March 21, 2009

I've been focusing on the book Book Notes - Quantitative Trading Strategies. I've read the first 90 pages.

Wrote some new EasyLanguage strategies to implement ideas in the book:

• Channel Breakout Strategy, pg. 90
• Channel Breakout With Filtering And Stop Loss and Stop Limits - My enhancements to test the effectiveness of Online Trading Academy rules for overall market, Trin, and Tick as filters.

On two stocks I am trading recently, GS and MET, the Channel Breakout Strategy has a dismal record when applied to one minute chart. OTA rules to help limit the losses, but they don't make it profitable.

Mar 25, 2009

It turns out that $TRIN and $TICK are not available all the way back to 1/1/1990, so a trading strategy depending on them will not generate trades until "max lookback bars" after the first date of that data. So I prepared a simpler strategy that doesn't use those extra filters.

Unfortunately, this book doesn't include the actual strategy code that was used. I have learned EasyLanguage well enough to be able to code the strategies described. Unfortunately, I am unable to replicate the results in the book. There isn't enough information in the book to really figure out why the differences. This article details a number of the issues that I ran into: Channel Breakout Strategy Including Shares According To Volatility. One of the biggest issues is the number of shares to trade.

Also some of the strategies are only generating about 70 trades over a 10 year period of daily price charts. This is not very many trades.

To simplify the situation, I will probably evaluate the strategies based on 100 share trades, and see if there is some correspondence in performance. Unfortunately, given the varying share size formula in the book, which apparently hasn't been properly described, not only is the net profit very different in my code, the Sharpe and K-ratios seem to be affected, sometimes drammatically. I should as an exercise run all the stocks listed in the book against my channel breakout formula, and see if I arrive at a chart or numbers anywhere close to the book. Unfortunately, given that TradeStation strategies can only trade one stock at a time, and there is no way to combine trades of multiple symbols in a single backtest performance report, it will require a lot of work to collate the reports for 34 different symbols.

Given that I want to find patterns for intraday trading, it's questionable how much guidance this book will really provide. I'm still planning to write programs to replicate each strategy, and test at least a few of the more promising stocks.

Just read the reviews on AMAZON.COM and find that generally it is badly reviewed, only two stars. Maybe I shouldn't have bought the book, but I do feel I have learned from it, if nothing else, to be skeptical. I am convinced I need to find out more about books like "How I became a Quant", however that book is just a collection of Wall Street "war stories", probably not actual techniques.

For the upcoming Online Trading Academy course I'll be taking, at least I should be better prepared to back test their suggested strategies. Of course having a computer recognize technical analysis patterns on a graph is not childs play, it is certainly more difficult than candlesticks. Perhaps the next book I should focus on is Book Notes - Money-Making Candlestick Patterns