Quant insti webinar on algorithmic trading for technocrats!

Quantitative Trading For
Engineers
Gaurav Raizada
Quantinsti
© Copyright 2010-2014 QuantInsti Quantitative Learning Private Limited

What is it exactly?
Base Salary
+ Bonus
Quant Trading
Flexi
Timings
Objective
Evaluation
Flat
Hierarchy
Trading
through
Computers
Cutting
Edge
Technology

Why Quant Trading
Programming
Trading
Quant
Trading
Implementing
Ideas
Direct Approach to
Making Money

Bazaar – Since Ever
• Participants are Producers, Consumers
• Mix of Barter & Coinage
• Trading Roles – Hedgers, Traders, Arbitrageur
• Speed of information travelled at the speed of Horse/Bullock
• Mostly Physical Trading

The Native Share & Stock Brokers'
Association
• Now known as Bombay Stock Exchange
• Set up in 1877
• Trading in ownership rights of the firms
• Variously called as ‘allotments’, ‘scrips’ and ‘shares’
• Delivery based Trading
• Trading was localized – through brokers
• Pit Traders, Brokers

Circa 1992
• Screen Based Trading System
• Localized behavior of the exchange was now globalized
• Anonymity of Orders
• Costs and Errors Reduced
• Reduction in manipulation
• Derivatives and Dematerialization

Now
• Trade matches in microseconds
• Complete Transparency
• Volumes are all time high
• Complex Instruments, Derivatives
• Extremely Democratic
• Much better control over Trading

Concepts of Diverse fields
Statistics
Finance
Computer
Science
Operations
Research
Economics
Psychology
History
Mathematics
Strategy

Search For Alpha
• Alpha is the ability to predict the future. Alpha is defined as
the additional return over a naive forecast.
• Finding alpha is the job of a quant research analyst.
• Alpha comes from following sources:
1. Information
2. Modeling
3. Speed

Speed First
• Simplest of All sources.
• For two strategies, doing the same, the faster one will do
better.
• Understanding and Implementing this is simpler and more
objective

Having an Estimate is Better than None at all
• Informational Alpha is sources of information
1. Proprietary information sources
2. Tick by Tick
3. Extraneous sources
• Modeling Alpha is development of Trading Models
1. Models provide trading edge
2. Valuation, Hedging etc.

Quantitative Trading Segmentation
Market
Making
 Get Inside the bid-ask spread and buy low, sell high
Arbitrage  Take Advantage of things trading at different prices
on different exchanges or through similar
instruments
Momentum  If it goes up, it keeps going up
Mean
Reversion
 If it has gone up, then it is bound to come back

Consequence of Definition
• Strategy must finish the day flat, HFTs must exhibit balanced
bi-directional (i.e. “two-way”) flow
• HFTs can't accumulate large positions
• HFTs can't deploy large amounts of capital
• HFTs have little need for outside capital or leverage, and
tend to be proprietary traders
• HFTs can't “blow up” (they don't use much leverage, and
don't have much capital, so they can't lose much capital!)
Workshop on Algorithmic & High Frequency Trading

Understanding HFT
• HFTs take the opposite side of trades of long-term investors
• Long term investors impact many securities besides the
ones they are directly trade, because stocks are correlated
• This creates opportunities for Statistical Arbitrageurs,
whose activity keeps correlated stocks “fairly priced” with
respect to one another
• r
• HFT comes in, when volatility is high, liquidity is in short
supply, and it becomes very profitable to provide it
• HFTs benefit from volatility, so they can not cause it

Contrasting HFT and Long Term Investing
HFT Long Term Investing
Profit Margins Small Large
Transaction Costs Small Large
Workshop on Algorithmic & High Frequency Trading
Capital
Requirements
Small Large
Consistency of
Profits
High Low
Total Profit Potential Small Large

Economics of HFT
• Opportunities for short-term returns follow a Gaussian
(Normal) distribution
– large expected returns are rare; tiny expected returns are
abundant
• HF Traders target opportunities that are tiny (expected
returns ~ 0.15 Rs before costs)
• Long-term investors don't have the cost-structure to target
such trades! (Cost being 0.35 Rs)
• typical HF trade: expected return = 0.15 Rs after costs,
standard deviation = +/- 5 Rs
• The risk/reward of such trades is not meaningful to long-term
investors

Economics of HFT
• Small returns are appealing to HFT because they are very
plentiful
• typical HF trade: expected return = 0.15 Rs after costs,
standard deviation = +/- 6 Rs
• after 100 such trades: expected return = 0.15 Rs; standard
deviation = +/- 0.6 Rs
• if one does 100 such trades per day, for full year: sharpe
ratio of 4.0
• if one does 10,000 such trades per day, for full year: sharpe
ratio of 40.0

Capturing HFT opportunities requires use of
advanced technology

Vicious Cycle
More
Volumes
Opportunities
Lower Costs
More
Higher volumes lead to gains in efficiency through the use of
technology, leading to lower transaction costs. Technology
is the enabler of the virtuous cycle, but cost is the driver.
As costs approach zero, volumes will peak as a result.

Market-making opportunities arise because long-term
investors desire immediacy when making trades
Investor 1 has to wait for 1
hour to find a Counterparty
T1 = 10 AM T2= 11 AM
Investor 1 comes to buy
shares at 100.05 or lower
Investor 2 comes to sell
shares at 99.95 or higher
Investor 1 buys from HFT at
100.05 at 10 am and Investor
2 sells to HFT at 11 am
T1 = 10 AM T2= 11 AM
Investor 1 comes to buy
shares at 100.05 or lower
Investor 2 comes to sell
shares at 99.95 or higher

Statistical Arbitrage
Reliance
Futures
Reliance
Put/Call
Nifty
Put/Call
Reliance
Stock
Nifty
Futures
Statistical correlations arise because securities are driven by systematic factors such
as inflation, regulatory policies, currency prices, economic growth, and so on.
Because there are far fewer systematic drivers than there are securities which
depend on them, correlation between securities is guaranteed to exist!

Understanding HFT
Structural vs Statistical Correlations
 Structural correlations tend to be strong,
steady, and robust.
 profitable opportunities tend to be very
easy to identify, and are thus heavily
competed for.
 Competition prevents structural price
divergences from growing large – Small
bets
 tremendous speed is required in order
to access them before competitors
 mainstay of HFTs, who specialize in fast
trading
 Statistical correlations tend to be weak,
time-varying, and non-stationary
 profitable opportunities based on
statistical correlations tend to be harder
to model, and more persistent in terms
of their duration
 size and duration of these opportunities
facilitates large bet-sizes and overnight
positioning
 Such opportunities tend to be favoured
by large quantitative hedge funds
specializing in statistical analysis

E-PAT course structure
Core Content
Statistics and Econometrics
Financial Computing &
Technology
Algo & Quant trading

E-PAT Course Structure: Statistics and
Econometrics
Core Content
Technology
 Probability and Distribution
 Statistical Inference
 Linear Regression
 Correlation vs. Co-integration
 ARIMA, ARCH-GARCH Models
 Multiple Regression
Basic Statistics
Advanced Statistics
Time Series Analysis
 Stochastic Math
 Causality
 Forecasting

E-PAT Course Structure: Financial Computing
& Technology
Core Content
Technology
 Intro to Programming Language(s)
 Programming on Algorithmic
Trading Platforms
 Linear Regression
 System Architecture
 Understanding an Algo Trading
Platform
 Handling HFT Data
Programming
Technology for Algorithmic
Trading
Statistical Tools
 Excel & VBA
 Financial Modeling using R
 Using R & Excel for Back-testing

E-PAT Course Structure: Financial Computing
Core Content
Technology
 Statistical Arbitrage
 Market Making Strategies
 Execution Strategies
 Forecasting & AI Based Strategies
 Machine readable News based
 Trend following Strategies
 Option Pricing Model
 Time Structure of Volatility
 Dispersion Trading
 Volatility Forecasting & Interpretations
 Managing Risk using Greeks
 Position Analysis
 Order Book Dynamics
 Market Microstructure
Trading Strategies
Derivatives & Market
Microstructure
Statistical Tools
 Hardware & Network
 Regulatory Framework
 Exchange Infrastructure & Financial
Planning (Costing)
 Handling Risk Management in
Automated systems
& Technology

E-PAT course mapping

Program Delivery
• Part-time program
– 3 hrs sessions on Saturday & Sunday both
– 4 months long program
– 100 contact hours including practical sessions
• Convenience - webinars
• Open Source
• Virtual Classroom integration
• Student Portal

Mapping Skill Set
Trading
Knowledge
Software
Development
Quantitative Skills
Trading Sales Trading Algo Trading Broking
Asset
management
– Mid Office
Asset
management
– Front
Office

Opportunities for Technologists
Brokerages/Banks Trading
Trading
Front
Office
Asset Management/MF
 Hedge Funds, Prop Funds
– Modeling, Coding –Excel
– 20-25 L
Proprietary Trading
– Trading, Modeling (MATLAB, R,
Kdb)
– 25-50 L
Trading
Mid
Office
Quants, Sales Trading
 Banks, Brokerages
–Modeling, Coding-
MATLAB/R/Excel
– 12-18 L
Technology, Operations
– Development (C++, Java,
Python)
– 20-30 L

Opportunities

Entrance Test
• Check your pre-requisite knowledge by taking the entrance
test:
http:/www.quantinsti.com/epat_scholarship_test.php

Coming dates
• http://www.quantinsti.com/importantdates.h
tml

Q&A
• Please type your questions in the chat
window.

Quant insti webinar on algorithmic trading for technocrats!

More Related Content

What's hot (20)

Viewers also liked (12)

Similar to Quant insti webinar on algorithmic trading for technocrats! (20)

More from QuantInsti (20)

Recently uploaded (20)

Quant insti webinar on algorithmic trading for technocrats!