Overview

A stock trader powered with deep q-network.

Dependencies

Python 3. To install all the libraries, run pip3 install -r requirements.txt

Table of content

• agent.py: a Deep Q learning agent
• envs.py: a simple x-stock trading environment
• model.py: a multi-layer perceptron as the function approximator
• utils.py: some utility functions
• run.py: train/test logic
• requirement.txt: all dependencies
• data/: stock price data

How to run

To train a Deep Q agent, run python3 run.py --mode train. There are other parameters and I encourage you look at the run.py script. After training, a trained model as well as the portfolio value history at episode end would be saved to disk.

To test the model performance on validation set, run python3 run.py --mode validate --weights <trained_model>, where <trained_model> points to the local model weights file. Validation data portfolio value history at episode end would be saved to disk.

To test the model performance on test set, run python3 run.py --mode test --weights <trained_model>, where <trained_model> points to the local model weights file. Test data portfolio value history at episode end would be saved to disk.

Visualize

To visualize performance run python3 visualize.py --file <pickle_file>, where <pickle_file> points to the local portfolio value history file for a specific train/validate/test file.

Credits for starter code go to ShuaiW.