Model and method for the stock market forecast

Abstract

The price fluctuations of the stock market are relevant for any economy. A method to construct an expert system that predicts the stock prices evolution is proposed. The Expert's rules are modeled by Fuzzy Logic Enhanced Time Petri Net (FLETPN). This model guides the inference rules to get the answers related to the market evolution. The model is capable of describing the periodic behavior as well as the reaction to the asynchronous relevant events that influence the market. The expert's rules are tuned by a genetic algorithm. The monitoring information of the market from the past evolution is used to improve the fuzzy logic rules (i.e., the expert system). The application of the method on the market evolution is used to evaluate the method forecast quality.