Forecasting Volatility with Component Conditional Autoregressive Range Model

The purpose of this research is to achieve a suitable model for forecasting volatility of a broad market index. In this paper the CCARR model is proposed for forecasting volatility and its estimation results are compared with the popular GARCH, CGARCH and CARR models. The model is intuitive and convenient to implement by using the maximum likelihood estimation method. GARCH and CGARCH models use price returns and CARR and CCARR models use price ranges to predict volatility. CCARR and CGARCH models assume that the price range comprises both a long run (trend) component and a short run (transitory) component, which has the capacity to capture the long memory property of volatility. Daily data of The Tehran Stock Exchange index from 2009 to 2022 including high, low and close prices are used. The results show that range-based models such as CARR and CCARR models fit the data better than return-based models. Also two-component models fit the data better than one-component models. In general, the CCARR model generates more accurate out of sample volatility forecasts than the popular GARCH, component GARCH and CARR models.