What can explain the price, volatility and traded volume of Bitcoin?

Abstract

We study which variables can explain and predict the return, volatility and traded volume of the cryptocurrency Bitcoin. The explanatory variables which we investigate are return, volatility, traded volume, transaction volume, change in the number of unique Bitcoin addresses, the VIX index and Google searches for the term Bitcoin. As a volatility measure we use realized volatility calculated from high-frequency data. Studies about the price formation of Bitcoin have been conducted before, but our study is the first with an extensive analysis of the realized volatility of Bitcoin. We find that the heterogenous autoregressive model for realized volatility (Corsi, 2009), which has been recognized as a very suitable model for realized volatility of other assets, is suitable also for modeling the volatility of Bitcoin. Moreover, we find that traded volume improves volatility forecasts even further. We find that transaction volume can predict Bitcoin returns and the traded volume of Bitcoin can be predicted from Google searches.