Date created: | Last Updated:

Category: Project

Description: This paper introduces a hybrid machine learning framework for algorithmic trading in cryptocurrency and foreign exchange (FX) markets. The framework integrates high-frequency OHLCV data, order book microstructure signals, and sentiment analysis powered by Large Language Models (LLMs). It combines LightGBM, LSTM, and Transformer networks with a regime-switching mechanism and stacking ensemble. To ensure reliability, the study applies walk-forward validation, realistic cost and slippage modeling, and ATR-based risk controls. Backtests (2019–2024) show improved Sharpe ratio, lower drawdowns, and higher profit factors compared to benchmarks. Results highlight the value of LLM-driven sentiment features, offering a scalable and adaptable solution for crypto and FX systematic trading.