(Analysis of Bitcoin Volatility with Generalized Autoregressive Conditional Heteroskedastic Models)

Abstract

Purpose – In the last decade, digital innovations in the field of finance emerge mainly due to Blockchain technology. The most widely used product of Blockchain technology in the world is cryptocurrencies. Bitcoin draws attention to both market capitalization and transaction volume. The analysis of the volatility of Bitcoin has great significance both theoretically and practically. Therefore, the research aims to determine the most suitable volatility model in Bitcoin investment analysis and present a model suggestion that can be used for predicting the future. Design/methodology/approach – In this study, daily logarithmic return series between 29.04.2013-17.04.2019 are used to analyze Bitcoin volatility. While calculating the return series, daily closing prices of Bitcoin are based on. As a research method, generalized autoregressive conditional heteroskedastic (GARCH) models are tested. In this study, ARCH, GARCH, GJR / TARCH, EGARCH, APARCH, and CGARCH type models are compared for volatility modeling of Bitcoin logarithmic return series. Findings – The study found that the best method for the analysis of Bitcoin volatility is the EGARCH model. Another important finding of the study is that negative shocks are more effective on Bitcoin returns than positive shocks. Discussion – The study contributes to the literature in terms of analyzing the dynamics of Bitcoin volatility, which is a new and digital product, with various models and current data set. Bitcoin investors and researchers can predict returns by using the volatility equation in result of this research. Another output of the study is that negative news effects Bitcoin returns more than positive news. Using this information, traders will be able to take positions on Bitcoin prices.