Study on stock price prediction based on BP Neural Network

In this paper, two kinds of methods, namely additional momentum method and self-adaptive learning rate adjustment method, are used to improve the BP algorithm. Considering the diversity of factors which affect stock prices, Single-input and Multi-input Prediction Model (SIPM and MIPM) are established respectively to implement short-term forecasts for SDIC Electric Power (600886) shares and Bank of China (601988) shares in 2009. Experiments indicate that the improved BP model has superior performance to the basic BP model, and MIPM is also better than SIPM. However, the best performance is obtained by using MIPM and improved prediction model cohesively.