REAL WORLD

Developing Systems
With A Rule Based Approach

by Jeffrey Owen Katz, Ph.D.,
with Donna L. McCormick
This month, Katz discusses a trading system he developed using a rule-based approach.

"Previously, I explored the development of simple rule-based trading systems. Specifically, I looked at a rule-based system that revolved around a theory, the thrust-retracement hypothesis, about how the market functions. This time, I'll also explore rule-based trading systems, but rather than constructing a set of rules to reflect a hypothesis of market behavior, I will engage in "data mining" -- that is, empirically determining tradable rules."

A genetic algorithm solves a problem using the same processes as biological evolution: It works by recombination and mutation of gene sequences, which in our case may be sets of number series.

ABOUT THE RULES
In this current effort, I focus on simple rules that may easily be manipulated and optimized without the aid of fancy technology. Recently, I developed a Standard & Poor's 500 trading system using this approach. I had two rules, each with a parameter or two, and I simply played with different values of the parameters until I found a set that traded the way I liked.
FIGURE 1: Here are the performance summary and parameters.

BUT FIRST, SOME INTERPRETATIONS
Before discussing the results, certain elements of the performance summaries presented in Figures 1 through 9 require some explanation. Depending on whether the solution is for the long or short side, values will be presented in the Long or Short columns, respectively; the values in the Total column are the result of adding both columns.
 

For more information on system implementation and the TradeStation system module, click on the sidebar "System Implementation."

Jeffrey Owen Katz is a professional trader and consultant in Selden, NY. His firm, Scientific Consultant Services (516 696-3333), specializes in custom programming, provides expert consultation on systems development and the use of Omega Research's tools, and develops publicly available software for traders.
Donna L. McCormick is a writer and vice president of Scientific Consultant Services.
Excerpted from an article originally published in the January 1997 issue of Technical Analysis of STOCKS & COMMODITIES magazine. 
© Copyright 1996, Technical Analysis, Inc. All rights reserved.
Return to January Contents