Based On The Bp Neural Network Of Stock Index Futures Price Forecast

The market of stock index futures is an unstable, open and complex system with nonlinear dynamical variation. The variation of the price of futures in stock index futures market is affected by many factors such as financial, economic, political, and social ones as well as investors’psychology. Its change process is nonlinear, chaotic and long-memory. Taking the features of stock index futures market into consideration and basing on historical data, the author uses BP neural network to study the variation trend of and make prediction on the prices of stock index futures in the short term.First, the author summarises several methods of forecasting the price of stock index futures. After a brief comparative analysis of the pros and cons of different methods, the author selects BP neural network as a basic approach for forecasting the price of stock index futures. Then, the author studies the structure of BP neural network, its adaptability in making prediction in financial market, as well as its algorithm and implementation of the algorithm using Mat lab tools, etc. The experimental method adopted shows that the structure of BP neural network is single input multiple output (SIMO) and of a single hidden layer with multi-node. At the same time, this author designs three sets of control experiments which use the BP neural network to improve the accuracy of prediction on the price trend. The first is based on a single factor, i.e. the closing price. Several factors such as the closing price and trading volume is taken into account in the second set of experiments. The third one combines the empirical mode decomposition (EMD) algorithm with BP neural network to build a model to forecast the prices of stock index futures.The experimental results show that the prices of stock index futures in the short term can be forecasted with higher accuracy based on BP neural network model. The control experiments also demonstrate that combining EMD algorithm with BP neural network can achieve a more accurate prediction on the prices of stock index futures.