On Developing Trading Systems by Jeffrey Owen Katz, Ph.D.,with Donna L. McCormick
This neural network developer details steps in developing a trading system using advanced technology.
STOCKS & COMMODITIES readers have one thing in common: We have all taken on the challenge of attempting to forecast and trade the financial markets. For me, I first considered this challenge as the result of a chance remark. Since I had also become curious
about trading at the time, I decided to take on the challenge of modeling the markets. That was more than a decade ago.
Initially, I developed several models that provided only entry signals for the Standard & Poor's 500/OEX. While these signals were often 80% to 90% accurate, I found myself second-guessing them. Moreover, I had to rely on my own subjective determinations of such critical factors as when to exit and where to place stops. As a result, my first attempts at trading produced more losses than wins. Emotional decision-making is a killer when it comes to trading!
Eventually, I realized that in order to succeed, I had to do four things: First, spend more time on the problem than I had previously; second, learn more about trading and the markets; third, develop specialized tools, such as a good trading simulator, with which to develop computerized trading systems; and fourth and most important, develop a computerized system to eliminate subjectivity.