
TurboTax Online
File your taxes with confidence, get your maximum refund guaranteed, and access expert help when you need it.

The Open-Source Framework for Reinforcement Learning in Quantitative Finance.

FinRL is the first open-source framework that provides a full-stack pipeline for deep reinforcement learning (DRL) in quantitative finance. Developed by the AI4Finance Foundation, it bridges the gap between financial engineering and AI by offering a modular architecture that separates data processing, environment creation, and agent training. In the 2026 market, FinRL stands as the backbone for automated trading systems, allowing developers to leverage high-performance DRL algorithms such as PPO, DDPG, and SAC. Its ecosystem has expanded to include FinRL-Meta, which follows a DataOps paradigm to provide real-time data access from various markets including stocks, crypto, and forex. The architecture is designed to be plug-and-play, integrating seamlessly with OpenAI Gym-style environments and high-level libraries like Stable Baselines3 and Ray RLlib. For enterprise-grade solutions, FinRL facilitates robust backtesting with realistic market constraints, including transaction costs and liquidity slippage, making it a critical tool for institutional strategy development and academic research in financial AI.
FinRL is the first open-source framework that provides a full-stack pipeline for deep reinforcement learning (DRL) in quantitative finance.
Explore all tools that specialize in optimize investment portfolios. This domain focus ensures FinRL delivers optimized results for this specific requirement.
Explore all tools that specialize in backtesting. This domain focus ensures FinRL delivers optimized results for this specific requirement.
A DataOps-based universe that provides unified access to diverse financial data sources with automated cleaning and feature engineering.
Ability to train multiple agents simultaneously to simulate competitive or cooperative market scenarios.
Combines multiple DRL algorithms (e.g., PPO + A2C) to leverage different market condition strengths.
Integration with ElegantRL for high-throughput training on massive datasets using CUDA.
Built-in support for Stockstats and TA-Lib to inject domain knowledge into the RL state space.
Modular interface to define rewards based on risk-adjusted metrics like Sortino or Calmar ratios.
Native integration with Optuna for Bayesian optimization of agent parameters.
Install the FinRL library using pip install finrl or clone the GitHub repository.
Configure the data provider using FinRL-Meta (e.g., Alpaca, Yahoo Finance, or Binance).
Define the state space, including price data and technical indicators like MACD or RSI.
Configure the action space, such as buy, sell, or hold for specific assets.
Define the reward function to optimize for Sharpe ratio or cumulative returns.
Initialize the environment using the OpenAI Gym-style interface provided by FinRL.
Select a DRL agent from supported libraries like Stable Baselines3 or ElegantRL.
Execute the training loop on historical training data with hyperparameter tuning.
Perform out-of-sample backtesting to validate model performance against benchmarks.
Deploy the trained model for paper trading using broker-specific APIs.
All Set
Ready to go
Verified feedback from other users.
"Highly praised by the quantitative research community for its modularity and extensive documentation, though noted for a steep learning curve for non-coders."
Post questions, share tips, and help other users.

File your taxes with confidence, get your maximum refund guaranteed, and access expert help when you need it.

Unlock the power of open finance with Truv's verification platform.

Automate financial close processes to reduce risk and improve efficiency.

Where the world does markets, joining 100 million traders and investors.

Automated crypto trading bot for maximizing profit.

AI-powered stock trading platform providing real-time momentum detection and actionable trading signals.