We investigate the design of trading systems using a genetic algorithm GA. Technical indicators are used to define entry and exit rules. The choice of indicators and their associated parameters are optimized by the GA which operates on integer values only. Holding time and profit target exit rules are also evaluated. It is found that a fitness function based on winning probability coupled with a profit target and one based on the Sharpe ratio are useful in maximizing percentage of winning trades as well as overall profit. Strategies are developed which are highly competitive to buy and hold.
R. Tymerski, Ethan Ott, G. Greenwood
Journal name not available for this finding