This is the repository for Seasons of Code- WnCC IITB'21. https://docs.google.com/document/d/1BYZIN4ct_Ap34xpvxxxZrYDiUTBdTFO2wU14VRbhrYk/edit https://docs.google.com/document/d/1AiOjZM5y7BYH9zs_boZMupaAQ0KlAv9odD5Vv54bTpA/edit
The initial phase of the project was mainly learning based and involved learning in deep about reinforcement learning.
The initial learning experience included:
- Learnt about Neural Networks through academics during the initial period.
- Completed the 4 lecture videos of RL course by David Silver
- Reading about RL, kinds of algorithms and lots more from this blog- https://spinningup.openai.com/en/latest/
This gave me a working understanding of how RL works.
Going further down the timeline, I started learning more on RL through :
- Q Learning- through this video series: https://www.youtube.com/watch?v=yMk_XtIEzH8&list=PLQVvvaa0QuDezJFIOU5wDdfy4e9vdnx-7
- SUDARSHAN RAVICHANDRAN - HANDS-ON REINFORCEMENT LEARNING WITH PYTHON book
- Also tried and worked on these tutorials:
https://towardsdatascience.com/creating-a-custom-openai-gym-environment-for-stock-trading-be532be3910e
https://towardsdatascience.com/trade-smarter-w-reinforcement-learning-a5e91163f315
https://towardsdatascience.com/visualizing-stock-trading-agents-using-matplotlib-and-gym-584c992bc6d4
but couldn't complete so as tensortrade is not working anymore, so faced a lot of issues with the same.
1.py and 2.py are the files
- Read the paper:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3197726
and summarised about the same. - Also completed the first 2 weeks of the coursera course : https://www.coursera.org/learn/advanced-methods-reinforcement-learning-finance#syllabus
- Currently trying to complete this tutorial ( Debugging due to data discrepancy) https://github.com/borisbanushev/stockpredictionai
- Set the FinRL environment : https://github.com/AI4Finance-LLC/FinRL and now trying to run these: https://github.com/6-Billionaires/trading-gym
So basically the final task is in progress (FinRL and OpenAI gym)