Predicting Close price and Volume based on Neuro-Fuzzy system for Thai Stock

摘要

Prediction of stock markets has been a challenging task and the great interest for researchers as the actual fact that a stock market is highly volatile in its behavior. If successful, may involve substantial pecuniary rewards. This paper discusses about the prediction of close price and volume on next day of Thai stock market trends. We introduce an intelligent decision-making model based on the application of Neuro-Fuzzy system (NFs) to verify that the subset of Technical Analysis is the most appropriate to predict stock trends in each industry group. . Our study are forecasting close price and volume model based on Intelligence system such as Neural network and our proposed neuro-fuzzy which trained and tested using one-year past stock index data. We compared our forecasting model with another intelligence system. The results show that our proposed neuro-fuzzy has minimum forecasting error and can be considered as a good method for close price and volume forecasting. This method is designed to predict 22 days of stock returns in advance. Selection of inputs based on technical analysis affects the performance of the prediction. The correlation of each data set will be suitable for different industries. Therefore, the researcher is investigated the appropriate data together with effectively using the methods as power set of technical analysis, correlation and Principal component analysis. From the 7 securities, the results shows that the PTT stock market in resources industry has been the most efficient example by using Neuro-fuzzy System (NFs) shows the lowest error value. Although the test Error was low we also want to try with sliding window technique to update the information that could help to reduce the effects of error in the future. Thus, our proposed close price and volume forecasting model can be implemented in a Decision-Trading System during the trading day.