An Application of Deep Reinforcement Learning to Algorithmic Trading

This codebase builds upon the original implementation available at https://github.com/ThibautTheate/An-Application-of-Deep-Reinforcement-Learning-to-Algorithmic-Trading.

It supports and extends the experimental framework presented in the following research paper:

Thibaut Théate and Damien Ernst.
An Application of Deep Reinforcement Learning to Algorithmic Trading (2020).
https://arxiv.org/abs/2004.06627

As explained in the report, the modified, improved, and fixed files are: TDQN.py, dataDownloader.py, main.py, tradingEnv.py, and tradingPerformance.py.

Additionally, the file TActorCritic.py implements a new algorithm (reusing the TDQN.py as template) that was not included in the original report.

Dependencies

The dependencies are listed in the text file "requirements.txt":

Usage

To run the code, execute the main.py file.

Selecting an Algorithm

To run a different algorithm, modify line 29 in main.py by changing the value of default='TActorCritic' to the algorithm you want to run. The two available options are:

• 'TDQN'
• 'TActorCritic'

Changing Company or Index

To run the model on a different stock or index, change the value of default='S&P 500' to the name of the desired stock or index (e.g., 'Apple').

• Note: The full list of available options can be found in the dictionaries: companies, indices, and stocks in the top of the file tradingSimulator.py.

Changing Dates

If you want to change the date range, modify the values in the dates.py file.