Analysis of Bitcoin’s volatility using HAR-GARCH-X models†
Seungmi Lee1 · Chanbin Oh2 · Eunju Hwang3
123Department of Applied Statistics, Gachon University
Correspondence to: † This research was supported by the National Research Foundation of Korea (Grant NRF-2023R1A2C1005395).
1 Undergraduate student, Department of Applied Statistics, Gachon University, Gyeonggi-do 13120, Korea.
2 Undergraduate student, Department of Applied Statistics, Gachon University, Gyeonggi-do 13120, Korea.
3 Corresponding author: Professor, Department of Applied Statistics, Gachon University, Gyeonggi-do 13120, Korea. E-mail: ehwang@gachon.ac.kr
1 Undergraduate student, Department of Applied Statistics, Gachon University, Gyeonggi-do 13120, Korea.
2 Undergraduate student, Department of Applied Statistics, Gachon University, Gyeonggi-do 13120, Korea.
3 Corresponding author: Professor, Department of Applied Statistics, Gachon University, Gyeonggi-do 13120, Korea. E-mail: ehwang@gachon.ac.kr
Received August 9, 2024; Revised September 13, 2024; Accepted September 20, 2024.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
- Abstract
- This study considers time series models that combine the HAR model and GARCH/TGARCH models for forecasting analysis of bitcoin volatility. Furthermore, HAR-GARCH-X models with exogenous variables are proposed by adopting each of three financial factors: stock price index, volatility index, and won/dollar exchange rate. These three factors are predecessors of bitcoin volatility, which is revealed through the Granger causality test. The proposed models can be useful time series models that help to better understand the volatility structure of bitcoin with conditional heteroscedasticity and volatility clustering as well as include the affects of main financial factors. The model-fit criteria such as AIC and BIC report that the HAR-GARCH-family models significantly improve the existing GARCH-family models. In order to compare the predictability of the HAR-GARCH-family models, MSE, RMSE, MAE and MAPE are evaluated to assess the forecasting performance of the models. Among the total of eight models of HAR-GARCH-family, the HAR-GARCH-X (VOL) model, which has the volatility index as an exogenous variable, was selected as the optimal model with the best predictive performance. This study contributes greatly in that the model-fit and predictability are improved by involving the HAR model and exogenous variables in the models.
- Keywords : Bitcoin, exogenous variable, HAR-GARCH model, HAR-GARCH-X model, volatility
Full Text(PDF) Free
