Evolutionary stock trading decision support system using sliding window

Abstract

A novel evolutionary decision support system (EDSS) that is effective for making profitable trading decisions in the stock market is presented. A genetic algorithm is incorporated with a sliding window scheme to effectively estimate the most profitable trading decision, namely buy, hold or sell. Because of its data-driven nature and the genetic change to the positive course, EDSS bypasses the complicated steps of network establishment and subsequent training, and it can directly deal with the problem of structural instability that plagues traditional rule-based systems. Empirical tests are conducted on the weighted price index of the Taiwan stock market (TSEWSI). Compared to the buy-and-hold strategy, the EDSS can achieve a significant improvement in profit gains.