FinRL-Trading is the home of FinRL-X, an AI-native infrastructure for quantitative trading built by the AI4Finance Foundation. It unifies data processing, strategy composition, backtesting and live broker execution in one modular system, so a strategy you research in a notebook is the same code that trades in production.
The project is open source on GitHub with about 3,840 stars and active development in 2026. FinRL-X succeeds the original FinRL framework with a modernized architecture designed for the LLM and agentic AI era. Its weight-centric design makes the target portfolio weight vector the single interface between strategy logic and execution, which keeps backtests deployment-consistent. The repository includes reinforcement-learning strategies such as PPO, A2C and DDPG, ensemble methods, stock-selection models and portfolio-allocation tools, together with the accompanying research paper. The project grew out of the widely cited FinRL research line and maintains an active community of contributors, example notebooks and strategy templates, so newcomers can start from a working end-to-end pipeline instead of assembling data loaders, environments and execution code from scratch. Documentation, tutorials and community support channels are linked from the repository, lowering the barrier for students encountering reinforcement learning for finance for the first time.
FinRL-Trading is free and open source under the Apache-2.0 license. There are no subscription tiers; costs are limited to your own compute, market data and brokerage accounts.
Quantitative researchers, finance students and algorithmic traders who want a production-minded open framework for training and deploying AI trading strategies.