Prediction of Gold Bullion Return Using Garch Family and Artificial Neural Network Models Dr. Sarangi Prasant*, Dublish Shashank** *Faculty, School of Management, Apeejay Institute of Technology, Greater Noida, Uttar Pradesh, India **Faculty, Management Institute, Raj Kumar Goel Institute of Technology, Ghaziabad, Uttar Pradesh, India Online published on 4 October, 2013. Abstract History considered being the ultimate asset, gold bullion bars have been used for centuries as a store of value. With limitations on its availability, every one from king to central banks has used gold investments as a foundation of their wealth. This metal trading is of high interest for analysts, researchers, more particularly to investors as it will present them with an opportunity to benefit financially by investing their resources. This study is an attempt to make a comparison between the most successful and widely used GARCH family of models to that of newly implemented but more famous among computer professionals Artificial Neural Networks (ANN) models. Eighteen various specifications of GARCH family of models and twenty various ANN models with four architectures are constructed to predict gold market return. Forecasting errors when calculated by using six forecasting error measures, it has been proved that the 3-5-1 (α=0.7 and ε=0.7) of ANN ranked best with minimum forecasting error. MSE is found to be best suitable for estimating forecasting errors for gold return series. Top Keywords ANN, GARCH family of models, EGARCH, Forecasting, GJR-GARCH, Volatility. Top |