Modelling and forecasting volatility using models based on the volatility range

This paper shows advantages of using price range in volatility modeling and forecasting. It is known that price range, defined as a difference between the logarithms of the highest and the lowest price of an asset, is a useful volatility approximation. In this paper three different range-based models are compared with commonly used residual-based GARCH model in terms of goodness of fit and forecasting accuracy. Each model is estimated on daily data covering six currency pairs quoted to PLN. Despite being equally simple as residual-based GARCH model, range-based models generally perform better. Forecasts generated by range-based models are more precise, moreover they seems to better capture volatility clustering phenomenon.