A Markov-fuzzy Combination Model For Stock Market Forecasting

This paper presents a simple combination of Markov model and fuzzy time series model (called MC-fuzzy) for forecasting stock market data. The fuzzy time series model is used to partition the dataspace into states and also solves the fuzzy data in stock index future price. The Markov model then is used to identify data patterns and forecast future states. The appropriate fuzzy rule then defuzzies the data for forecast value. Experimental results of the proposed model show that the performance is better than other forecasting models such as, ARIMA, artificial neural network (ANN), Hidden Markov Model (HMM) -based models and equivalent to HMM-Fuzzy models.