M. A. Cebrián Hernández, E. Jiménez Rodriguez, J. M. Feria Domínguez
Bitcoin, the world's largest digital currency by market capitalization, is characterized by its highly volatility. The digital currency has experienced sharp price fluctuations since it was launched in 2009. The main aim of this paper is to predict such volatility by applying General Autoregressive Models with Conditional Heterocedasticity (GARCH) on a historical data set of Bitcoin's daily returns from (2016) to (2018), extracted from Datastream database.
We first run a basic GARCH model (1,1) in order to measure the effectiveness of the estimates provided by this method and, secondly, we contribute to the existing literature by applying a Multivariate GARCH model considering other exogenous variables highly correlated with the volatility of such cryptocurrency. Our results provide better volatility estimations when using a multivariate GARCH model with respect to univariate GARCH models.
Keywords: Cryptocurrencies, Bitcoin, volatility, Multivariate-GARCH, back-testing.
Scheduled
GT15-3 Risk Analysis
September 6, 2019 12:40 PM
I3L8. Georgina Blanes building