Friday, June 11, 2010

Harris, Short-term Trading with Price Patterns

I have admitted more than once on this blog that I am no systems designer. To be a good at creating trading systems you must be imaginative, comfortable in the world of probability and statistics, and have basic or better programming skills. I may have imagination, but that’s about it. The best blend that I’ve found among active bloggers is CSS Analytics.

But let’s go back ten years, to Michael Harris’s Short-term Trading with Price Patterns (Traders Press, 2000). The book is obviously dated and some of its methodology is—I’m looking for a more scientific description but what comes to mind is—clunky. Nonetheless, this book is an interesting attempt to combine price patterns and probability. Harris focuses on the futures markets.

The basic idea is to identify profitable short-term trading patterns, normally between three and seven bars, and then combine them into a trading system. That is, an entry would be triggered if Pattern A or Pattern B or Pattern C, etc. was identified.

Harris goes beyond simple pattern trading, however, by introducing his p-Indicator. It is in effect a weighted probability function. Let’s say that your system consists of three patterns. You first hypothesize a target and a stop for each pattern. (Harris normally uses the same figure for both the target and the stop, measured in points.) You then look at the entire price history of the particular futures market, analyze both long and short trades, and determine how often each pattern occurred and the percentage of the time that each hypothetical trade met its target before it was stopped out. The formula will return a weighted profitability percentage. The closer the value is to 100%, the better the trading results. Harris says that 60-70% offer good results. The p-Indicator, he claims, provides both an entry signal and a way to size trades.

There are many ways that one could improve on Harris’s methodology, which I’ve sketched in only the broadest strokes here. But I like the idea of having an entry signal which is not simply binary. I know that others have swum in these non-binary waters. For instance, MarketSci Blog had a post more than a year ago about transactional vs. confidence-based trading strategies and Condor Options distinguished between binary and polynary strategies. But unless there’s a shark or a tar ball alert there should be ample opportunity for many more to join in the fun.

1 comment:

  1. I just discovered your blog. I find it very interesting with a different than usual angle on the markets. The discussions of the different books are very helpful. thank you very much.