Stock Market Index Prediction Using Artificial Neural Network

Abstract

Often, nonlinearity exists in the financial markets while Artificial Neural Network (ANN) could be used to expect equity market returns for the next years. ANN has been improved its ability to forecast the daily stock exchange rate and to investigate several feeds using the back propagation algorithm. The proposed research utilized five neural network models, Elman network, Multilayer Perceptron (MLP) network, Elman network with Self-Optimizing Map (SOM), MLP with SOM filter and simple linear regression, for estimating new values. Results were examined to investigate the predicting ability and to provide an effective feeds for future values. The result of the proposed simulation showed that SOM could greatly improve the convergence of the neuron networks; whereas Elman network did a better performance to capture the temporal pattern of the symbolic streams generated by SOM.A benchmark of linear regression model was also employed to show the ability of neural network models to generate higher accuracy in forecasting financial market index.