bullish formations. For descending scallops,
the declines average 24%; that is quite good (we usually see a 20% decline for
bearish patterns). The measure rule is weak in both species and especially
so with the descending variety—only 52% of the chart patterns reach their
price targets.
One surprising finding is that consecutively ascending scallops get nar
rower and shorter, on average, when compared with prior scallops in a series.
For example, in a line of four ascending scallops, the first one will be wider and
taller than the last one. The relationship for descending scallops is unknown
because of a dearth of consecutive formations.
Table 36.1 outlines identification guidelines for both ascending and descend
ing scallops. One difference between the two types is the price trend leading to
the formation. As the name implies, ascending scallops appear when prices are
Table 36.1
Identification Characteristics of Ascending and Descending Scallops
Pricetrend Prices should be rising leading to ascending scallops and declining
toward descending scallops.
j shape Both ascending and descending formations have two price peaks
with a rounded recession in between. The ascending variety has a
higher right peak, whereas the descending scallop has a higher left
peak. Ascending scallops look like the letter J and descending
scallops look like a reversed J.
Volume Ascending scallops often show a Ushaped volume trend that gets
heavier over time, but there is no discernible volume trend for
descending scallops.
moving higher over the intermediate to longterm trend, that is, over 3, 6,
or more months. The descending variety occurs when prices are moving
steadily down.
Figures 36.1 and 36.2 show examples ofascending and descending scallop
formations, respectively. Figure 36.1 shows three ascending scallops with the
first one being an especially large one. It looks like a rounding bottom except
that the minor high, where the formation ends on the right (in midApril), is
well above the minor high on the left (during early December). This is typical
for ascending scallops—the right side should be above the left. However, it is
Great Atlantic and Pacific (Grocery, NYSE, GAP)
Scallop
36
Oct 92 Nov Dec |an 93 Feb Mar Apr May Jun Jul
Figure 36.1 Three ascending scallops. The formation resembles the letter).
20

CNF Transportation Inc. (Trucking/Tramp. Leasing, NYSE, CNF)
Feb93 Mar Apr May |un |ul
Figure 36.2 Four consecutive descending scallops.
all right if the two peaks are close to each other in price. This often signals an
end to the series of scallops and the rising price trend.
The Jshaped pattern appears on the smallest scallop in Figure 36.1. I
highlight this formation with some consternation. When hunting for scallops,
one should look at the price lows, not the highs. Ifyou connect the minor lows
of the first two formations, you see that prices have a bowl shape. The bowl
shape is not clear in the smallest formation unless you trace along the highs.
The smallest formation in Figure 36.1 also has the best volume pattern—
a Ushaped trend. This is common for ascending scallops but should not be
viewed as a requirement. The first scallop does not have an easily recognizable
bowlshaped volume trend but it is there. The volume spikes are higher near
the formation ends than in the center.
Figure 36.2 shows four descending scallops. You can see that the overall
price trend is downward. It starts on the left at about 20 and saucers down to
about 15. The descending scallops appear like reverse J patterns. The minor
high on the right is below the left minor high and between the two peaks is a
rounded recession. You can see that the last scallop has minor highs that are
nearly equal. This often suggests the receding price trend is nearing an end. In
this case, prices reach the ultimate low in less than 2 months at 13s
/8, quite close
to the last bowl low of 143
/4.
The volume trend is irregular. I have noticed a tendency for a volume
spike to appear near the center of the formation as prices switch from moving
downward to upward. In Figure 36.2 you can see the spikes in late March and
midJune.
^^ Focus on Failures 491
Focus on Failures
Scallops suffer from what I call 5% failures. A 5% failure is when prices break
out in the intended direction but fail to continue moving in the same direction
by more than 5%. They double back and head in the opposite direction, some
times causing an investor to lose money. Figures 36.3 and 36.4 show examples
of failures. There is nothing wrong with the ascending scallop in Figure 36.3
in the AprilMay period. Prices round up nicely and continue higher while the
volume pattern is bowlshaped if you disregard the twin spikes in early May.
However, the late June formation marks the high for the stock. Again, there is
really nothing wrong with the pattern. The J shape is pronounced and smooth.
The volume pattern is somewhat rugged but higher on either end than in the
center. The narrowness of the formation is a clue to its failure. It is about 2
weeks wide, which is quite narrow for scallops (the average width is about 2
months). From the high at 19, the stock heads down in a choppy manner until
the end of the study (midJuly 1996) where it is at 13'/2.
Descending scallop failures are similar to the one shown in Figure 36.4.
The formation sometimes acts as a reversal of the downward trend since prices
move up substantially after the formation ends. The price trend leading to the
formation is downward. This is not entirely clear from the figure, but the stock
reaches a high of 365
/s in late September 1993, then prices move down until
AbitibiConsolidated (Paper & Forest Products, NYSE, ABY)
High Point
Mar 95 Apr May Sep Oct
Figure 36.3 An ascending scallop failure in late June. Most scallops act as con
solidations of the trend but the narrow ascending scallop in late June marks the
high for the stock.

American Express (Financial Services, NYSE, AXP)
Statistics 493
Apr 94 May |un Jul Aug Sep
Figure 36.4 A descending scallop failure. This descending scallop acts as a reversal.
they reach a low of 251
A in earlyJuly. After the formation reverses, prices climb
and reach a high of 50% in early April 1996.
Ifyou measure from the right minor high to the ultimate low (as marked
on the figure), the decline is less than 5%—a 5% failure. The only thing that
stands out about this formation is that it occurs well into a downtrend imply
ing a trend reversal is likely.
We see in the Statistics section that most descending scallops form near
the yearly low, suggesting the trend has been moving down for some time.
This downward trend coupled with a poor measure rule showing, where about
half the formations fall short of their price targets, leads one to believe that the
performance may be worse than the statistics suggest. In short, be careful about
investing in a descending scallop that forms well into a downtrend.
Statistics
Table 36.2 lists statistics for ascending and descending scallops. These forma
tions are plentiful—I logged between 400 and 600 formations of each. Most of
the formations act as consolidations of the prevailing trend. If the trend lead
ing to the formation is upward (for ascending scallops), then prices resume
their uptrend shortly after the formation ends.
I log a failure if prices move in an unexpected direction for too long. For
ascending scallops, a failure occurs when prices move downward and slip below
Table 36.2
General Statistics for Ascending and Descending Scallops
Description
Ascending
Scallops
Descending
Scallops
Failure rate
formations
(measure rule)
Percentage of scallops occurring
near 12month price low (L),
Percentage gain for each 12
Consecutive number of formations
613
492 consolidations,
121 reversals
151 or 25%
33%
20% to 30%
327 or 71%
2 months (57 days)
8 months (231 days)
L2%, C14%, H84%
L56%, C40%, H31%
2 = 67%, 3 = 38%,
4 = 18%, 5 = 8%,
6 = 3%, 7 = 2%,
8 = 1%
414
310 consolidations,
104 reversals
14 or 3%
24%
20%
206 or 52%
L51%, C34%, HI5%
L24%, C23%, H27%
2 = 62%, 3 = 27%,
4 = 9%, 5 = 3%,
6 = 1 %, 7 = 0%
the formation low without first moving 5% above the right formation edge.
Failures for descending scallops are similar in that prices need to rise above the
right edge high without first dropping by more than 5%. The failure rates—at
25% and 3% for ascending and descending scallops, respectively—measure
from the high reached in the right scallop edge to the ultimate high or low. As
explained earlier, this measurement penalizes ascending scallops and assists
descending ones. The measurement accounts for the different failure rates and
explains the subpar 33% gain and exceptional 24% average loss for the two
chart patterns. Despite these variances, I consider the measurement method to
be acceptable since the right minor high, when completed, is easy to identify
and serves as a good reference point.
I compute the most likely gain or loss by using a frequency distribution of
the gains and losses. Figure 36.5 shows a graph of the gains and Figure 36.6
shows a graph of the losses.
In Figure 36.5 the two highest columns represent the most likely gain
because they have the highest frequency. Even though the average gain for all
successful formations is 33%, the most likely gain rests between 20% and 30%.

Percentage Cain
Figure 36.5 Frequency distribution of gains for ascending scallops. The most
likely gain is between 20% and 30%.
6%
25 30 35
Percentage Decline
50 >50
Figure 36.6 Frequency distribution of losses for descending scallops. The most
likely loss is 20%.
494
V. Statistics 495
Figure 36.6 shows a distribution ofthe losses. The tallest column, at 20%,
represents the most likely loss since it has the highest frequency. However, the
two adjacent columns are quite close to the tallest one so the likely decline may
range between 15% and 25%. I view these figures as representations of what
you might expect to make ifyou trade these formations.
I discuss the measure rule in the Trading Tactics section, but it involves
computing the height of the formation and adding the result to the right edge
high or subtracting it from the lowest low in the formation. The result is die
minimum price target. For ascending scallops, 71% of the formations meet or
exceed their predicted price targets, whereas only 52% of descending scallops
meet theirs. I view values above 80% to be reliable, so scallops fall short of the
benchmark. What does this mean? Ifyou buy a stock and see a scallop develop
near the beginning of the price trend, prices will likely meet the target. How
ever, if the trend has been in existence for a long time (say, over 4 months),
then there is a greater chance that prices will not meet the predicted target. Of
course this depends on the overall market. Declines during a raging bull mar
ket may be shortlived, so factor in the market accordingly.
The average formation length is quite short, less than 2 months for both
types of scallops. Some, such as that shown in Figure 36.1, will be longer than
average and some will be shorter, but overall, there seems to be a tendency for
the formation to last between 1 and 3 months.
The time it takes to reach the ultimate high or low is about double that
for ascending scallops than for descending ones. Coupled with the average rise
or decline, this suggests the declines are steeper, more violent, and the rises
more drawn out and sedate. It also suggests investors should be patient, at least
for ascending scallops, and let the stock play out before selling.
Where in the yearly price range do the formations occur? For ascending
scallops, most ofthem occur within a third ofthe yearly high as measured from
the high at the right end of the formation. For descending scallops, most occur
within a third of the yearly low. Overlaying the performance on the yearly
price range, we discover that ascending scallops occurring within a third of
their yearly low perform best, with a 56% average gain. I hasten to add that
there are only 8 formations in this category, well short of the 30 samples
needed to make definitive conclusions. However, die other two categories have
enough samples and suggest that the lower diey occur in the yearly price range,
the better performance will be. Descending scallops, on the otiier hand, are
mixed. Performance is essentially flat, ranging between 23% and 27%.
I included a tabulation on the number of consecutive formations in a
stock. In essence, I wanted to know how likely it is diat a second, diird, or
fourth (and so on) formation would occur in a trend. It turns out that it is quite
common to have more than one scallop in a single uptrend or a single down
trend. In odier words, about twothirds of the time a second scallop appears

after the first one (in the same trend). A third scallop will appear about a third
to a quarter of the time. Table 36.2 shows the remainder of the percentages.
Why is this significant? If you are considering shorting a stock that shows
three cascading scallops, there is a good chance that you will be making a mis
take. Only 9% of the formations have four descending scallops in a row and
that suggests the end of the downtrend is near. The more scallops you find, the
higher the likelihood the end of the trend is approaching.
Do ascending scallops change their shape as they climb? I measured the
width of each scallop and where it occurs in a series of consecutive scallops
(over a single uptrend). Figure 36.7 shows the results. Only four points are
shown because the sample size diminishes beyond four scallops in a row. The
first scallop has an average width of 62 days but by the time the fourth scallop
in a series appears, the width decreases to 47 days, on average. So, if you dis
cover a narrow ascending scallop well into an uptrend, you might consider
avoiding the stock. Otherwise, you might be investing near the top.
Do ascending scallops become flatter as they climb? Figure 36.8 shows
the result of the analysis. The first scallop has an average height, as measured
from the highest high to the lowest low, of about 18%. This drops to 14% by
the time the fourth scallop in a row appears. Again, I graphed only four points
because of the rarity of trends containing more than four ascending scallops in
Number in Series
Figure 36.7 Graph showing narrowing width of ascending scallops over time.
The more consecutive ascending scallops that appear in an uptrend, the narrower
they become. If you see a narrow ascending scallop forming after a long uptrend,
you might avoid taking a position in the stock.
Trading Tactics 497
Figure 36.8 Graph showing formation height of ascending scallops. Ascending
scallops become slightly flatter as they climb.
a row. I tried to apply the analysis to descending scallops without success.
Only three points are available (because of the sample size) and the results are
inconclusive.
Trading Tactics
Table 36.3 shows trading tactics for ascending and descending scallops. The
first trading tactic is to determine how far prices are likely to move once the
formation completes. This is called the measure rule because it involves mea
suring the formation height and applying it to the breakout point.
The measures for both ascending and descending scallops begin by com
puting the formation height in the same way. Subtract the lowest low reached
in the bowl from the high reached on the right side of the formation. Once you
have the height, add the value to the highest high on the right side of the for
mation for ascending scallops or subtract it from the lowest low for descend
ing scallops. The result is the minimum expected price target. An example
makes the calculation clear. Consider the ascending scallop that forms during
late September as shown in Figure 36.9. Apply the measure rule to this forma
tion by subtracting the formation base from the right side high. Point B shows
the base low at 12'/2 and the right side high, point A, is 16. The difference of
3 /2 is the formation height. For ascending scallops, add the difference to the
rightside high (point A) to get the target price of 19'/2. Prices meet the target

498 Scallops, Ascending and Descending %J
Table 36.3
Trading Tactics for Ascending and Descending Scallops
Measurerule Compute the height of the scallop by taking the difference
between the rightedge high to the lowest low in the formation.
For ascending scallops, add the difference to the highest rightedge
high, and for descending ones subtract the difference from the
lowest low. The result is the minimum expected price target.
Lip retrace Once prices crest the right lip high, prices fall. If they drop below
the bottom of the formation in an uptrend, then the formation is a
failure and you should avoid the stock or close out your position.
Likewise, if prices rise significantly above the top of the right edge
in a downtrend, then the formation is also a failure.
Buy point Take a position in the stock once prices drift below the rightedge
high. For ascending scallops, wait for prices to bottom out. For
descending scallops, sell short immediately once prices reach the
rightedge high and head down.
Stops Ascending scallops: '/B below the lowest low. Descending scallops:
VB above the right lip high.
Avoid Be careful of shorting a stock showing a descending scallop that
forms many months into a downtrend. The downtrend may be
near the end.
in late April (not shown on the chart). If the scallop is descending, then sub
tract the difference from the formation low (point B) to get the target price. In
such a case, the target would be 9 (12'/2 3'/2).
In both cases, the formation height uses the rightside high, not the left. For
ascending scallops, this makes the measure rule much harder to fulfill because
the right side is much higher than the peak on the left. For descending scallops,
the measure rule is easier to meet because the right side is lower than the left
side. Still, ascending scallops meet their targets more often than descending
ones by 71% to 52%. This statistic implies that the descending variety form
near the end of the trend—something you should keep in mind if you intend
to short a stock containing a descending scallop.
Once a scallop completes, prices decline. They retrace all or a part of
their gains (that is, from the rightedge high to the bowl low) before heading
higher (in the case of ascending scallops) or continuing down (for descending
scallops). In Figure 36.9, you can see that the retrace after the first scallop
brings prices down to the height of the left scallop lip at 14'/4. The retrace after
the center scallop sees prices return to near the bowl low.
For ascending scallops, once prices crest on the right side and begin
declining, wait for the decline to end. In some cases, another scallop will form
and it will be relatively easy to buy during formation ofthe bowl. For descend
ing scallops, you will want to sell short as soon as the right side peak becomes
obvious and prices head down.
Sample Trade 499
CKE Restaurants, Inc. (Restaurant, NYSE, CKR)
Ascend ng Scallop Ascending Scallop
Figure 36.9 Three consecutive ascending scallop formations. Kristy bought the
stock at point C once prices rose above the top of the ascending scallop. The last
scallop has a Vshaped bowl and a right rim that almost makes it to the high of the
left side. She sold at point D.
Stoploss points should be '/s beyond the support or resistance level. In
Figure 36.9 place a stoploss order at 123
/8, or '/s below the formation low
(point B) for the first scallop. If the loss from the purchase point is too large,
consider moving the stop to just below the left peak. As you can see in Figure
36.9, the left peak is an area ofsupport, but it varies from formation to formation.
For descending scallops, place the stop for short trades l
/% above the right
peak. In Figure 36.2, the scallop on the left would have a stop placed at 19'/2.
Sample Trade
How do you trade these formations? Sometimes it helps to have inside infor
mation. That is what Rristy's boyfriend is doing time for in a lowsecurity
prison. She has become the primary trading arm of the relationship, a hobby
she had long before her beau came along.
Kristy was intrigued by the scallop formation shown on the left in Figure
36.9. The Vshaped look to the bowl concerned her as did the poorly shaped
volume pattern. But she liked the prospects for the restaurant company and her
fundamental analysis was thorough and tasty.
Before she bought the stock at point C, she computed the estimated gain
and compared it to the risk ofa loss. The targeted rise was to 183
/4 (she calculated

using the rightside peak 3 days earlier). The risk point was 14, the high of the
left side and a massive support area reached in early 1994. At her purchase point
of15'/4, theriskwas 11
A (15'/4 14) and thepotential rewardwas 3'/2 (183
4 15!/4).
The nearly three to one ratio was high enough to risk a trade.
She felt gratified when prices closed at the high for the day, suggesting
prices the following day would move higher still. When she looked at the stock
the next day, prices did reach a new high but closed lower. As she posted her
daily quotes for the stock, the declining price trend over the next week or two
concerned her, but not unduly so. Rristy recognized the rounding turn of
another scallop forming and saw that her stop held.
Day by day she followed the stock and did not like the third scallop in the
series (the rightmost one). The bowl shape was irregular and the volume pat
tern was unconvincing. When prices stopped at the old high before collapsing,
she knew the rise was at an end. She pulled the plug on the operation at 16%,
shown as point D in the figure.
In the short term, Kristy was right in that prices headed lower. They
moved down until reaching the low of the bowl but then rebounded. By mid
June, theyhad nearly doubled, reaching a high of28 A, 10 points above the tar
get price of 18%. Still, on her 1,000 shares, she cleared almost $1,500 on the
trade.
37
Shark32
Upside Breakouts
Appearance
Failure rate
Average rise
Volume trend
Throwbacks
Surprising finding
See also
Downside Breakouts
Appearance
Failure rate
Average decline
A 3day symmetrical triangle with consecutively
lower highs and higher lows. Breakout is upward.
41%
Downward
64%
Horizontal symmetry improves results.
Symmetrical Tops
A 3day symmetrical triangle with consecutively
lower highs and higher lows. Breakout is downward.
Shortterm (up to 3 months) bearish reversal (55% of
formations) or consolidation
44%
21%, with most likely decline less man 10%
501

502 Shark32
Volume trend
Fullbacks
Surprising finding
See also
Downward
58%
Horizontal symmetry improves results.
Symmetrical Tops
Despite this formation being only 3 days long, I applied the 5% failure rule to its
performance. A 5% failure occurs when prices move less than 5% in the breakout
direction before reversing and moving significantly in the new direction. For
sharks with upside breakouts, the failure rate is exceedingly high at 41 %. The fail
ure rate is even worse for downside breakouts at 44%. I consider formations to be
reliable if the failure rate is below 20%. The reason for the poor showing is, in
part, because ofthe compactness of the formation. Prices break out in one direc
tion, reverse course, and shoot out the opposite side of the formation, scoring a
failure. With larger chart patterns, the stock has more opportunity to resume the
original breakout direction before making it to the other side of die pattern.
The average rise from upside breakouts is 32%, well below the 40%
posted by wellperforming formations. Downside breakouts perform in line
with their bearish cousins, scoring a 21% average decline.
I used a frequency distribution to compute the most likely gain or loss.
Both types of breakouts have likely gains or losses of about 10%. These results
are typical for many formations and suggest that you have to be ready to pull
the trigger quickly ifyou want to keep any profit.
Shark32 patterns with both breakout types show a receding volume
trend over the 3 days, just as any other symmetrical triangle (sharks are very
short symmetrical triangles).
Some analysts say that sharks appearing symmetrical about the horizon
tal axis perform better than those that are not symmetrical. I found this to be
true generally for both breakout directions. I explore this further in the Statis
tics section later in this chapter.
The narrowing highs and lows of a shark32 pattern remind me of a spring
being wound tighter and tighter. Eventually the spring releases and prices
shoot out of the formation. That is the theory, anyway, but the reality shows
that prices quickly return, causing a failure. More about failures later.
As far as formations go, this one is easy to identify and quite common. I
offer no table of identification guidelines because it is so simple. Locate a 3day
price pattern that looks like a short symmetrical triangle, that is, a chart pattern
having lower highs and higher lows on each of 3 consecutive days. No ties
^ Focus on Failures 503
allowed and the shortest day should have a nonzero price range (that is, the
high and low cannot be the same price).
Figure 37.1 illustrates what I am talking about. Ifyou look closely, the tops
slope downward and the bottoms slope upward, forming a 3day symmetrical
triangle. Each succeeding high is below the prior day's high and each succeed
ing low is above the prior day's low. Together, they form a 3day, triangular
shaped shark fin, hence the name shark32.
The volume trend is nearly always downward. This is sometimes not
clear, but I used linear regression to compute the slope of the line over the 3
day formation. All three chart patterns shown in Figure 37.1, for example,
have downward volume trends.
At first you may be skeptical that the formation on the right has a down
ward volume trend. Think of it this way: If you substitute dots instead of vol
ume bars in die figure then draw a line so that it is equidistant from the three
dots, you will find that the slope of this line is downward (higher on the left).
In essence, this is linear regression, a mathematical way of placing the line
evenly between the dots. The slope ofthe resulting line gives the volume trend.
Focus on Failures
The shark32 formation sports a very high failure rate, so finding failure exam
ples is easy. Figure 37.2 shows a shark formation failure. Regardless of what
you call this formation—a reversal or consolidation of the prevailing trend—
Alien Telecom, Inc. (Telecom. Equipment, NYSE, ALN)
Sep93 Oct Nov Dec
Figure 37.1 Three shark32 patterns. Each has lower highs and higher lows on 3
consecutive days.

504 Shark32
Alza (Drug, NYSE, AZA)
|u!91 Aug Sep Oct
Figure 37.2 A shark32 failure. Prices break out downward, pull back, and move
substantially higher from this shark32. The downward plunge takes prices less
than 1 % lower, a socalled 5% failure.
once prices break out of the formation, they should continue moving in the
same direction without reversing substantially.
First some definitions: A breakout is when prices close outside the for
mation. Since the first day has the widest price range, it is the benchmark to
gauge a breakout. A breakout occurs ifprices close above the first day's high or
below its low. Expect prices to continue moving in the breakout direction.
However, throwbacks to the triangle apex from the top or pullbacks from
the bottom are both permissible. Throwbacks or pullbacks occur over halfthe
time (sometimes both occur, but I only consider to be valid those in the break
out direction). Occasionally, prices throw back or pull back then keep moving.
If a throwback drops below the first day's low or a pullback rises above the first
day's high (and closes there), then the formation is a failure.
Figure 37.2 shows a shark32 pattern in a rising price trend. Since the
shark pattern functions as a consolidation ofthe trend most of the time, we can
expect prices to continue rising after the pattern completes. It does not; prices
reverse and head down.
The apex ofthe triangle, where two imaginary trendlines drawn along the
boundaries of the formation meet on the right, is at a price of 32'/s. Two days
after the formation completes, prices drop to a low of 305
/s and close at the low
for the day. The next day prices move even lower to 307
/i6. The first day of the
formation has a low of 313
/4, so prices have clearly broken out downward (the
close of 305
/8 is below the first day's low of 313
/4). We can expect prices to
v
Statistics 505
continue heading down, but they stop and pull back to the triangle apex. The
decline, as measured from the breakout day low to the ultimate low, is
just 0.6%.
Since the breakout direction is unknown ahead of time, an investor must
wait for prices to close above or below the triangle before taking a position in
the stock. Thus, I use the breakout dayhigh orlow to determine the gain or loss
because that is the first day in which an investor would likely take a position.
You can see in Figure 37.2 that an investor selling short the day after the
breakout would have gotten creamed. Prices pull back to the triangle apex the
next day and 2 days later they move above the top of the triangle. When that
happens, the formation fails. Prices continue up until midJanuary when they
reach a high over 55.
There is a saying that large formations are stronger than smaller ones.
This adage certainly appears to be the case with this formation. It is even more
pronounced after factoring in the high failure rates. The larger symmetrical
triangles have failure rates that are about onetenth the rate of sharks.
Statistics
Table 37.1 shows general statistics for the shark32 pattern. Since chart pat
terns perform differently depending on their breakout direction, I separated
the formation into two categories: upside and downside breakouts. These for
mations are plentiful enough that I only examined 100 stocks over 5 years to
log almost 300 patterns. You can see in the table that most act as consolidations
of the shortterm price trend with downside breakouts having slightly more
reversals.
The failure rate for both breakout directions is just over 40%. This is
double the 20% maximum I consider acceptable, suggesting that you should
Table 37.1
General Statistics for Shark32 Formations
Description
1991 to 1996
Failure rate
Average rise/decline of successful formations
Upside
Breakout
160
111consolidations,
49 reversals
66 or 41%
32%
10% to 15%
1 46 or 91 %
Downside
Breakout
139
63 consolidations,
76 reversals
61 or 44%
21%
Less than 1 0%
119 or 86%

506 Shark32
stay away from this chart pattern. The average rise is 32 % for upside breakouts
and the decline for downside breakouts averages 21%. These statistics are both
worse (40% average gain) and better (20% average loss) for typical bullish and
bearish formations. The frequency distribution in Figure 37.3 shows the most
likely gain or loss for the shark32 pattern. The tallest column is the one with
the highest frequency and the one I consider to be the most likely gain or loss.
For upside breakouts, the most likely gain is about 10% to 15%, with a quar
ter of the formations having gains over 50%. Fifty percent is quite large and is
responsible for boosting the overall average up to 32%. I do not consider the
right column to be representative of what a typical investor can expect to
receive, so I disregard it.
For downside breakouts, the most likely loss is 10%; that is the tallest col
umn on the graph. A third of the formations have losses under 10%, so ifyou
are considering shorting a shark32 pattern with a downside breakout, you
might reconsider.
As mentioned earlier, I performed linear regression on the 3day volume
series and discovered that the slope ofthe regression line is downward over the
course of the formation.
Table 37.2 shows breakoutrelated statistics for the shark32 pattern. The
shark32 pattern differs from a symmetrical triangle in that the breakout for a
Figure 37.3 Frequency distribution of the most likely gain or loss for shark32
chart patterns with upside and downside breakouts. The likely return is less than
10% or 15%.
Table 37.2
Breakout Statistics for Shark32 Formations
Description
Days to breakout
Throwback/pullback
Average time to throwback/pullback
completion
ultimate high/low
near 1 2month low (L), center (C),
or high (H)
Upside Breakouts
8 days
102 or 64%
8 days
6.5 months
(192 days)
L9%, C27%, H64%
L32%, C41%, H31%
Downside Breakouts
8 days
80 or 58%
8 days
2 months
(66 days)
L29%, C35%, H35%
L18%, C21%, H23%
Statistics 507
shark pattern, by definition, comes after the formation completes. For sym
metrical triangles, the breakout usually occurs well before the triangle apex.
For the shark pattern, 8 days typically lapse before prices close above the high
est high or below the lowest low posted during the first day of the pattern (the
widest of the 3 days).
Throwbacks occur when prices return to the triangle apex after an upside
breakout. Throwbacks happen 64% of the time and are too infrequent to base
a trading plan on, but they do give an investor another opportunity to take a
position in the stock. Unfortunately, not all throwbacks (and pullbacks, too)
rebound and continue moving up. Figure 37.2, for example, shows a pullback
that, after returning to the shark pattern, fails to continue moving down again.
Pullbacks occur after a downside breakout when prices return to the triangle
apex, which happens 58% of die time.
The average time for a throwback or pullback to return to the price level
ofthe triangle apex (the center of the shortest day in the shark32 pattern) is 8
days. This is a day or two earlier than most formations. If you miss the initial
breakout, you might have another chance to invest or add to your position
within the coming week.
I determine the ultimate high or low by a significant change in trend, usu
ally 20%, but stop short ifprices return to the shark32 pattern and cross to the
other side.
The number of days to reach the ultimate high or low varies depending
on the breakout direction. For upside breakouts, prices reach the ultimate high
over 6 months later. For downside breakouts, prices reach the ultimate low in
about 2 months. I measure both from the breakout point.
Where in the yearly price range does this formation occur? Most of the
shark32 patterns (64%) with upside breakouts form within a third of the

508 Shark32 w
yearly high. For downside breakouts, the pattern distributes more evenly with
35% appearing within a third of the yearly high or midrange. Mapping per
formance onto the yearly price range, the picture changes somewhat. The best
performing patterns for upside breakouts are those that land in the center third
of the price range. They score an average gain of 41%. For downside break
outs, the best performing formations begin tumbling within a third of the
yearly high; they post losses averaging 23%.
The last statistic concerns symmetry. Shark32 patterns that are symmet
rical about the horizontal axis perform better than those that are not. The fol
lowing formulas measure symmetry:
Symmetry = 0.10 to 0.50
Apex Price = (H + L)/2
Base Height = H[2] L[2]
(A variable number that determines how
symmetrical the pattern needs to be.)
(This is the third or smallest day in the
formation.)
(This is the daily range from the highest to the
lowest during the first day.)
To be symmetrical, a shark32 pattern must satisfy the following equation:
Apex Price < H[2] (Base Height * Symmetry)
Apex Price > L[2] + (Base Height * Symmetry)
In essence, all we are doing is making sure that the middle ofthe narrow
est day is within a given distance from the center of the widest day. The dis
tance I tested ranged from 0.10 to 0.50 in steps of 0.02. A value of 0.50 means
that 50% of the price lies above the center and 50% lies below—perfectly
symmetrical. I ran all successful formations through the various combinations
for both upside and downside breakouts. Those formations with upside break
outs and symmetry values between 0.40 and 0.50 perform better than their
nonsymmetrical counterparts. They have gains that range from 31% to 34%,
whereas their nonsymmetrical counterparts have returns of 30% to 32%. For
downside breakouts, the best performing symmetrical range is narrower at
0.44 to 0.50. The symmetrical sharks have losses ranging between 21% to
25%, whereas the nonsymmetrical sharks have a flat 20% return.
I consider die performance improvement to be marginal and the forma
tion needs to be almost perfectly symmetrical to show any meaningful
improvement. Still, it is an interesting finding.
In case you are wondering what a nonsymmetrical triangle looks like,
look back to Figure 37.1. With a symmetry setting of0.44, only the middle for
mation is symmetrical. The other two have first days in which the low price is
too far down (look at the distance between the two tops and two bottoms ofthe
first and second days—they are uneven and thus asymmetrical).
Trading Tactics 509
Trading Tactics
If your worst enemy tells you this is the formation to trade, ignore him; he is
trying to lead you into bankruptcy. With a failure rate nearly the same as a fair
coin toss, why risk a trade? Since you might ignore my advice and trade this
one anyway (or perhaps you have found a way to make it work), Table 37.3 lists
a few helpful suggestions about the shark32 formation.
If you add up the number of consolidations and the number of reversals
listed in Table 37.1, you will find that the shark32 pattern usually acts as a
consolidation of the prevailing trend. Knowing this, you should anticipate a
breakout in the direction of the shortterm trend. Figure 37.4 shows this type
of behavior. The price trend is moving downward when the formation appears.
After the formation completes and prices drop below the first day's low, they
continue moving down.
The best performing shark32 patterns are symmetrical about the hori
zontal axis, or nearly so. Run the daily high and low prices through the formu
las shown in the Statistics section using symmetry values of 0.44 or higher. If
you are lazy and want to take an easier approach, visually find the midpoint of
the first and third days' price range. If the two points are close to one another,
then the shark32 is symmetrical.
Since you cannot be sure in which direction the breakout will occur (but
lean in the direction of the prevailing trend), always wait for the breakout. I
determine that a breakout occurs when prices either close above the shark's
first day's high or below the first day's low.
Once prices break out, be aware that there is a better than even chance of
a throwback (upside breakouts) or a pullback (downside breakouts). Prices
Table 37.3
Trading Tactics for Shark32 Formations
Profitable suggestion
Trade with the trend
Symmetry
Wait for breakout
Watch for throwback,
pullback
Halfmast formation
Stops
Save your money and do not trade this one.
Since the shark pattern is usually a consolidation, expect a
breakout in the direction of the prevailing price trend.
Choose only near symmetrical patterns.
The breakout direction is unknown, so wait for prices to close
above or below the first day's (the widest of the 3 days) high
or low, then trade with the trend.
More than half the time, a throwback from the top or pullback
from the bottom occurs. Place a position or add to it once
prices resume their original course.
Sometimes the pattern appears midway through a price move.
Place a stoploss order above (downside breakouts) or
below (upside breakouts) the tallest day in the chart pattern.

510 Shark32
Airborne Freight (Air Transport, NYSt, ABF)
Oct91
Figure 37.4 The shark32 pattern sometimes acts as a halfmast formation, mark
ing the midpoint of a move. Here, it bisects the move from points A and B in a con
solidation of the downward trend.
return or come close to the center of the 3day formation (the triangle apex).
When prices return to their original breakout direction, that is the time to
either place a trade or add to it.
Take a look at Figure 37.4. Notice anything peculiar? The formation
apex (at 227
/s) positions neatly between the minor high—point A at 257
/s, and
the minor low—point B at 191
A. The shark32 pattern acts as a midpoint or
halfmast formation. This is useful in trying to gauge the length of the ulti
mate move.
If you walk back into your computer room and find that your young son
has placed a trade in a shark32 formation while you were away, do not panic.
Place a stop at the other end ofthe formation. For example, in Figure 37.4, you
would place a stop at 237
/s, or H above the highest high in die formation (since
the trend is moving down). If the trend was moving up, the stop would be
placed '/g below the lowest low in the shark32 pattern. That way, when 40%
of the formations fail, you will not lose too much money.
38
Triangles, Ascending
RESULTS SNAPSHOT
Appearance
Failure rate
for upside breakout
Average rise
Volume trend
Premature breakouts
Breakout distance to apex
Throwbacks
Percentage meeting
See also
Triangle shape with horizontal top, upsloping
bottom
32%
2%
Downward
25%
63%
58%
89%
HeadandShoulders Tops; Triple Tops
Have you ever heard someone say, "I just happened to be in the right place at
the right time?" Perhaps you have even said it yourself. Investing is a lot like
that—being in the right stock just before it takes off. That is one ofthe reasons
the ascending triangle is one ofmy favorite formations. You can make a bundle
511

ofmoney ifyou trade it properly. But before we get to trading tactics, let us look
more closely at ascending triangles.
The Results Snapshot shows the important findings. The ascending tri
angle has a poor failure rate of 32%. However, ifyou wait for an upside break
out, then the failure rate drops to just 2 %.
For upside breakouts, the average rise is a strong 44%. A frequency dis
tribution of the gains suggests that the most likely gain is 20%. With such a
strong showing, prices fulfill the measure rule 89% of the time. I consider val
ues above 80% to be reliable, so this chart pattern stacks up well.
Tour
Figure 38.1 shows a good example ofan ascending triangle. A horizontal trend
line drawn across the minor highs and an upsloping trendline connecting the
minor lows form the characteristic triangular pattern. Volume diminishes as
prices bounce between resistance at the top and support at the bottom. A pre
mature breakout gives a hint of the coming action; less than 2 weeks later,
prices break out again and move higher.
Why do ascending triangles form? Imagine you are the manager of a
large mutual fund. Over the years your fund has purchased a few hundred
thousand shares of the company shown in Figure 38.1. After seeing the stock
CUC International (Financial Services, NYSE, CU)
Ultimate High
May 93 Oct Nov
Figure 38.1 A good example of an ascending triangle. The horizontal top and
upsloping trendline on the bottom mark the boundaries of this bullish formation.
The premature breakout on high volume is often indistinguishable from the real
breakout. The volume trend is downward until the premature breakout.
rise for almost a year, you are getting nervous about continuing to hold the
stock. You believe the stock is trading well above its fair market value and you
have spotted a more promising situation in another company.
You tell the trading department to dump all your shares as long as it
receives at least 18'/2. For 2 days, starting on June 4, 1993, the trading depart
ment sells shares. Since your fund has a large block of shares to get rid of, the
price cannot climb much above 18'/2 without the fund selling shares and forc
ing prices back down. The selling puts a ceiling on the stock. Word gets
around that you are selling and other institutional investors jump on the band
wagon and sell too. Their aggressive selling satiates demand and the stock
starts declining. It tumbles to a low of 16 <4 on June 9, where buying demand
halts the decline. Buyers, viewing the price of the stock as a steal, demand
more shares. The buying pressure turns the decline around and prices start ris
ing—quickly at first but more slowly as more investors become willing to sell
their shares. When the stock hits IS1
/: again on June 16, your fund sells more
shares, effectively halting the advance. The stock struggles at that level for 3
days. Again, the selling pressure forces prices down and they cross to the other
side of the nowforming ascending triangle. Prices rebound one last time, and
hit the sell zone and stay there for about a week before being turned away by
an excess supply. A call from the trading department confirms that the stock
has been completely sold.
Without an overhanging supply to halt the stock's rise, prices gap up on
increasing demand and soar to 19!
/4. Your fund is out of the picture, but the
forces of supply and demand are not finished with the company. Others still
selling their shares force the stock price back down into the triangle proper.
Prices race to the other side of the triangle, rebound off the lower trendline,
then march back up out the top. Prices dance along the top trendline for a day,
then catapult higher and move up.
IfI had to sum up the price action of an ascending triangle, I would say it
forms because of a supply of shares available at a fixed price. Once the supply
depletes, shares quickly break out of the formation and move higher. If
demand continues to be strong, prices rise. Otherwise, the stock collapses back
on itself and either regroups for another try or continues down.
Finding an ascending triangle in a chart of daily price data is simple, perhaps
too simple. I recently read a tutorial in a popular magazine in which nearly half
the illustrations purporting to be triangles were incorrectly identified. If you
have any doubt about the validity of a chart pattern, others may share those
doubts. If others do not see the same shapes you do, chances are the pattern
will not work as you expect. Under those circumstances, where there is some
doubt about correct identification, do not trade the formation. Save your

money for a trade where you are sure the formation is valid. I discuss identifi
cation problems later in this section.
The triangle pictured in Figure 38.1 is nearly a classic example of an
ascending triangle. The horizontal top line ofresistance repels prices and they
rebound off a steadily rising support line below. The two narrowing lines, one
horizontal and the other sloping up, outline a triangular shape. The ascending
trendline predicts a rise in prices, hence the name ascending triangle.
Table 38.1 lists ascending triangle characteristics. The top horizontal
trendline should have prices that approach and withdraw at least twice (in odier
words, two distinct minor highs). Similarly, the upsloping trendline should be
supported by two distinct minor lows. The two trendlines meet at the triangle
apex, but prices usually break out of the formation well before then.
Table 38.1
Identification Characteristics of Ascending Triangles
Triangle shape
Horizontal top line
Upsloping bottom trendline
Crossing pattern
Volume
Premature breakouts
Upside breakout
Price action after breakout
Two price trendlines, the top one horizontal and the
bottom one sloping up, form a triangle pattern. The two
lines join at the triangle apex.
Prices rise up to and fall away from a horizontal resistance
line at least twice (two minor highs). Prices need not
touch the trendline but should come reasonably close
(say, within'/s). The line need not be completely
horizontal but usually is.
Prices decline to and rise away from an upsloping
trendline. Prices need not touch the trendline but should
come close (within '4). At least two trendline touches
(minor lows) are required.
Prices should cross the chart pattern several times; they
should not leave a vast amount of white space in the
center of the triangle.
Volume is heavier at the start of the formation than near
the end. Volume is usually low just before the breakout.
Somewhat prone to premature breakouts, both up and
down. Volume on a false breakout is also heavy, just as
the genuine breakout.
Volume is heavy (but need not be) and continues to be
heavy for several days.
Once prices pierce the horizontal resistance line
confirming a breakout, prices move up and away from
the formation. Throwbacks to the formation top are
common. If prices continue to climb rapidly, volume will
probably remain high. For downside breakouts, volume is
high at first and usually tapers off unless the price decline
is rapid, in which case volume will probably remain high.
As the triangle forms, volume is heavy at first but tapers off until the day
of the breakout. Often volume is abnormally low a few days before the break
out, as if the formation is gathering strength for the final push. When the
upside breakout comes, volume can rise substantially and usually does, but
heavy volume on a breakout is not a prerequisite.
How can you be sure the breakout is not a premature breakout? You
cannot. A premature breakout is a close outside the boundaries of the two
trendlines. After a few days, prices return to the confines of the triangle and
eventually break out for good by soaring above the top trendline. Volume on
premature breakouts is indistinguishable from normal breakouts and both
occur at about the same distance to the triangle apex.
Once a triangle has a genuine breakout on the upside, what is the behav
ior like? Prices rapidly climb away from the triangle but occasionally throw
back to the top of the formation. Volume is usually heavy, supporting the rise,
and continues to be heavy as momentum gathers speed. Once prices level out,
volume returns to normal. If prices rise over several weeks, the volume pattern
usually appears erratic and heavy when compared to earlier in the year.
What about support and resistance? If you consider the triangle as the
momentary intersection of two trendlines, you can guess where support and
resistance will be. It will be along the two trendlines. Figure 38.2 shows an
example of this on the weekly time scale. Notice the generally downsloping
Gillette Co. (Toiletries/Cosmetics, NYSE, C)
93
Figure 38.2 Ascending triangle on weekly time scale. The price rise generally fol
lows the upsloping support line of the triangle. The numbers count the minor
high and low touches of the trendline.

volume trend from the formation start to the week before the breakout. Vol
ume spikes upward on the breakout then generally declines as prices round
over and approach 1994. Prices start climbing again, essentially hugging the
trendline started by the ascending triangle. The upward trend continues for
several years following the triangleinitiated support line. Although the match
between the sloping trendline and the slope of the later price action is not
exact, the trend is clear. The triangle sports two minor high touches of the top
trendline and three on the bottom, numbered in the figure. Trendline touches
and prices crossing the triangle are important selection criteria.
By now you may feel comfortable with correctly identifying an ascending
triangle. However, there are some situations that may fool investors new to the
formation. Figure 38.3 shows the first one. Cover up the right half of the fig
ure and ask yourself ifwhat you see on the left looks like an ascending triangle.
The horizontal line, arbitrarily drawn to rest on top of the central peak,
extends to the left and right until it intersects prices. Although the lower trend
line has several instances where prices decline to and bounce off of the up
sloping line, the top trendline does not have such a situation. Only in the
center of the formation do prices rise up to and decline away from the hori
zontal trendline. The intersection of prices with the horizontal trendline on
the left coincidentally touches a daily high. Otherwise, the price trend is one
Baker Hughes (Oilfield Svcs./Equipment, NYSE, BH1)
1 2 3
24
15
Nov 92 Dec Jan 93 Feb Mar Apr 92 Dec |an 93 Feb Mar Apr
Figure 38.3 Two views of an incorrectly identified ascending triangle. What looks
like an ascending triangle on the left clearly is not on the right. The three down
ward spikes in December, identified by the numbers near the top of the figure,
mark a headandshoulders bottom with a horizontal neckline, not a triangle.
that has been declining for several days in a row. At the start, there is no minor
high supporting the horizontal trendline.
The same can be said of the formation's right side. There is no minor
high on which the horizontal trendline can rest. Looking at the right side of
the chart. Does this still look like an ascending triangle? I can hear you asking
me to lower the horizontal trendline until it touches the two minor highs in
early to midDecember (below number 1 and midway between numbers 2 and
3). That is not a bad guess, but it is still wrong. What you are really looking at
is a headandshoulders bottom. The left shoulder has a large volume spike
(under number 1). Located under number 2, the head shows a smaller volume
spike. The right shoulder shows volume that recedes even further (number 3).
A true ascending triangle has at least two minor highs forming the top trend
line and at least two minor lows forming the bottom.
Figure 3 8.4 shows another example of a falsely identified ascending tri
angle. This chart has too much white space in the central portion of the trian
gle. A welldefined ascending triangle has prices that bounce from one side of
the formation to the other as it nears the apex. Take a good look at the figure.
This represents one ofthe most common identification mistakes. Novices will
find a rounding bottom and draw a horizontal line across the top and another
tangent to the bottom price action then yell, "Eureka! An ascending triangle!"
Wrong.
Brinker International (Restaurant, NYSE, EAT)
|an96
Figure 38.4 This pattern is not a valid ascending triangle. There are not enough
crossings between the two trendlines to illustrate a valid triangle construction. The
minor highs and lows are numbered.

Tultex Corp. (Apparel, NYSE, TTX)
|un95
Figure 38.5 An excellent example of a correctly constructed ascending triangle.
The number of minor highs and lows is good and there are plenty of crossings
from the top trendline to the bottom. The volume trend is downward too, until the
upside breakout.
Contrast Figure 38.4 with Figure 38.5. In Figure 38.5 notice the number
of times prices move from one side of the triangle to the other. Even though
prices do not rise very far before throwing back to the triangle apex and mov
ing down, it is still a nicely formed ascending triangle. Also note the generally
decreasing volume, especially near the breakout.
Focus on Failures
Figure 38.5 shows the first failure type: a 5% failure. Strictly speaking this is
not a 5% failure (because prices climb by 6%), but it is typical of what one
looks like. A 5% failure is when prices break out and move less than 5% higher
before curling around and moving below the formation low. In this case, prices
leave the formation at 6*4 and reach a high of 6'/2—a 6% move, far enough
away for the chart pattern to be a success. For ascending triangles, 5% failures
are rare, occurring just 2% of the time.
Figure 38.6 shows the more common failure type when prices break out
downward and continue moving down. I have numbered the minor highs and
lows to help with identification. The top trendline is horizontal and the bottom
one slopes up. As you run through the identification characteristics in Table
38.1, there is only one thing that looks odd with this ascending triangle: the
Statistics 519
Tenneco Inc. (Auto Parts (Replacement), NYSE, TEN)
Mar 94 Apr May |un
Figure 38.6 A classic ascending triangle failure. Prices break out downward and
continue moving down. The numbers show the trendline touches.
volume pattern. The volume trend in this chart pattern slopes up. I usually
ignore the volume pattern because I stumble across many instances when I
have a perfectly valid chart pattern but the volume trend is wrong. That is the
case here. I would not throw out the formation just because of the volume
trend, but I would take one additional step before investing: I would wait for
the breakout. If you wait for the breakout, you will pass up losing opportuni
ties such as this one. You will learn in the Statistics section that a third of the
ascending triangles break out downward.
Statistics
Table 38.2 shows the general statistics for ascending triangles. There are fewer
ascending triangles than I expected in the 500 stocks I examined—725. That is
less than two each over the 5year period. Several stocks have from three to six
triangles, whereas many others have none. Still, the number oftriangles exam
ined allows their statistical performance to become clear. The vast majority
(529) of ascending triangles act as consolidations of the prevailing trend. The
other 196 are reversals.
Do ascending triangles ascend? Yes. More than two out of three (68%)
have a meaningful rise after an upside breakout. However, if one waits for a
breakout (a dose above the horizontal trendline), the failure rate improves signif
icantly. Even if an investor buys on a premature upside breakout, the likelihood

Table 38.2
General Statistics for Ascending Triangles
to 1996
Failure rate
(5% failure)
Average decline of failed formations
Most likely rise
Breakout distance to apex
Start to breakout
725
230 or 32%
13 or 2%
44%
21%
20%
63%
439 or 89%
2 months (64 days)
179*
40 50 60
Percentage Cain
Figure 38.7 Frequency distribution of gains for ascending triangles. The most
likely gain is 20%, but the rightmost column skews the average gain upward.
ofmaking a successful trade rises from 68% to 85%.Just 2% ofthe upside break
outs curl around before soaring less than 5% (the 5% failure rate). The implica
tions ofthis are obvious: Wait for an upside breakout from an ascending triangle
before buying the stock.
Those triangles that perform as expected show an average rise of 44%,
whereas those that fail tumble by an average of 21 %. I did a frequency distrib
ution of the gains to better judge what the gain would be for a typical stock.
Figure 38.7 shows that most of the gains occur in the 11% to 20% range.
Almost half the formations (49%) have gains less than 30%. The large column
on the right pulls the average upward.
I plotted the average price rise versus the distance to the apex. I wanted to
verify the claim that the most powerful breakouts occur twothirds of the way
to the triangle apex. Figure 38.8 shows the results. The chart resembles a bell
shaped curve tilted on its side, suggesting that the further away from the aver
age breakout of 63%, the weaker the breakout becomes.
One surprise with ascending triangles is the high number that meet the
measure rule (89%). I discuss the measure rule further in Trading Tactics but
it describes a way to predict the minimum price move. You simply compute
the height of the triangle at the start of the formation and add the result to
the price level of the horizontal trendline. The result is the expected mini
mum price rise. Almost 9 out of 10 ascending triangles hit their price targets.
However, the 89% value is misleading. The average height of the formations
in the database is just 10% of the stock price. This means the predicted price
target is only 10% higher than the breakout price, so it is relatively easy to
400
Price Rise c
.
Figure 38.8 Scatter diagram of average price rise versus distance to apex. The
most powerful breakouts occur twothirds of the way to the triangle apex.
521

meet the target. Perhaps the biggest surprise of all is that the 89% figure is
not higher.
Another unexpected discovery is the relatively high number ofpremature
breakouts, shown in Table 38.3. One ofevery four formations has either a pre
mature upside or downside breakout (or both). There are slightly more pre
mature upside breakouts (22%) than downside ones (12%). The volume is
above average, at 175% and 106% of the 25day volume moving average of
upside and downside premature breakouts, respectively. The above average
volume makes a premature breakout look just like a genuine breakout and they
occur about the same distance from the apex as normal breakouts (60% and
63%, respectively). I noticed a tendency for a premature breakout to echo on
the other side of the formation with another premature breakout. Just over a
third (36%) ofthe formations that have premature breakouts have both up and
down premature breakouts.
Do premature breakouts predict the direction of the genuine breakout? If
you exclude horizontal breakouts, premature upside breakouts occur with gen
uine upside ones 40% of the time, and downside breakouts 46% of the time.
Likewise, downside premature breakouts pair with genuine upside and down
side breakouts 42% of the time.
What does all this mean? Nothing. When you have a premature break
out, either up or down, the final breakout can still go either way. That is the
Table 38.3
Premature Breakout Statistics for Ascending Triangles
Number of premature breakouts (up or down) 184 or 25%
Number of premature upside breakouts 163 or 22%
Number of premature downside breakouts 88 or 12%
Volume at upside premature breakout as percentage of
25day moving average 175%
Volume at downside premature breakout as percentage
of 25day moving average 106%
Upside premature breakout distance to apex 60%
Downside premature breakout distance to apex 63%
Number of upside and downside premature breakouts 67 or 36%
Premature breakout up, genuine breakout up 65 or 40%
Premature breakout up, genuine breakout down 75 or 46%
Premature breakout down, genuine breakout up 37 or 42%
Premature breakout down, genuine breakout down 37 or 42%
Note; Premature breakouts are indistinguishable from real breakouts and offer no clue as
to the direction of the genuine breakout.
Statistics 523
conclusion I reached after reviewing the statistics shown in Table 38.3 and sev
eral others not included. Premature breakouts do not predict the final breakout
direction or success orfailure ofthe formation.
Putting premature breakouts aside, what about the genuine breakout?
Both upside and downside breakouts occur nearly die same distance to the
apex, 63% and 68%, respectively (Table 38.4). In only 13 instances does an
upside breakout result in failure (a 5% failure, from Table 38.2). Five per
cent of the downside breakouts move down by less than 5%, curl around, and
soar upward. The overall result means that if the formation has an upside
breakout, it will continue up. If it has a downside breakout, it likely will con
tinue down.
Almost two out of three times (62%) an ascending triangle will break out
upward and 33% of the time it will breakout downward. Those statistics are
not as good as other formations, but there are ways to improve on the win/loss
record as we see in Trading Tactics.
Throwbacks occur when prices break out upward and quickly return to
the formation. Fullbacks are similar except that the breakout is downward and
prices rise back to the upsloping trendline. Throwbacks at 58% of formations
with upward breakouts are more prevalent than pullbacks at 49% offormations
with downward breakouts.
The statistics reveal that, on average, throwbacks and pullbacks complete
in about 2 weeks (15 and 13 days, respectively). Since this is the average, prices
generally start returning to the triangle before this period. For successful for
mations, prices reach the ultimate high in 6 months (186 days), on average. A
frequency distribution of the time between an upside breakout and the ulti
Table 38.4
Breakout Statistics for Ascending Triangles
Upside breakout distance to apex
Downside breakout distance to apex
Downside breakout but success
Upside breakout
Downside breakout
Horizontal breakout
Throwback
Fullback
63%
68%
3 3 or 5%
450 or 62%
238 or 33%
3 7 or 5%
262 or 58%
117 or 49%
15 days
13 days
6 months (186 days)
Note: If the formation breaks out upward, it continues to rise.

mate high is somewhat puzzling. The majority ofstocks (224 or 45%) take less
than 3 months to reach their ultimate high followed by 96 (or 19%) for inter
mediateterm moves and 35% (175) for durations over 6 months. Although I
consider an ascending triangle to have shortterm investment implications, a
stock could have a price move lasting much longer.
Table 38.5 outlines a study ofvolume for ascending triangles. Using lin
ear regression, I examined the volume characteristics from formation start to
the day before the breakout. In 68% of the cases, the volume trends downward
over the period. Linear regression is a fancy way of computing a line that best
fits the points such that the distance from each point to the line is minimal.
Specifically, I used the slope of the linear regression line to determine the vol
ume trend.
Table 38.5 shows a comparison of the average breakout volume with the
25day moving average of the volume ending the day before the breakout (I did
not want the actual breakout volume influencing the moving average). Notice
that the Any after an upside breakout is the highest volume day (176% or 76%
above the moving average). It suggests that once people discover an upside
breakout, they buy the stock, forcing the volume to spike the day after the
breakout. From that point on, volume recedes but is still comparatively high a
week later.
Some analysts say throwbacks are more likely after a low volume break
out. I checked this and found that it is not true. A throwback occurs after a high
volume breakout 54% ofthe time and after a low volume breakout 46% ofthe
time. It is more accurate to say that a throwback is more likely to occur after a
high volume breakout.
Do high volume breakouts push prices higher? Not really. When the
breakout volume is over 50% above average, prices rise by 42%. When the
breakout volume is less than 50% below the 25day moving average, a 43% rise
results.
Table 38.5
Volume Statistics for Ascending Triangles
Volume for breakout day and next 5 days compared with
Percentage of throwbacks after high volume breakout versus
low volume breakouts
Performance of high volume breakouts versus low volume
breakouts
495 or 68%
130%, 176%, 151%,
144%, 128%, 121%
54% versus 46%
42% versus 43%
Note; The trend of volume is downward until the breakout day when it spikes upward.
Trading Tactics 525
Trading Tactics
Now that you can identify ascending triangles and know their behavior, how
do you trade them? Before I give an example of a trade, I discuss trading tac
tics and the measure rule (see Table 38.6).
The shape of the ascending triangle suggests prices will rise, but how far?
If you compute the height of the formation and add the result to the price of
the horizontal trendline, the result is the minimum predicted price. This is
called the measure rule.
An example makes the calculation clear. Consider the stock shown in Fig
ure 38.9. Calculate the height of the formation by subtracting the low (14 at
the sloping trendline) from the high (175
/s denoted by the horizontal trendline)
at the formation start. The difference is 35
/g. Add the result to the highest
high—the value of the horizontal trendline—and you get a target price of2 iVi.
Prices reach the target on July 16, 1992, when they climb to a high of 2lYs,
about 6 weeks after the upside breakout.
A more visual and conservative approach is to draw a line from the start
of the formation (the top left corner) parallel to the upsloping trendline. The
value of the line the day prices break out of the formation becomes the target
Table 38.6
Trading Tactics for Ascending Triangles
Measure rule
Sell on measure rule
Sell on downside breakout
Short sales
Compute the height of the formation at the start of the
triangle. Add the result to the price of the horizontal
trendline. The sum is the minimum price target.
Buy the stock the day after a breakout (when prices close
above the top trendline by at least %). If you miss it, hope
for a throwback then buy when prices resume the
breakout direction after the throwback completes.
For shortterm traders, sell when prices near the target
price (see measure rule) or when prices pierce a support
trendline. For intermediate and longterm traders, hold
the stock until fundamentals or market conditions change.
If you own the stock and it breaks out downward, sell. If
you do not own it, sell it short. Should the stock pull
back, that is another opportunity to sell, sell short, or add
to your short position.
If you short the stock and an ascending triangle appears,
you have a one in three chance that it will break out
downward. Cover the short immediately if it breaks out
upward.
Note: Buy the stock after a confirmed breakout (the stock closes above the triangle top)
and sell when it nears the target price.

526 Triangles, Ascending Sample Trade 527
Jan 92 Feb Mar Apr May )un jul Aug Sep
Figure 38.9 Measure rule applied to ascending triangles. There are two ways to
predict the price move of an ascending triangle. Compute the formation height by
subtracting the low from the high at the start of the formation (denoted by the
two circles). Add the result to the price marked by the top trendline. The combi
nation is the price to which the stock will climb, at a minimum. Alternatively, draw
a line parallel to the upsloping trendline beginning with the left top corner of the
formation. At the point where prices break out of the formation, the price level of
the line becomes the target price.
price. The figure shows the new line. Be careful when determining where the
formation begins since tagging the beginning of the formation too soon will
cause an abnormally high price target.
Since a third of the formations break out downward, you must wait for an
upside breakout before investing. Once prices close above the top trendline,
buy the stock. Although you will be buying at a higher price, the chances of
having a failure are small (2% versus 32% ifyou do not wait).
Once prices rise, use the measure rule to estimate gains. Since die mea
sure rule is not perfect, be ready to take profits once prices near the target. Use
past resistance zones to finetune the prediction.
If prices break out of the triangle downward, then sell your holdings
immediately. This is also a time to go short. Look for prices to drop up to 20%.
If a pullback occurs, wait for prices to resume their downward direction then
add to your short position. Close out the trade if the fundamentals improve or
if prices pause at a support zone.
Sample Trade
Dan is an investor with a few years ofexperience. He is new to technical analy
sis and discovered ascending triangles by accident. After doing some research
to familiarize himselfwith the formation, he found that if he delayed buying a
stock until after a breakout, he would increase his chances ofsuccess. However,
he would also give up part of his gains as the fastest portion of the rise occurs
at the start. That was a tradeoffhe was willing to make.
Dan took an interest in the company shown in Figure 38.10 when he
noticed an ascending triangle forming in the stock. He believed that the break
out was nearly at hand when volume suddenly sank to 2 3,400 shares on August
19. Two days later, on higher volume, prices crossed the triangle and peaked
out the top. For the next few days, prices balanced themselves on the top hor
izontal trendline and waited for demand to send them higher. The decisive
breakout occurred on August 26, even though volume was tepid. Dan grabbed
his calculator and computed the breakout distance to the apex and discovered
that the breakout occurred at the 70% mark. This signaled a potentially strong
breakout.
Fastenal Company (Retail Building Supply, NASDAQ, FAST)
|un94 Nov
Figure 38.10 Trading an ascending triangle. Dan bought 500 shares of the stock
at point 1 after the stock threw back to the formation. He sold it at point 2, the day
after the stock hit the price target of 221
/2. Note the downsloping volume trend
during creation of the formation and the two support lines parallel to the two tri
angle borders.

However, volume told a different story. Although volume had been
steadily receding throughout the formation as one would expect, there was not
enthusiastic volume on the breakout. With this stealthy signal, Dan decided to
wait before buying the stock.
Believing that a profitable opportunity was at hand, he computed the tar
get price to see if it afforded a profitable move. At the formation start, the hor
izontal trendline marked the high for the stock at 19i4. At the same point, the
upsloping trendline marked the low. At the start of the formation, the lower
trendline was at 163
/8. This predicted nearly a threepoint climb, or a 15%
move from the 19'/4 launch price. To Dan, the small move was not terribly
exciting, but it was much better than the interest rates the banks were paying.
Two days after the breakout, the stock started declining and returned to
the top ofthe formation. That is when Dan pulled the trigger and bought 500
shares at the high for the day, 19l
/2 on September 2, 1994. Immediately, he
placed a stoploss order to sell the stock should it decline below the lowest low
of the formation. The formation low occurred on July 25, 1994, at a price of
163
/4. He told his broker to sell the stock I
/B below this, or 165
/g. That would
limit his loss to a steep 15%, but it was also slightly below the nearest support
level (the bottom of which was also at 163
/4). He reasoned that there was a
decent chance that ifthe stock declined, growing demand would repulse prices
and not trigger his stop.
Then he waited and watched the stock. It peaked at 21 '/4 on September 26
before leveling off and heading back down. Since the stock was not near the
price target of 22 H, Dan decided to hold on. The stock continued sinking
until it found support at the horizontal triangle trendline at 19 on October 5.
At that point, the stock started moving up again. On Halloween, the stock
reached his price target by hitting a daily high of 23. He decided to sell the
stock the next day and received a fill at 22l
/2.
Dan evaluated his results and reviewed the trade. He had a net gain of
$1,450 or almost $3 a share. That is a 15% gain in 2 months or almost a dou
bling of his money if he kept up the performance for the entire year. He also
decided that he was lucky as he sold near the top. When the stock returned to
the support level in early October, it could have continued down. He decided
that once a stock rises by 10%, he should raise his sell stop to breakeven even
though, in this case, it would have cashed him out prematurely.
39
Triangles, Descending
RESULTS SNAPSHOT
Appearance
Failure rate
Failure rate if waited for
downside breakout
Average decline
Premature breakout
Breakout distance
to apex
Fullbacks
Percentage meeting
See also
Triangle shape with horizontal bottom and down
sloping top
45%
4%
19%, with mostlikelydecline between 10% and 20%
Downward
22%
69%
64%
67%
The Results Snapshot shows performance results for descending triangles. The
failure rate at 45% is well above the 20% cutoff for reliable formations. How
ever, ifyou wait for a downside breakout, then the failure rate drops to just 4%.
529

The average decline, at 19%, is about what you would expect from a bearish
formation. The most likely decline, at 10% to 20%, is evenly distributed across
the range. Premature breakouts occur in nearly a quarter of the formations
(22%), so that is something to watch out for.
Tour
Figure 39.1 shows a descending triangle that is typical in many respects. Prices
rise to meet a downsloping trendline on the top of the pattern and fall back.
Then, they rebound off a horizontal trendline along the base ofthe formation.
The volume pattern is unusual for a descending triangle. Normally, volume
recedes as the breakout approaches, but this one appears to have a Vshaped
trend—higher at the beginning and end and weaker in the center. The break
out is downward and occurs on low volume. A bearish breakout can have high
or low volume but volume is usually heavy. After the breakout, prices pull back
to the triangle boundary before continuing down.
Why do descending triangles form? The descending triangle shown in
Figure 39.1 begins forming in October 1994 as part of a consolidation in a
downward trend. Imagine you believe the fair value of this stock is 73
/s but is
overvalued at prices much above that. You tell your broker to buy the stock
Fllene's Basement Corp (Retail (Special Lines), NASDAQ, BSMT)
Sep94 Oct tan 95
Figure 39.1 A nicely formed descending triangle with unusual volume pattern.
Typically, volume trends downward and is quite low just before the breakout. Also
shown is a pullback, repulsed by the horizontal resistance level.
should it fall to 73
/8. After reaching a minor high at 83
/s on October 11, the
stock begins declining for a few days. It descends and reaches the buy price 2
days later. Your broker buys the stock.
You are not alone. Other investors, believing the stock is retesting the low
that occurred a week earlier, also buy the stock. Together, the buying puts a
momentary floor on the stock. For the next 2 days, the stock returns to the 73
/s
level before buying demand pushes the price higher. This time the stock does
not climb as high as the prior minor high; it only reaches a value of 8:
/8 before
turning down. Again, when the stock reaches a low of 73
/8, buying demand
increases enough to halt the decline at that level and to send the stock moving
back up. During the next 2 weeks or so, you and other investors buy the stock.
Enthusiasm for the stock quickly wanes and a series of lower highs outline a
downsloping trend. The floor, at 7%, becomes the horizontal support level.
Eventually, investors buy enough of the stock and have either run out of
money to buy more or decide they already own enough. The stock slips below
the support line on November 9, and closes at the low for the day at 7l
/s. The
stock hovers near that price for a few more days before continuing down in
earnest on higher volume.
Quickfooted investors, realizing that the floor is no longer holding firm,
sell the stock. The price begins declining rapidly now but soon levels off. For
a few days, selling pressure meets buying demand and the decline halts, turns
around, and begins moving up. It nears the base of the triangle and the smart
money quickly disposes of any remaining shares in their portfolios. The pull
back completes and the stock rounds over and starts heading down again. In 3
months' time the stock reaches the ultimate low ofjust under 3 before leveling
out. That is a decline of 60%.
Descending triangles have distinctive chart patterns making them easy to iden
tify. Consider the triangles shown in Figure 39.2. A descending triangle appears
during March and April 1993 and marks the end of a long rise started in late
1992. On above average volume, the stock moves up in early March then
quickly rounds over and heads down. It declines to a low of about 29'/2 where it
finds support. Prices bounce back up again, not carrying as high this time, then
return to the support level. As April dawns, the stock bounces one last time
before falling through the support line and heading down on high volume.
Like a ball bouncing along the floor, each bounce from the support line
is less high than the previous bounce, giving the formation a downsloping
appearance along the top. The support region at 29'/2 is flat. These two ingre
dients, a down sloping trendline on the top and a horizontal support line on the
bottom, are the two main characteristics of descending triangles. A receding
volume pattern throughout the formation rounds out the picture.

532 Triangles, Descending Identification Guidelines 533
Figure 39.2 Two descending triangles. The March triangle forms after a long
climb beginning in late 1992. The nicely formed chart pattern has a receding vol
ume trend especially in the latter half of the formation. The July formation is a fail
ure since it does not immediately descend as expected. Nearly half of descending
triangles break out upward.
The July formation is also a descending triangle although not as well
formed. The volume pattern rises through the first half of the formation before
moving downward toward the triangle apex. Prices momentarily move down
out of the formation on August 3 and stay below the horizontal support line for
2 more days. Then, prices start rising. They sail up through the base ofthe for
mation and shoot out the top, reaching a peak of 30'/s in late August. Prices
start moving down, slowly at first, then plunge down on exceedingly high vol
ume. The stock declines to a low of 23u
/ie on December 21, 1993, a decline of
about 15%. Using the same ultimate low point, the first triangle shows a
decline of almost 20% from its horizontal support line.
I am sure that ifyou owned stock in this company and sold during either
of the descending triangles, you would be pleased. Although the second for
mation is a failure because it rises above the triangle top, prices do start down
within the month. Sometimes failed formations prematurely alert you to a
trend change, as theJuly example shows.
Table 39.1 outlines the identification characteristics for descending tri
angles. The triangularshaped appearance makes the descending triangle easy
to identify. Prices rebound from the base of the formation following a hori
zontal trendline, whereas prices along the top obey a downwardsloping trend.
Volume throughout the formation also follows a downward trend especially as
Table 39.1
Identification Characteristics of Descending Triangles
Triangle shape
Horizontal bottom support line
Downsloping top trendline
Volume
Premature breakouts
Downside breakouts
Price action after breakout
A triangularshaped pattern bounded by two trendlines,
the bottom one horizontal and the top one sloping
down, that intersect at the triangle apex.
A horizontal (or nearly so) base acts to support prices.
Prices should touch the base at least twice (at least two
minor lows that either touch or come close to the
trendline).
A downsloping price trend that eventually intersects
the horizontal base line at the apex. Prices should rise
up and touch (or come close to) the sloping trendline
at least twice, forming two distinct minor highs.
Volume recedes and tends to drop off just before the
breakout.
Are rare but occur on high volume making them appear
like genuine breakouts.
Usually occur on very high volume that diminishes over
time. However, prices can also break out on low volume.
Prices usually move down quickly, reaching the ultimate
low in a straightline fashion. Fullbacks occur about
twothirds of the time.
it nears the breakout day. Volume on that day is typically very high when com
pared to the prior day. After a breakout, prices drop away from the formation
quickly and reach their ultimate low rapidly. Since the impact of a descending
triangle is short to intermediate term, prices soon recover.
Support and resistance appear along the two trendlines. Throwbacks to
the top of the formation usually stop at the sloping trendline, whereas pull
backs to the bottom halt at the horizontal trendline. After a breakout, prices
often follow the sloping trendline down. During the recovery process after a
descending triangle, prices rise to meet the level of the horizontal trendline
then pause. Sometimes it takes several tries before prices push up through the
horizontal resistance line.
Triangles, as a group, are easy to spot. However, there are some situations
that dictate a careful approach. Figure 39.3 shows such a case. It is an example
ofwhat looks like a descending triangle, but is not. The volume trend does not
conform to the usual pattern for a descending triangle. Volume should recede
as the breakout nears. In Figure 39.3 volume rises alongwith prices at the start
of the triangle, then tapers off at the top when prices round over. However,
volume climbs as prices descend then shoots up the day after the breakout.
Comparing the volume at the start and end of the formation, you can see that
the trend—although somewhat downward—does not recede over time. It is
more of a bowl shape.

Crompton and Knowles Co. (Chemical (Specialty), NYSE, CNK)
12
Apr May |un Sep Oct Nov
Figure 39.3 An invalid descending triangle. There is only one minor high leaving
too much white space in the center of the formation. Prices should cross from side
to side several times forming at least two minor highs and lows.
Volume for many formations is not a crucial factor, and you should not
attach too much significance to it. However, a volume pattern that is not char
acteristic for a chart pattern raises a warning flag. Coupled with other factors,
it might cause you to bypass the stock and look elsewhere for a more promis
ing situation.
The price picture is even worse. Only one minor high composes the
entire triangle. Wellformed descending triangles have prices that cross from
side to side several times. There is no massive amount of white space in the
center of the triangle. Contrast Figure 39.3 with Figure 39.1.
Focus on Failures
A 45% failure rate is shameful, but that is what descending triangles have. Fig
ure 39.4 shows an example ofa failure. Perhaps the first thing you notice is that
prices rise; they shoot out the top of the triangle. Why? I can only speculate an
answer to the question but Figure 39.4 does provide some clues. First, the tri
angle appears in a rising price trend. In almost two out of three cases, prices
continue in the direction ofthe existing trend. In this case, that trend is upward.
Figure 39.4 shows a premature upside breakout in the beginning of Feb
ruary. It may sound silly, but that is a strong clue that prices will continue
Statistics 535
Bane One Corp (Bank, NYSE, ONE)
Oct 92 May |un
Figure 39.4 A descending triangle failure. Since descending triangles act as con
solidations of the trend, this one breaks out upward.
upward. If you bought the stock the day after the breakout, you may question
the veracity of that comment. A day later, prices start a throwback that does not
stop at the top trendline. Prices continue down and almost touch the lower
trendline before resuming their uphill trend.
Still, waiting for a downside breakout before shorting a stock or selling an
existing holding is always a wise course with these formations. Imagine that
you sold your holdings once you recognized the chart pattern as a descending
triangle. You would have missed out 011 the rise to 40—a 20% mistake.
The volume pattern for this formation is difficult to decipher. As prices
rise, so does volume; as prices descend, volume recedes too. Ifyou were to run
the volume data through a linear regression formula, you would find that the
slope of the resulting line tilts downward. In other words, volume recedes, just
as in a wellbehaved descending triangle.
Statistics
Table 39.2 shows general statistics for descending triangles. Fewer valid
descending triangles appear in the database than do ascending triangles.
Almost two out of three (422 out of 689) are consolidations of the current
trend. This simply means that ifthe price trend is downward going into the tri
angle, it is still moving downward after leaving it.

Table 39.2
General Statistics for Descending Triangles
from 1991 to 1996
Failure rate
Failure rate if waited for downside
breakout
Average rise of failed formations
Most likely decline
Start to breakout
Start to apex
689
309 or 45%
27 or 4%
19%
42%
10% to 20%
256 or 67%
2 months (61 days)
3 months (87 days)
Note: Only about half the descending triangles work as expected and twothirds of those
reach their predicted price targets.
25 30 35
Percentage Decline
Figure 39.5 Frequency distribution of declines after a downside breakout from a
descending triangle. There are relatively few large declines that distort the average
decline. The most likely decline is in the 10% to 20% range.
In theory, descending triangles are wonderful formations because the top
trendline predicts the breakout direction: downward. However, only 55% of
the descending triangles work in this fashion. That is a little better than a coin
toss and certainly not good enough on which to base a trade. However, ifyou
wait for the breakout to occur, your chances of success rise to 96%. Unfortu
nately, the downside move is not terribly exciting at 19%, just a little behind
other bearish formations that typically show a decline of 20%. To derive the
most likely decline, I did a frequency distribution by sorting the percentage
decline for each formation and grouping the values into 10 bins. Figure 39.5
makes it clear where most of the declines occur: the 10% to 20% range. That
is what I call the most likely decline.
Descending triangles with upside breakouts do quite well, soaring 42%
above the breakout price. The numbers suggest that you should trade with the
trend. If prices break out upward, go long. If your triangle has a downside
breakout, then short the stock.
For those stocks with descending triangles performing as expected, 67%
meet or exceed their price targets. That is to say they decline below the pre
dicted price. I consider values above 80% to be reliable, so descending triangles
fall short.
On average, it takes about 2 months (61 days) before the triangles have a
breakout. The overall formation length from start to the triangle apex where
the two trendlines meet is 87 days.
Table 39.3 shows the statistics for premature breakouts. Only 22% ofthe
formations have premature breakouts in either direction. A premature break
Table 39.3
Premature Breakout Statistics for Descending Triangles
Description
Number of premature breakouts (up or down)
Number of premature upside breakouts
Number of premature downside breakouts
Volume at upside premature breakout versus 25day
moving average
Volume at downside premature breakout versus 25day
moving average
Number of upside and downside premature breakouts
Premature breakout up, genuine breakout up
Premature breakout up, genuine breakout down
Premature breakout down, genuine breakout up
Premature breakout down, genuine breakout down
Upside premature breakout distance to apex
Downside premature breakout distance to apex
Statistic
151 or 22%
38 or 6%
137 or 20%
127%
143%
24 or 1 6%
15 or 39%
12 or 32%
72 or 5 3%
42 or 31%
71%
64%
Note: Premature breakouts are indistinguishable from real breakouts and offer no clue as
to the direction of the final breakout.
537

out is when prices close outside the formation boundary but quickly return
within a few days (they should not venture very far, either). A genuine break
out soars outside the formation and usually continues in the breakout direc
tion. A premature breakout is just a few days long, whereas a throwback or
pullback often takes over a week before prices near the trendline again.
Upside premature breakouts are exceedingly rare, occurring only 6% of
the time on above average volume. Downside premature breakouts are more
likely at 20%, and display volume that is 43% above the average (or 143% of
the total).
Of the formations having premature breakouts, only 16% have both
upside and downside premature breakouts. The next four lines in the table try
to determine if there is a relationship between premature breakouts and the
direction of the final breakout. About half the formations (53%) with prema
ture downside breakouts later break out upward. The other variations have
substantially fewer hits. The table does not show the few remaining premature
breakouts associated with horizontal breakouts.
The last two table entries try to determine if there is a way to eliminate
premature breakouts by knowing that they break out sooner than genuine
breakouts. The answer is no. Although premature downside breakouts occur,
on average, about 64% of the way to the apex, it is close enough to the 69%
genuine breakout distance that you will not be able to tell the difference. Since
the volume pattern is also the same for premature and genuine breakouts, there
is no way to differentiate a premature breakout from a genuine one.
Where do the most powerful breakouts occur? I graphed the distance to
the apex versus the percentage decline but the graph shows a random relation
ship. However, some analysts have suggested that the most powerful breakouts
occur about twothirds of the way to the triangle apex.
Table 39.4 shows the statistics for genuine breakouts. Upside breakouts
occur at nearly the same location as downside breakouts: 70% and 69%,
respectively. Only 4% of the formations have downside breakouts that fail to
continue falling, and a similar number of formations that break out upside
continue down. To put this another way, prices continue moving in the direc
tion of the genuine breakout. If the breakout is down, for example, prices con
tinue dropping.
Upside breakouts occur 41% of the time, downside breakouts 54% ofthe
time, and the remainder are horizontal—prices run flat and pierce the apex.
Since descending triangles are supposed to descend, these statistics are alarm
ing. The 54% downside breakout value is little better than the flip ofa coin and
warns you not to anticipate the breakout direction.
For formations with upside breakouts, throwbacks occur 39% of the time.
Fullbacks from downside breakouts do much better, having a 64% rate. The
pullback rate is a little too low to depend on it while trading; in other words,
do not depend on a pullback before shorting the stock.
Statistics 539
Table 39.4
Breakout Statistics for Descending Triangles
Upside breakout distance to apex
Downside breakout distance to apex
Upside breakout but success
Upside breakout
Downside breakout
Horizontal breakout
Throwback
Pullback
Percentage of breakouts occurring near 12month low (L),
70%
27 or 4%
19 or 3%
283 or 41%
370 or 54%
36 or 5%
109 or 39%
236 or 64%
12 days
12 days
2 months (66 days)
L33%, C32%, H35%
, C17%, H17%
Note: Prices continue moving in the direction of the breakout, but do not try to anticipate
the breakout direction.
Both throwbacks and pullbacks complete in 12 days, on average, after the
breakout. I arrive at this value by removing any throwback or pullback that
occurs more than 30 days after the breakout. Prices returning to the apex after
a month are due to normal price action, not a throwback or pullback.
On average, it takes slightly over 2 months (66 days) to reach the ultimate
low after a downside breakout. A frequency distribution of the results shows
that the majority (72%) reach the ultimate low within 3 months, qualifying the
formations as having shortterm investment implications.
Where in the yearly price range do breakouts occur? Most downside
breakouts happen when prices are within a third of the yearly high, but the
other twothirds are close. Mapping performance over the yearly price range
shows that the best performing triangles have breakouts near the yearly low,
scoring declines of21 %. The other two ranges perform less well. The number
suggests that stocks having trouble (they are near the yearly low), continue to
do poorly. In other words, do not short a stock making new highs; instead, look
at those making new lows.
The last statistics table (Table 39.5) concerns volume. The volume trends
downward for most descending triangles (72%) as measured using the slope of
the linear regression line from the triangle start to the day before the breakout.
I compared the breakout volume with a 25day moving average of the volume.
The table lists the results. The heaviest volume occurs the day after a breakout,

Table 39.5
Volume Statistics for Descending Triangles
with 25day moving average
Percentage of high volume breakouts subject to
pullbacks versus low volume breakouts
Performance of high volume downside breakouts
versus low volume downside breakouts
494 or 72%
158%, 168%, 136%,
136%, 135%, 123%'
56% versus
18% versus 17%
Note: The trend of volume is downward until the breakout day when it spikes upward.
suggesting that endofday traders notice the breakout then jump on the trend
the following day. Volume remains heavy throughout the next week.
Do pullbacks occur after low volume breakouts? No. I used 50% above
and below the 25day volume moving average as the benchmark for high and
low volume, respectively. Then, I sorted the pullbacks according to the break
out volume. Most pullbacks (56%), occur after a high volume breakout, not a
low volume one.
Lastly, I determined that there is no significant relationship between high
volume downside breakouts and large price moves. To check this, I separated
downside breakouts with high volume and their corresponding price moves
from their low volume counterparts. The average decline for high volume
breakouts is 18% and for low volume breakouts, 17%. So, do not get too
excited if you have a high volume downside breakout; it does not mean that
your stock will fall any further than a low volume breakout.
Trading Tactics
Table 39.6 shows trading tactics for descending triangles and it begins with the
measure rule. As you would expect, the measure rule tries to predict the value
to which prices decline after a downside breakout. Compute the height of the
formation by subtracting the price of the lower trendline from the upper one
at the formation start. Then, subtract the height from the value of the lower
horizontal trendline. The result is the target price.
Compute the height of the triangle shown in Figure 39.6 by taking the
difference between the two trendlines (marked by the black dots). The value is
l7
/s (that is, 9'/4 73
/8). Subtract the height from the value of the horizontal
trendline, or 73
/8 l7
/8, giving a predicted price decline to 5'/2. Prices re

Encyclopedia-of-Chart-Patterns_PDFdrive.pdf

Encyclopedia-of-Chart-Patterns_PDFdrive.pdf

More Related Content

What's hot (8)

Similar to Encyclopedia-of-Chart-Patterns_PDFdrive.pdf (20)

Recently uploaded (20)

Encyclopedia-of-Chart-Patterns_PDFdrive.pdf