Forecasting volatility of SET with artificial neural network-GARCH models

Abstract

This research aimed to compare the effectiveness of two forecasting models, ANN-GARCH and ANN-EGARCH. The two hybrid models were formed by a combination of Generalized Auto Regressive Conditional Heteroskedasticity (GARCH) and Artificial Neural Network (ANN) models, used to forecast the volatility of the Stock Exchange of Thailand index (SET). The results showed that both of the ANN-GARCH and ANN-EGARCH forecasts were highly accurate, measured by root mean square error (RMSE) and mean absolute percentage error (MAPE) values. With key variables included, the models accurately described volatility forecasts.