Optimized Selection of Stock Portfolio by using the Fuzzy Artificial Neural Networks Web Model, ARIMA & Markowitz Model in Tehran Stock Exchange

Abstract

Financial issues have always been the main topics of scholars’ research. Fuzzy logic is one of the techniques that are widely used in this study in order to model the environmental uncertainty. The aim of this paper is combination of fuzzy logic and neural networks to select a basket (portfolio) of stocks. Web forecasting system based on fuzzy artificial neural networks that discovers fuzzy rules by using the past time data and predicts it, is also applied in this learning algorithm. The data of this study have been collected weekly from the Tehran Stock Exchange. Output data Simulation had been collected from the stock market base using obtained output data. This paper first deals with the study of financial markets. After that the research models were described and by using the other linear techniques such as Markowitz and ARIMA models, stock price was predicted. Then performance of the models was investigated with two population mean test (t-student) at the 95% confidence level. At the end Fuzzy artificial Neural Web network was selected as the best model for decision- makers. To perform research models and analysis of Java source code and software PASW 18 and ISP Server and also JDK3.1 were used. Finally, practical suggestions were given.