Reproduce “Integrating Large Language Models and Reinforcement Learning for Sentiment-Driven Quantitative Trading” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Wo Long, Wenxin Zeng, Xiaoyu Zhang, Ziyao Zhou
Researchers developed a quantitative trading system that integrates FinGPT-generated sentiment signals with traditional technical indicators using a Twin Delayed Deep Deterministic Policy Gradient (TD3) reinforcement learning algorithm. This RL-driven strategy achieved a 23.65% annualized return and a 1.38 Sharpe ratio in out-of-sample tests, surpassing a buy-and-hold benchmark's 17.17% return and 1.20 Sharpe ratio.
Key insights
- LLM-generated sentiment signals from financial news can provide incremental value when combined with traditional technical indicators for quantitative trading strategies.
- Reinforcement learning, particularly the TD3 algorithm, offers an effective framework for dynamically synthesizing heterogeneous signals (sentiment and technicals) into adaptive and robust trading decisions.
- Strict adherence to data cutoffs and masking of identifying information is essential to ensure the reliability and practical applicability of LLM-based financial research by mitigating look-ahead bias.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.