Machine Learning-Based Bitcoin Trading Under Transaction Costs: Evidence From Walk-Forward Forecasting
arXiv:2606.00060v1 Announce Type: cross Abstract: This paper investigates whether machine learning forecasts of hourly BTC-USDT returns can be converted into economically meaningful trading performance after transaction costs. Using approximately 70,000 hourly observations from 2018-2026, XGBoost, LSTM, and iTransformer are evaluated in a 27-fold walk-forward protocol. All three models produce positive gross trading performance in selected configurations, but naive sign-based strategies fail once transaction costs of ten basis points are imposed. A cost-aware execution filter, which prevents t
The proliferation of advanced machine learning models and increasing data availability for cryptocurrency markets are enabling more sophisticated algorithmic trading strategies.
This research provides empirical evidence that AI can generate economically meaningful returns in volatile assets like Bitcoin, even considering significant transaction costs, indicating a maturation of AI in financial applications.
The perceived efficiency and predictability of crypto markets are evolving, as AI models begin to overcome frictional trading costs, potentially increasing institutional participation and liquidity through algorithmic means.
- · Sophisticated algorithmic trading firms
- · AI/ML developers specializing in finance
- · High-frequency traders in crypto markets
- · Retail traders with high transaction costs
- · Hedge funds reliant on less sophisticated models
- · Exchanges with high transaction fees
Increased deployment of AI-driven trading strategies across various cryptocurrency pairs beyond BTC-USDT.
Greater competition among AI models leading to higher market efficiency and potentially tighter spreads.
Regulatory scrutiny of AI-driven market manipulation and the potential for flash crashes due to autonomous trading agents.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG