Forecasting Malaysian stock price using artificial neural networks (ANN)

Abstract

Predicting a stock price is a very difficult task because it is complex and involves many factors. This has led to drop in the investment level in the Malaysian stock market. It is difficult to predict the stock market because its environments are unstable and dynamic. Recently, the demand for neural network in the business arena is on the increase. It is need to analyze vast data in order to search for information and knowledge that do not exist by using traditional methods. This included stock market prediction that is a very significant research in business area. In regard to Bursa Malaysia, Artificial Neural Network (ANNs). ANNs was only used to predict main index, i.e. Kuala Lumpur Composite Index (KLCI), but no attempt to predict share price and in particular banking sector. Since ANN has potential to predict non-linear behavior, this research attempts the use of ANNs to predict banking sector stock price in FTSE Bursa Saham Malaysia Kuala Lumpur Composite Index (FBM KLCI). One of the interesting topics of stock-market research is stock market prediction. Precise stock forecasting becomes the greatest challenge in the investment industry because stock data distribution changes over time. This paper investigates the use of ANN to predict Malaysian stock price, in particular Maybank Berhad stock price. The feedforward neural back-propagation network with Training Function Gradient Decent Training Algorithm is used in this study. The outcome of selected stocks, namely Maybank, are modeled and simulated and the results show that ANN offers a very accurate stock model and also generates competitive systems using all four trading strategies. The results also show that, neural network is a good tool to predict stock price movement with accuracy higher than 95%. Closing price is a good input for neural network model for stock price prediction.