FinRL

FinRL is an open-source framework that provides deep reinforcement learning tools for quantitative trading, from data ingestion to live execution.

FinRL is an open-source platform that enables the development of financial trading agents using deep reinforcement learning. It functions as a comprehensive stack covering market simulation, learning algorithms, performance evaluation, and live trading execution. Users ingest data from sources like Yahoo Finance, Alpaca, CCXT, or WRDS, and wrap it in Gymnasium-compatible environments. The platform allows for training models via PPO, SAC, A2C, DDPG, TD3, and DQN using backends like Stable-Baselines3, ElegantRL, and Ray RLlib. It supports multi-asset trading including equities, ETFs, crypto, futures, and FX, allowing practitioners to move from research backtesting to live deployment without rewriting code. Developed by the AI4Finance Foundation, it provides a consistent API for defining market states, action spaces, transaction costs, and slippage controls.

FinRL at a glance

Best for: Finance, Researcher
Runs on: open source, API
Pricing: Free open-source plan; commercial edition available via inquiry
Released: July 2020
Category: AI research tool
Website: finrl.ai

Key features

Deep reinforcement learning for quantitative trading
Integration with Stable-Baselines3, ElegantRL, and Ray RLlib
Support for equities, ETFs, crypto, futures, and FX
Modular data layer and environment architecture
Backtesting with Sharpe ratio and drawdown analysis
Live trading deployment for policy objects
Deterministic seeding for reproducibility

Use it for

Developing and training deep reinforcement learning trading agents
Running backtests on historical market data
Paper trading strategies before live execution
Executing trades across multi-asset portfolios

Alternatives to FinRL

FinGPT — Access open-source financial language models and APIs at a low cost.
FinCatch Studio — Simulates forward scenarios and stress-tests financial models on primary facts.
TradingAgents — It uses specialized AI agents to collaboratively evaluate market conditions.
FinSight AI — It features recoverable long-running workflows and evidence-grounded hybrid retrieval.
AI Berkshire — It forces clear conclusions with specific price ranges and multi-agent debate.
FinRobot — Automates the creation of professional financial analysis and research reports.
Vibe-Trading — Provides reproducible, inspectable research artifacts for financial markets.
Clawock — It keeps an accumulating, verifiable record of agent decisions and outcomes without hand-tuning.
BellRun — Influence an autonomous AI intraday stock trader in a simulated rodeo market game.
rLLM — Enables reinforcement learning training for AI agents with minimal code changes.
Investor Skills — It connects wallets and brokerages to execute trades with AI agents.
Fincept Terminal — Provides institutional-grade AI research and quant tools for a fraction of legacy terminal costs.

All FinRL alternatives