AI4Finance-Foundation/FinRL

FinRL®: Financial Reinforcement Learning. 🔥

What it solves

FinRL provides an open-source framework for automating quantitative trading using deep reinforcement learning (DRL). It simplifies the process of developing, training, and testing trading agents that can make sequential decision-making policies in financial markets.

How it works

The framework is organized into three core layers:

  1. Market Environments: Simulates the financial market for the agent to interact with.
  2. DRL Agents: Implements various algorithms (such as A2C, DDPG, PPO, TD3, and SAC) to learn optimal trading strategies.
  3. Financial Applications: Applies the learned policies to specific tasks like stock trading, cryptocurrency trading, and portfolio allocation.

It follows a train-test-trade pipeline: downloading and preprocessing data (including technical indicators like MACD and RSI), training agents using libraries like Stable Baselines 3, and backtesting performance against baselines like the DJIA index.

Who it’s for

This repository is primarily intended for learners, developers, and researchers who want to experiment with financial reinforcement learning and create research prototypes.

Highlights

  • Broad Data Support: Integrates with numerous data sources including Yahoo Finance, Alpaca, Binance, and WRDS.
  • Coded Pipeline: Provides a complete end-to-end workflow from data acquisition to backtesting.
  • Algorithm Variety: Supports multiple DRL agents via Stable Baselines 3, ElegantRL, and RLlib.
  • Educational Focus: Includes tutorials and examples for stock trading workflows.

Related

  • Project
    Tencent-Hunyuan/UniRL

    UniRL is an Apache‑2.0 Python framework that applies a single reinforcement‑learning loop to many multimodal generative models (diffusion, video, LLM, VLM, and unified AR+diffusion). It offers composable trainers, multiple rollout back‑ends (vLLM‑Omni, SGLang), standard and novel RL algorithms (GRPO, Flow‑DPPO, DRPO, CPPO), and ready‑to‑run recipes for dozens of state‑of‑the‑art models. Users launch training via Hydra configs, and the system handles distributed rollout, scoring, advantage computation, and weight syncing.

  • Dispatch
    PipelineRL: Optimizing LLM Reinforcement Learning via Inflight Weight Updates

    Hugging Face and ServiceNow Research have open-sourced PipelineRL, an experimental RL implementation that uses inflight weight updates to eliminate the trade-off between inference throughput and on-policy data collection.

  • Project
    AI4Finance-Foundation/FinRobot

    FinRobot is an open‑source, multi‑agent AI platform for high‑stakes financial analysis. It separates deterministic Python‑based valuation (DCF, LBO, etc.) from LLM‑generated narration, orchestrates specialized agents (data, analysis, modeling, synthesis, reporting, plus debate agents), and ships three versions: a production desktop app (V2), a self‑hosted web report generator (V1), and an educational AutoGen library (V0). The system integrates multiple market data sources, provides 56 analyst playbooks, and can be run locally via a CLI, a browser UI, or a native Tauri desktop client.

  • Project
    AI4Finance-Foundation/FinRL-Meta

    A market environment and benchmark layer for financial reinforcement learning that provides automated data pipelines and standardized environments to develop and evaluate trading agents.

  • Project
    alibaba/ROLL

    ROLL is an open‑source library for reinforcement‑learning training of large language models at scale. It provides Ray‑based distributed execution, supports many RL algorithms (PPO, GRPO, Reinforce++, etc.), works with Megatron‑Core, vLLM, SGLang, and LoRA, and offers pipelines for RL‑based fine‑tuning (RLVR) and multi‑turn agentic RL. The project includes extensive docs, hardware‑agnostic support (NVIDIA, AMD, Ascend), and built‑in observability.