As a Data Science Master's student at Rutger's University. I took upon an Independent Study under the guidance of Professor Charles W. Cowan (a great mentor). The goal of this study was to design and develop a Stock Trading Agent using Reinforcement Learning techniques which earn's reward/profit at the end of a time interval.
Over the course of a semester, I learned and built this system using:
- Simulation Techniques to generate stock ticker data.
- Built a Python module to simulate stock market to facilitate buy/sell of more than one stock over the period of time.
- Implementing various Deep Q-Learning Networks using Tensorflow in Python.
- Trying out various ways of feature engineering and using various reward functions to improve models performance.
To successfully create a profitable Trading Agent in the simulated environment.
I am using this github repository as a medium to provided detailed description of my work in the form of a report.
Please reach out to me at email (mandhyanaayu.gmail.com), if you have any questions or just want to have a conversation about Reinforcement Learning or the Trading agent.
Best,
Aayush Mandhyan.