Combine neural network for stock price prediction

This paper is focused on the prediction of a stock market price behavior by an innovative model with combination of artificial neural network (ANN). For this purpose, three types of data that reports daily in Iranian stock market have been used. The structure of this hybrid model consists of two-levels: base predictors in the first level, are responsible for forecasting daily data with different characteristics of a stock i.e. three independent neural networks for prediction of stock price, option volume and rate of return are used. On the second level, the other networks, as a Combinator, the final prediction and analysis of predictive information of the first level will be done. Experimental results on one set of Iran’s stock data showed the superior performance of the suggested model in comparison with current predict model.