Fuzzy - neural model with hybrid genetic algorithms for stock price forecasting in auto industry in Tehran security exchange

Selection of appropriate time and price in trading stocks has an important role in investment decisions on profit and loss of investors in capital markets. Nonlinear intelligent systems, such as artificial neural networks, fuzzy- neural networks and genetic algorithms, would be used to forecast stock prices motions. In this article,a model of stock prices motions has been designed using Adaptive Neuro- Fuzzy Inference System (ANFIS)integrated with genetic algorithm, in which two different groups of fundamental and technical variables have been employed as model inputs. According to Model outputs,the rate of forecasting errors in both groups of inputs is not significant and these systems are able to forecast daily stock prices. The Mann-Whitney test has been used to measure the accuracy of models and it was found that there is no significant difference between results of prices forecasted in both methods. Both methods are able to forecast next day price with an insignificant error provided that at least one of the inputs in both methods has a linear dependence with price, . Also, results show that these systems do not work properly to forecast prices of high volatility stocks.