Predicting the return of the Japanese stock market with artificial neural network

The Japanese stock market is one of the largest stock markets in the world, and the prediction of it is a hotspot research area. As the most widely quoted average of Japanese equities, Nikkei 225 index is a benchmark to value the Japanese economy. Forecasting the stock market returns is a major activity of financial firms and private investors when they make investment decisions. Accurate prediction of the stock market returns is a highly challenging task due to the highly nonlinear nature of the financial time series. In this study, we applied the artificial neural network which can map any nonlinear function without a prior assumption for predicting the return of Nikkei 225 index. Due to the complexity of stock market data, first we selected the 18 input variables from the data of 71 variables that covered financial and economic information of Japanese stock market by the fuzzy surfaces. And then, in order to verify the prediction ability of the selected input variables, we predicted the return of Nikkei 225 index by the artificial neural networks with the learning algorithm of back propagation. For the neural network model, we compare linear regression model with it in the prediction ability of the stock market return. It was observed through empirical experiment that the artificial neural networks performed well, and had a more effective ability than the conventional linear regression in forecasting the Japanese stock market. In addition, the prediction effect of the combination of 18 input variables is effective and can be therefore a good alternative for stock market returns prediction.