Forecasting the Volatility of Bitcoin : The Importance of Jumps and Structural Breaks

This paper studies the volatility of Bitcoin and determines the importance of jumps and structural breaks in forecasting volatility. Using high-frequency data, we perform a model-free decomposition of realized variance into its continuous and discontinuous components, positive and negative semivariances, signed jumps and leverage components. We show the importance of this decomposition in the in-sample regressions using eighteen competing heterogeneous autoregressive (HAR) models. In the out-of-sample setting, we find that the HARQ-F-J model is the superior model, indicating the importance of the temporal variation and squared jump components at different time horizons. We also show that the HAR models with structural breaks outperform models without structural breaks across all forecasting horizons. Our results are robust to an alternative jump estimator and estimation method