Fluctuations in the stock market are closely related with all of investors’ benefit, so the research on the stock forecast is both theoretically and practical meaningful. Traditional forecasting methods are generally give a qualitative and long range forecast on the stock market, which has much limitations. Yet the neural networks (NNs) method, a kind of artificial intelligence model, is more and more widely used in the stock forecasting, due to its excellent learning ability, fault tolerance and other good characteristics.Based on the above background, this paper firstly introduces the concept of neural network. Then it makes the research of using genetic algorithms optimize the neural network to improve the speed and accuracy of forecasting. In the paper, the optimization is concentrated on the weights of the neutral networks. In the empirical analysis stage, the SSE50Index is used as a sample and forecast data. The previous day’s closing price and the opening price of the day is input sample, predicting the closing price of the day. The result shows that, it’s much more precise after the neutral networks are optimized by genetic algorithms.However, there are still a lot of problems in the stock forecasting with NNs, such as how to choose a more reasonable input samples and model parameters, etc.That is also the research focus in the stock forecasting with NNs in the future. |