Challenge
Algorithmic trading is a rapidly evolving domain, and Reinforcement Learning (RL) has emerged as a powerful approach within it. However, there remains a notable gap in effectively integrating text-based market information (e. g., financial news) and addressing risk management in RL-driven trading strategies.
Approach
We created an enhanced version of FinRLDeepSeek, a framework that integrates Large Language Model (LLM) signals into Reinforcement Learning (RL) for stock trading.
Outcomes
We empirically show through backtests on real market data that our new method consistently improves stability of stock valuations and risk-adjusted stock returns.
Project Status
Ongoing
Lead Researchers
Aydin Javadov
