FinRL
Deep reinforcement learning library for trading research and teaching.
GitHub statistics as of 2026-11-10. Status: active = commits in the last 90 days, slowing = in the last year, inactive = older, archived = read-only.
Overview
FinRL wraps market data into reinforcement-learning environments and trains agents with standard algorithms such as PPO, A2C and SAC for stock trading, portfolio allocation and crypto. It is mainly a research and education project with many notebooks; reported results depend heavily on the test period and settings.
Best for: Learning and experimenting with reinforcement learning on market data.
Strengths
- Many ready-made environments and tutorials
- Covers several RL algorithms
- Active academic community
Limitations
- Research code, not a production trading system
- RL results are easy to overfit; out-of-sample checks are essential
- Mostly Jupyter notebooks
Comparisons
Alternatives in AI and machine learning
All frameworks · Descriptions are our own summary and may lag behind each project's documentation. Check the project's license before using it in a product. Nothing here is investment advice.