Forecasting Bitcoin Volatility Using Two-Component CARR Model
Xinyu Wu, NIU SHENGHAO, Xie Haibin · ECONOMIC COMPUTATION AND ECONOMIC CYBERNETICS STUDIES AND RESEARCH · 2020
In this paper, we propose an extension of the range-based CARR model, the two-component CARR (CCARR) model to model and forecast the Bitcoin volatility. The extension inherits the strength of the original range-based CARR model, its capability of exploiting intraday information from the high and low prices to estimate volatility. Moreover, the CCARR model has the capacity to accommodate the long memory volatility. Empirical results show that the CCARR model outperforms the CARR model and the return-based GARCH and twocomponent GARCH (CGARCH) models in forecasting the Bitcoin volatility. The results highlight the value of using price range and including a second component of the conditional range for forecasting the Bitcoin volatility.