willinvest / neuralhft

A Pythonic deep reinforcement learning library for high frequency trading

                              

neuralHFT: Deep reinforcement learning for high frequency trading.

neuralHFT is a pythonic libarary that is written and optimized for high-frequency stocks and crypto trading. neuralHFT is compatible with state of the art DRL algorithms implemented with stable-baselines3, and PyTorch out of the box and with no boilerplate code required. Seamless integration with Alpaca API is supported by default allowing standalone and end-to-end support from training to launching HFT algorithms in your live/paper trading account. The Alpaca API provides low-latency, commission-free trading of stocks and crypto assets, making it an ideal platform for high-frequency trading strategies.

The end-to-end design enables users with granular control over dataset creation, feature engineering, model architecture design, market simulation, reward shaping and low level control over training in a clean and pythonic fashion. This enables users to shift focus from time-consuming initial setup and third party service configurations to more rapid prototyping, and model discovery.