A Neural Network Model for Predicting Stock Market Prices

Stock exchanges are considered major players in the financial sector of many countries. In such exchanges, it is Stockbrokers who execute stock trade deals and advise clients on where to invest. Most of these Stockbrokers use technical, fundamental or time series analysis in trying to predict future stock prices, so as to advise clients on appropriate investments. However, these strategies do not usually guarantee good returns because they guide on trends and not the most likely trade price of a future date. It is therefore necessary to explore improved methods of prediction. The research uses Artificial Neural Network (ANN) that is feedforward multi-layer perceptron (MLP) with error backpropagation to develop a model ANN of configuration 5:21:21:1 using 80% data for training in 130,000 cycles. The research then develops a prototype and tests it using 2008-2012 data from various stock markets, such as the Nairobi Securities Exchange (NSE) and New York Stock Exchange (NYSE). Results showed that the model predicted prices with MAPE of 0.71% to 2.77%. Validation done using Neuroph & Encog showed close RMSE. The model can therefore be used in any typical stock market predict.