Think Breadth, Not Depth
- 1. THINK BREADTH, NOT DEPTH Seong Lee O P E N D A T A S C I E N C E C O N F E R E N C E_ BOSTON 2015 @opendatasci
- 2. THINK BREADTH, NOT DEPTH HOW NEW DATA GENERATES NEW ALPHA
- 3. OUR AGENDA • My Story With Quantitative Investing • The Problem Of Small Fish Big Pond • The Solution: Big Fish Small Pond • How This Applies To Quantitative Finance • How You Can Do This Too • Conclusion
- 4. MY STORY • I work with algorithms, clients, and ideas • My job is to combine all three into one finished product • Everyone has different ideas but wants the same thing • Alpha but everyone uses the same data • Wall Street Analyst Estimates – PEAD • Build something that beats the market?
- 5. THE PROBLEM • Too many people compete in the same arena • Small fish in a big pond • Finance: Wall Street • Business: Samsung, Apple
- 6. THE SOLUTION • Where no one is, is the best place for you to succeed • Wall Street: Don’t go to MS, go to the boutique firm next door • Business: A smartphone for 3rd world toxin detection in water systems • AKA new data sources
- 7. REAL LIFE EXAMPLE • EVERYONE uses wall street analyst earnings estimates • Wall Street vs. Crowd (Estimize) • Estimize crowd predictions are 67% more accurate • Algorithm perform 2x better on Crowd vs. Wall Street
- 8. WALLSTREET
- 9. CROWDSOURCED
- 10. AVOIDING THE TWITTER LEAK • April 28, 2015 Twitter earnings were leaked • 20% drop in market cap • $5 billion drop in market cap • Accern (News Sentiment) analyzed +100,000 articles • 1 day before leak, Accern shorted Twitter • SunTrust downgraded Twitter -> Strong Negative sentiment
- 11. AVOIDING THE TWITTER LEAK
- 12. HOW YOU CAN DO THIS TOO • Kimono Labs • API for scraping websites • Quandl • Thousands and thousands of datasets • Social Media • CO Everywhere, Ground Signal • Location based social media information
- 13. CONCLUSION • Think breadth and diversity • Email: slee@quantopian.com
Editor's Notes
- #5: And from working with all of these different clients, I’ve realized that everyone has different ideas. Most of these different ideas are generally small tweaks on existing methods The question is not a yes/no answer, but what will give me the highest probability of beating JP Morgan? Morgan Stanley? Renaissance Establish the PEAD first An extremely well documented phenomon that happens on the earnings announcement date So a companuy like apple will announe earnings about four times a year When I first started my job at Quantopian, Jess Stauth, our VP of Quant Strategy, came to me with the challenge of writing a PEAD algorithm
- #6: Too many people are competing in the same arena Take Wall Street -> Boston College graduate competing against Harvard Graduate Take Business -> HTC or Nokia competing against AAPL and Samsung Finally, something that actually relevant -> let’s look at a bar If you’re a small business, you need to compete in a different arena
- #7: The world of quantitative finance, there are two different ways to compete in this arena: mathematics, data Does it seem weird that we have the solution in different slides -> can we make them all congruent? E.g. put this slide towards the end where wall street, business, and life are afterwards?
- #8: Explain WHY the crowd is unconventional Sticking point – Need to talk about the algorithm itself briefly? Looking at an earnings surprise, buy/sell and hold position for 5 days.
- #13: This should be extremely specific for different types of life situations