PyTorch-DDPG-Stock-Trading An implementation of DDPG using PyTorch for algorithmic trading on Chinese SH50 stock market, from Continuous Control with Deep Reinforcement Learning. Environment The reinforcement learning environment is to simulate Chinese SH50 stock market HF-trading at an average of 5s per tick. The environment is based on gym and optimised using PyTorch and GPU. Need only to change the target device to cuda or cpu. The environment has several parameters to be set, for example: the initial cash is asset, minimum volume to be bought or sold is unit, the overall transaction rate is rate and the additional charge on short position is shortrate (which genuinely exists in Chinese stock market). Model The Actor-Critic model is defined in actorcritic.py with act and target networks for them both. Complying to the original DDPG algorithm, the target networks are updated using soft-copy. The train-on-data process is same as the original DDPG algorithm using SARSAs from memory buffer. The policy gradience is fetched from the very first layer between actor & critic and directed t

⚠ No verified equity curve — no track-record source connected.

No score history is stored yet — only the current score is shown.

⚠ No live verification account connected — ask for proof before buying.

Alerts on changes: coming soon

Prop-firm compatibility not provided.

Data-completeness & trust index (not a profitability rating)