In this paper, we forecast Bitcoin's returns and return jumps using a self-exciting process embedded in a stochastic volatility model. We show the existence of the jump clustering feature, which varies depending on the frequency of the data. In an out-of-sample setting, we use a particle filter...