Pairs Trading from NYC Algorithmic Trading Meetup November '13
Download as PPTX, PDF26 likes28,864 views
The document is a guide to building a pairs trading algorithmic trading system, emphasizing that the concept is accessible to retail investors. It discusses the importance of understanding statistical methods, backtesting, and the risks involved in trading, while encouraging self-learning and automation. Key resources and next steps for aspiring traders are also provided, along with personal experiences of the author in algorithmic trading.
Pairs Trading from NYC Algorithmic Trading Meetup November '13
- 1. How To Build Your Own Pairs Trading Algorithm Trading System Quantopian Meetup Gary Chan
- 2. Introduction • Started programming at a young age • Played poker as main source of income for a few years • Money is just a way to keep score • Became interested in stock market • Read ~70 textbooks, mostly corporate finance, CFA curriculum • Currently running an algorithmic trading system out of my apartment
- 3. Expectations • I’m not here to divulge secrets • Teach a man to fish, not give him a fish • You will see how easy the pairs trading model is to understand and build • Deployable by retail investors • Seeing is believing • You will not be able to build your own tomorrow… but one day… if you want it badly enough • It’s a small world, so share ideas
- 4. Why you can beat the professionals • Definition of professionals • Algorithmic does not mean high frequency • HFT is not for retail investors • Strategy capacity • Find a niche
- 5. Disclaimers • A lot of learning, before this came to fruition • Content here will not be complete due to time constraints • Survivorship bias, Look ahead bias • Model risk, implementation risk, execution risk, over parameterization • Random models can have profitable backtests • I’m not responsible if you lose money
- 6. Algorithmic Trading is Easy • • • • • • What does easy / hard mean? Info, books, tools, cheaply or freely available Yahoo, Google, Morningstar, Edgar, Quantopian Visual Studio Express, MySQL, R Many textbooks written on the subject Low capital requirements (30K minimum) compared to other ventures • Backtest before you use real money!! • Higher quality data will cost more money
- 7. My Current System • • • • • Low frequency Single computer, 3 years old Runs on Wifi from my apartment Started coding Oct 2012 Started forward testing play money in March 2013 • Started forward testing real money June 2013 • Current performance consistent with backtests • Just started using cloud computing
- 8. My Cloud
- 9. My Cloud Part 2
- 10. Backtesting • Profitable backtests does not mean good • You need a valid model • Live trading WILL underperform backtests
- 11. How A Good Backtest Looks • Consistent, high R Squared • Similar parameters have similar equity curves
- 12. Key Info From Backtests • • • • Portfolio of 39 pairs Average of 12 to 18 pairs with opened positions, $5400 max drawdown $10k sized legs in backtests, $395 average profit per trade 81.3% winning trades, 4540 trades, 26 day average holding period
- 13. Pairs Trading Books • Contains Code • Complete manual to putting together your own system
- 14. Types of Investment Strategies • Mean Reversion – Buy low, sell high / Sell high, buy low – Fundamental Analysis – Statistical Arbitrage • Momentum – Sell low, buy lower / Buy high, sell higher – News • Technical analysis does not work
- 15. Necessary Math • • • • • • • Basic statistics Linear algebra Standard deviations, regressions Types of distributions (normal, logarithmic) Random walks Stationary (mean reverting) series Cointegration
- 16. Mean Reversion Example #1 • • • • • Keep an opened mind Everything can be measured in money If you don’t believe me, put it on EBay Fear can be measured in dollars Unwarranted fear -> Sell volatility to profit
- 17. Mean Reversion Example #2 • • • • • Stocks move in random walks Some stocks move together Spread between the stocks are mean reverting Buy low, sell high on the spread Statistical arbitrage means you win most of the time, not all the time • Buyouts and bankruptcies result in large losses • Diversification is a must • Learn some corporate finance / fundamentals
- 18. Case Study, GLD and IAU • Two gold ETF’s • Pull end of day prices from Yahoo Finance into csv files • Do a linear regression • Calculate the spread • Graph of spread is mean reverting • Find #Stdevs to enter / exit trade • Optimize parameters
- 19. Next Steps • Try higher quality data • You could trade this by hand + excel, but better to automate the process • Many systems are built in excel • Optimize the code • Backtest until you find a profitable strategy • Use a brokerage with an API, sample code • Forward test with play money
- 20. Next Steps Part 2 • • • • • • • Execution, order, position management Test with small amounts of real money Tweak your system Ramp up trading size if still profitable Exhaust universe for the strategy Find new strategies Never stop learning
- 21. The End • Ernest P Chan - Quantitative Trading: How to Build Your Own Algorithmic Trading Business (Wiley Trading) • Ganapathy Vidyamurthy - Pairs Trading: Quantitative Methods and Analysis (Wiley Finance) • R - http://www.r-project.org/ (like Matlab) Yahoo, Google, Morningstar, Edgar, Quantopian • Visual Studio Express, MySQL, R • garychan7@gmail.com • https://twitter.com/RITrading - I tweet my trades here