Modelling and Forecasting Bitcoin Returns Using Alternative GARCH Models: Evidence from International Cryptocurrencies
International Journal of Enhanced Research in Management & Computer Applications · 2026 · 15 (10) · 01-16
DOI: 10.55948/ijermca.15102601 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Enhanced Research in Management & Computer Applications, 2026. Verfasst von Nagendra Marisetty.
Abstract (Originalsprache)
This study examines the monthly returns and volatility characteristics of ten global cryptocurrencies, namely Bitcoin, BNB, Dogecoin, Ethereum, Solana, Tether, TRON, USDC, XRP and Zcash during 2021–2025. The study employs descriptive statistics, the Pearson correlation coefficient, and stationarity tests, namely the Augmented Dickey–Fuller (ADF), DF-GLS, Phillips–Perron (PP) and KPSS tests. The descriptive statistics indicate that chosen cryptocurrencies exhibit significantly distinct return and risk characteristics. Dogecoin has the highest average monthly return and standard deviation. Tether and USDC have relatively smaller fluctuations. Correlation analysis shows significant positive correlations between some cryptocurrencies, e.g., Bitcoin and Ethereum, and negative correlations between a couple of selected assets. The ADF, DF-GLS and PP tests indicate that all ten series of monthly returns are stationary. The study also investigates the relationships between the returns of the selected cryptocurrencies and the returns of Bitcoin using six volatility models, namely GARCH (1, 1), EGARCH (1, 1), PARCH (1, 1), CGARCH (1, 1), FIGARCH (1, 1) and FIEGARCH (1, 1). The findings indicate that Ethereum maintains a consistently positive and statistically significant association with Bitcoin returns for the first five model specifications, while the estimated relationships of other cryptocurrencies vary across the models. The ARCH and normality tests reported are indicative of satisfactory performance of the GARCH (1, 1), EGARCH (1, 1), PARCH (1, 1), CGARCH (1, 1) and FIGARCH (1, 1) models as indicated by the residual diagnostic tests. Model evaluation results show that GARCH performs best in terms of adjusted R-square, regression standard error, RMSE and MSE. EGARCH (1, 1) has the lowest AIC, SC and HQC whereas FIGARCH (1, 1) has the lowest MAPE. Taking into account joint goodness-of-fit and predictive performance measures GARCH (1, 1) is the overall preferred model. The relative performance of the models however depends on the evaluation criterion. The findings contribute to the knowledge of cryptocurrency return dynamics, interrelationships and volatility modeling during the study period.
Zitieren
APA
Marisetty, N. (2026). Modelling and Forecasting Bitcoin Returns Using Alternative GARCH Models: Evidence from International Cryptocurrencies. International Journal of Enhanced Research in Management & Computer Applications, 15(10), 01-16. https://doi.org/10.55948/ijermca.15102601
Vancouver
Marisetty N. Modelling and Forecasting Bitcoin Returns Using Alternative GARCH Models: Evidence from International Cryptocurrencies. International Journal of Enhanced Research in Management & Computer Applications. 2026;15(10):01-16. doi:10.55948/ijermca.15102601
BibTeX
@article{Marisetty2026,
author = {Marisetty, Nagendra},
title = {Modelling and Forecasting Bitcoin Returns Using Alternative GARCH Models: Evidence from International Cryptocurrencies},
journal = {International Journal of Enhanced Research in Management & Computer Applications},
year = {2026},
volume = {15},
number = {10},
pages = {01--16},
doi = {10.55948/ijermca.15102601}
}
EUVIMED