RLTrader provides a cryptocurrency trading environment utilizing deep reinforcement learning and OpenAI's Gym. It's designed for researchers and developers interested in building and optimizing trading agents, offering a framework for training and testing strategies on historical market data.

How It Works

The project employs deep reinforcement learning algorithms, specifically PPO2 from the stable-baselines library, to train trading agents. It uses historical cryptocurrency data to simulate trading scenarios, allowing agents to learn optimal buy/sell strategies. The approach emphasizes hyperparameter optimization via an optimize.py script, which stores promising configurations in an SQLite database for subsequent agent training and testing on unseen data, aiming for robust generalization.

Quick Start & Requirements

• Install with pip install -r requirements.txt (for NVIDIA GPU) or pip install -r requirements.no-gpu.txt (for CPU).
• Requires Python 3.6.8 (for Windows conda install).
• NVIDIA GPU with CUDA is recommended for performance.
• Data can be downloaded from https://www.cryptodatadownload.com/data/northamerican/.
• Official Medium articles detailing the agent creation and optimization are available: Article 1, Article 2.

Highlighted Details

• Hyperparameter optimization using optimize.py to find effective trading strategies.
• Agents are trained and then tested on separate datasets to verify generalization.
• Docker and Vagrant environments are provided for easier setup and testing.
• Google Colab support is available for GPU-accelerated training.

Maintenance & Community

• Development has slowed in favor of the successor framework, TensorTrade (https://github.com/notadamking/tensortrade).
• A Discord server is available for support and discussion: https://discord.gg/ZZ7BGWh.

Licensing & Compatibility

• The repository does not explicitly state a license in the README.
• Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project is noted as a predecessor to TensorTrade, with development having slowed. The README does not specify a license, which may impact commercial use. The optimization process can be time-intensive, potentially taking hours to days depending on hardware.