A Comparison of Genotype Representations to Acquire Stock Trading Strategy Using Genetic Algorithms

Automatic trading methods are important issues in recent financial markets. In this paper, we compare some genotype coding methods of technical indicators and their parameters to acquire stock trading strategy using genetic algorithms (GAs). In previous works, the locus-based representation is widely used for encoding technical indicators on chromosomes in GAs, and the direct coding is also widely adopted for encoding the parameters of the indicators. However, these conventional methods are not so effective for the GA search. Therefore, we propose a new genotype coding methods, namely the allele-based indirect coding. We examine the performance of the proposed and conventional coding methods in stock trading of twenty companies in the first section of the Tokyo Stock Exchange for recent ten years. In our empirical results, the allele-based indirect coding is superior to the other ones both on the cumulative profits and the computational costs.