Hybrid Intelligent Trading Approach XCS Neural Network Model for Taiwan Stock Index Trend Forecasting

Econometricians build precise hypotheses in advance when they use econometric models to discuss changing stock market trends. However, baffled by these unreasonable hypotheses, economics generally cannot effectively explain real stock market behaviors using mathematical models. Therefore, this study attempts to use genetic theories to produce a rule base that can be adapted to stock market behaviors, and then re-learns it to refine those rules, hopefully discovering knowledge hidden in the stock market. Artificial intelligence models recently have been frequently applied in financial analysis. Compared with econometric models, which require numerous hypotheses and suffer various other limitations, artificial intelligence models are more flexible, able to solve any nonlinear problems, and more suitable for analyzing dynamic environments such as stock markets. This study combines two artificial intelligence technologies: the extended classifier system and backpropagation neural network to establish a XCS-neural-network based trading system, and this system is then used to identify environmental patterns and predict the values of the test set. Experiments reveal that all test data in this study have accuracy rates exceeding 50%. Therefore, this study confidently concludes that the proposed system can help investors make more precise investment decisions.