High-Frequency Volatility Prediction for Bitcoin
Academic Conferences Series · 2026-04 · 21 (1)
DOI: 10.62381/acs.cess2026.06 ↗Kurz gefasst
Fachartikel in der Zeitschrift Academic Conferences Series, 2026-04. Verfasst von Junyu Wei.
Abstract (Originalsprache)
Cryptocurrency markets operate continuously and generate high-frequency order book data that provides detailed information about market liquidity, depth, and order flow. However, the high dimensionality and volatility of limit order book data make short-term price prediction challenging. This study investigates the prediction of future mid-price movements for BTC/USDT using minute-level limit order book data from Binance. The dataset contains 741,600 order book snapshots collected from January 2022 to May 2023, with five price levels on both the bid and ask sides. We construct key order book features, including mid-price, bid-ask spread, order imbalance, and market depth, and develop classification models to predict price direction across multiple forecasting horizons. The study provides a baseline framework for high-frequency cryptocurrency price prediction and establishes a feature engineering pipeline that can be further extended using advanced deep learning architectures, including Transformer and DeepLOB models.
Zitieren
APA
Wei, J. (2026). High-Frequency Volatility Prediction for Bitcoin. Academic Conferences Series, 21(1). https://doi.org/10.62381/acs.cess2026.06
Vancouver
Wei J. High-Frequency Volatility Prediction for Bitcoin. Academic Conferences Series. 2026;21(1). doi:10.62381/acs.cess2026.06
BibTeX
@article{Wei2026,
author = {Wei, Junyu},
title = {High-Frequency Volatility Prediction for Bitcoin},
journal = {Academic Conferences Series},
year = {2026},
volume = {21},
number = {1},
doi = {10.62381/acs.cess2026.06}
}
EUVIMED