PREDICTING STOCK PRICE PERFORMANCE: A NEURAL NETWORK APPROACH

Abstract

The prediction of stock price performance involves the interaction of many variables, making prediction very difficult and complex. Many analysts and investors use financial statement data to assist in projecting future stock price trends. The purpose of this paper is to apply the neural Network Approach to predict stock price performance by using Colombo Stock Exchange data in the period of 1993 January to 2005 June. This consists of 12 years data for the analysis. The sample is 2552 daily stock prices in seven companies which have been listed in Colombo Stock Exchange (CSE).  The data used in this study were gathered from Daily Price list published by CSE. This source provides company related data and market performance indicators. The input for the network was a list of nine variables. Output variable is price performance of seven companies separately. The three layered network correctly classified 85 percent of the mean training data and appropriately predicted 90.0 percent of the mean testing data.

Keywords:   Artificial Neural Networks, Back-Propagation, Topology, Root Mean  Square Errors

For full paper: fmscresearch@sjp.ac.lk