Comparison of statistical techniques for forecasting The Stock Exchange of Thailand index

Abstract

The purpose of this study is to compare four forecasting methods on SET Index, SET 50 Index, and SET 100 index. These methods are Simple Exponential Smoothing, Holt’s Method, Box-Jenkins Method, and Regression Analysis. Two sets of Index data were used in this study. The first one used as a training set was collected at the end of each day from January 2, 2008 to September 30, 2011. The other one used as a testing set was collected at the end of each day from October 3, 2011 to October 14, 2011. The diagnostic check was used to test each method. The MAE, MSE, and MAPE from each method shows that Box-Jenkins is more appropriate to the forecasting of SET Index, SET 50 Index and SET 100 Index because its MAE, MSE, and MAPE are fairly low and have no autocorrelations. This leads to the conclusion that the appropriate models for forecasting of SET Index, SET 50 Index and SET 100 Index are ARIMA (2,1,2), ARIMA (2,1,3) and ARIMA (2,1,3) respectively. The most stock market indices are correlated to SET Index, SET 50 Index and SET 100 Index is Composite Index of Indonesia