Coevolutionary grammatical evolution for building trading algorithms

Abstract

Advancements in communications and computer technology has enabled traders to program their trading strategies into computer programs (trading algorithms) that submit electronic orders to an exchange automatically. The work in this chapter entails the use of a coevolutionary algorithm based on grammatical evolution to produce trading algorithms. The trading algorithms developed are benchmarked against a publicly available trading system called the turtle trading system (TTS). The results suggest that out framework is capable of producing trading algorithms that outperform the TTS. In addition, a comparison between trading algorithms developed under a utilitarian framework, and using Sharpe ratio as objective function shows that they have statistically different performance. © 2011 Springer Science+Business Media B.V.