Skip to content
DSML.KZ
Publications

Code · Code Ocean · August 28, 2026

Multimodal Hybrid Deep Learning for Cryptocurrency Volatility Forecasting

Mohammed Alaa Ala’anzy, Yeskendir Kaliyev, Omar Alnaseri, Abdulmohsen Algarni

Mohammed Alaa Fadhil Al-hadeethi
Open via DOI

Abstract

Benchmarking framework for evaluating multimodal deep learning against econometric baselines in predicting cryptocurrency realized volatility across multi-step horizons (1, 3, and 7 days). Fuses market price data, NLP news sentiment (FinBERT), and on-chain network metrics into a sequential 1D-CNN, Bidirectional LSTM, and Multi-Head Self-Attention architecture. Includes Diebold-Mariano significance testing, regime-aware stress testing (LUNA & FTX crashes), and Monte Carlo Dropout uncertainty quantification.

Stock Market Forecasting MethodsFinancial Risk and Volatility ModelingBlockchain Technology Applications and Securityrealized volatilitymultimodal deep learningnews sentiment

More by Mohammed Alaa Fadhil Al-hadeethi