Forecasting volatility in the financial markets: a comparison of alternative distributional assumptions

This article analyses the volatility forecasting performance of the GARCH models based on various distributional assumptions in the context of stock market indices and exchange rate returns. Using rollover methods to construct the out-of-the-sample volatility forecasts, this study shows that the GARCH model combined with the logistic distribution, the scaled student's t distribution and the Riskmetrics model are preferable both in stock markets and foreign exchange markets. The exponential power and the mixture of two normal distributions are, however, less recommended. Furthermore, a complex distribution does not always outperform a simpler one, although the exact ranking depends on the application of underlying assets and the performance statistics being used.