Design of a Hybrid Expert System for Stock Market Prediction

The major objective of this study is to develop a proposal for building an expert system model that can be used to help the traders and investors take trading decisions before an event of price change takes place in future.

The proposed model uses an amalgamated approach of combining fuzzy logic, time series data, data mining and document analysis. The novelty of the proposed system is the use of a unique technique developed during the research process named as fuzzy document based information retrieval scheme (FDIRS).

For performing the experiments, four global stock market indices and thirteen Indian stock market stocks that cover all the major sectors that drive the economy of India, were used so that the performance of the proposed model could be validated properly. RMSE used as the performance parameter helps to test the forecasting ability of the proposed model. In order to assess the performance of the proposed model the RMSE values of the already established methods and the proposed method is compared. The data set used while testing the proposed model was the same data set given to the already established methodologies, this approach enabled to compare the outputs and performance factors more critically..