finrl

FinRL is the original end-to-end financial reinforcement learning library maintained by the AI4Finance community. It is structured around market environments, DRL agents, and financial applications, and uses a train-test-trade workflow with separate training, testing, and trading scripts. The repository is intended for learners, developers, and researchers; users building modern production or live-trading systems are directed to the separate FinRL-X / FinRL-Trading project. The included stock-trading tutorial shows how to download Dow 30 data from Yahoo Finance, add technical indicators such as MACD and RSI plus VIX and turbulence data, split the dataset into training and trading periods, train agents, and run a backtest. The example trains A2C, DDPG, PPO, TD3, and SAC agents through Stable Baselines3, saves the resulting models, and compares agent performance with Mean Variance Optimization and the DJIA index. The package includes application examples for stock trading, cryptocurrency trading, high-frequency trading, portfolio allocation, imitation learning, and the NeurIPS 2018 stock-trading task. Agent integrations include Stable Baselines3, ElegantRL, and RLlib. The meta layer provides data processors, preprocessing, ticker configuration, and environment implementations for stocks, cryptocurrencies, and portfolio allocation. FinRL documents multiple data sources, including Yahoo Finance, Alpaca, Binance, CCXT, Akshare, Baostock, Tushare, JoinQuant, RiceQuant, WRDS, IEX Cloud, QuantConnect, EOD Historical Data, Sinopac, and FXMacroData. Processors generally work with OHLCV data and can produce prices and technical indicators such as MACD, Bollinger Bands, RSI, DX, and moving averages; users can also add custom features. The repository provides installation instructions, examples, unit tests for environments and data downloaders, documentation links, and citation information. It is released under the MIT License, with a separate trademark notice for the FinRL name and logo. The README states that the code is shared for academic purposes and does not constitute financial advice or a recommendation to trade real money.