Online multiple-model approach to prediction for financial markets

Abstract

Financial market prediction, being a complex problem, has intrigued researchers for a long time. In this paper, we try to address the problem by treating it as a time-series and employing artificial neural networks (ANNs) to forecast the future stock value. Two types of neural network learning algorithms are illustrated for the current application: The backward propagation algorithm and an online sequential learning algorithm. Several training strategies are also proposed. The principle objective of this paper is to demonstrate the improvement in predictive performance using multiple neural networks. Towards this, an attempt is made at predicting the SENSEX value of the Bombay Stock Exchange.