Prediction Model of Stock Prices Based on Correlative Analysis and Neural Networks

There is a low forecasting accuracy when we forecast the price date by the traditional BP network only considering a single closing price as the time series vector. But, if we try to add other factors vector to the BP network input vector, we will rind there are low training accuracy caused by the a large number of factors In order to solve the issues raised above the author sets up a two-step Forecast approach with the combination between SOFM network and BP network. First, we uses the Gray Correlation Analysis to choose the set of variable which can describe the characteristics of the state of the stock market from a number of technical indicators. Then we can classify the state of stock market by the SOFM network which has the capacity of self-organizing classification. And base on the classification, we use BP network to accurately predict. The results of experiment showed that the predictive accuracy of SOFM-BP model is more improved than that of traditional BP neural network model. And it is feasible and effective to forecast China's stock market by SOFM-BP model, which has a prospective future.