A Comparative Study on Time-delay Neural Network and GARCH Models for Forecasting Agricultural Commodity Price Volatility

Abstract

In this paper, forecasting performance of time-delay neural network and GARCH models for predicting the volatility using monthly price series of edible oils in domestic and international markets is evaluated. An attempt has also been made to investigate whether the forecasting performance of two competing models can be improved by combining their individual forecasts. For this purpose, the individual models were combined to produce improved forecasts using non-parametric approach through the use of kernel. Further, the models were evaluated on their ability to predict the correct change of direction (CCD) for future values.