Predicting SP500 Index Using Artificial Neural Network

摘要

This paper studies artificial neural network algorithm as a means of modelling and forecasting the financial market data. Such method bypasses traditional statistical method to deal with financial time series data. A recurrent neural network model, Elman network, is implemented to incorporate autocorrelation in time series data. A 3-parameter model is chosen to fit and forecast S&P 500 index. The experimental data is from 2000-2007, to screen out the abnormal market environment after 2008 financial crisis.