Improved particle swarm optimization for fuzzy based stock market turning points prediction

Abstract

Stock prices usually appear as a series of zigzag patterns that move in upward and downward trends. These zigzag patterns are learned as a tool for predicting the stock market turning points. Identification of these zigzag patterns is a challenge because they occur in multi-resolutions and are hidden in the stock prices. Furthermore, learning from these zigzag patterns for prediction of stock market turning points involves vagueness or imprecision. To address these problems, this research proposed the swarm-based stock market turning points prediction model which is a combination of a zigzag patterns extraction method, and a mutation- capable particle swarm optimization method. This model also includes the stepwise regression analysis, adaptive neuro-fuzzy classifier, and subtractive clustering method. This study explores the benefits of the zigzag-based multi-ways search tree data structure to manage the zigzag patterns for extracting interesting zigzag patterns. Furthermore, the mutation capable particle swarm optimization method is used to optimize the parameters of subtractive clustering method for finding the optimal number of fuzzy rules of adaptive neuro-fuzzy classifier. Stepwise regression analysis is used to select the important features from the curse of input dimensions. Finally, adaptive neuro-fuzzy classifier is used for learning the historical turning points from the selected input features and the extracted zigzag patterns to predict stock market turning points. The proposed turning points prediction model is tested using stock market datasets which are the historical data of stocks listed as components of S&P500 index of New York Stock Exchange. These data are stock prices that are either moving upward, downward, or sideways. From the findings, the proposed turning points prediction model has the potential to improve the predictive accuracy, and the performance of stock market trading simulation.