A Bayesian study of changes in volatility of Bitcoin

Abstract

This paper is aimed at studying MS-GARCH model applied at Bitcoin. A Bayesian estimation of the model shows that Bitcoin's volatility can be modelled using two states of volatility, high and low. The modelled volatility is not stable over time. Twenty eight periods of high volatility were found, the largest period of volatility occurred during 2013. The findings help explain what happened during this high volatility periods.