Predicting volatility of cryptocurrencies: Deep learning and GARCH family models
Modern Finance · 20.03.2026 · 4 (1) · 95-113
DOI: 10.61351/mf.v4i1.370 ↗Kurz gefasst
Fachartikel in der Zeitschrift Modern Finance, online erschienen am 20.03.2026. Verfasst von Abdul Moiz und 1 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This paper examines the application of econometric models, deep learning architectures, and hybrid combinations of both methods for volatility forecasting in cryptocurrency markets. Using daily data on 10 major cryptocurrencies from 2020–2025, this work benchmarks GARCH-family models, deep learning models, and hybrid frameworks against a uniform 30-day realized volatility target under a strict walk-forward evaluation protocol. The findings indicate that traditional GARCH models perform adequately for relatively stable assets, whereas deep learning models improve accuracy by capturing nonlinear dynamics. Hybrid models, particularly those combining GARCH models with TDNN and GRU components, achieve the most accurate and statistically significant volatility forecasts. These findings highlight the added value of hybrid methodologies in cryptocurrency risk modeling.
Zitieren
APA
Moiz, A., & Raza, H. (2026). Predicting volatility of cryptocurrencies: Deep learning and GARCH family models. Modern Finance, 4(1), 95-113. https://doi.org/10.61351/mf.v4i1.370
Vancouver
Moiz A, Raza H. Predicting volatility of cryptocurrencies: Deep learning and GARCH family models. Modern Finance. 2026;4(1):95-113. doi:10.61351/mf.v4i1.370
BibTeX
@article{Moiz2026,
author = {Moiz, Abdul and Raza, Hassan},
title = {Predicting volatility of cryptocurrencies: Deep learning and GARCH family models},
journal = {Modern Finance},
year = {2026},
volume = {4},
number = {1},
pages = {95--113},
doi = {10.61351/mf.v4i1.370}
}
EUVIMED