This is a toy project to implement A3C reinforcement learning in a stock market environment I stitched together over a weekend in Feb 2018.
The market environment is adapted from kh-kim, the A3C is based on Grezgo's async-rl project. The network concatenates an LSTM with a state vector for the portfolio.
Go to Model/A3C/ and open Run.ipynb to start training.
Run tensorboard --logdir=Model/A3C/Graph/
to inspect the performance of the model.