Forecasting Cryptocurrency Volatility Using a MS-EGARCH Model

This study analyzes the cryptocurrency index (CRIX) using the generalized autoregressive conditional heteroskedasticity (GARCH) and extended GARCH models. Using the daily cryptocurrency index for July 31, 2014, to March 22, 2019, from the CRIX (https://thecrix.de/), we examine the CRIX return volatility forecast performance of three GARCH models. This empirical research investigates the importance of asymmetry in cryptocurrency volatility, which is not accounted for by the standard GARCH model; thus, asymmetric model variations are applied. The results show that the regime-switching model resolves the single-regime model’s problem of elevated forecasts for high-volatility periods. Additionally, we show that the forecasting performance of the Markov-switching exponential GARCH (MS-EGARCH) model is superior to that of other models. This suggests that the MS-EGARCH model outperforms other models in accounting for cryptocurrency index volatility. Hence, the regime-switching model, which applies asymmetry, has greater explanatory power than the standard GARCH model.