While on vacation I tried to learn reinforcement learning for finance. This is the code I came up with that accompanies the blog post
In a nutshell, I tried to make an agent that picks a portfolio of stocks and take transaction costs into account. It doesn't really work but I learned a lot and if you submit some PRs it will get better
main.ipynb is a notebook that I used to run the code and see what's happening SingleSine.ipynb is the first notebook I did, with experiments on a single sine wave. learners is a directory you can add more learning algorithms to learers/a2c is an actor critic implementation I copy pasted from Denny Britz env has stuff related to the environemt env/pricegenerator.py has code that makes a cool synthetic market, that looks random but has learnable stuff going on env/Env.py is an environemnt for RL agents. It takes actions and gives rewards
All of this code is "doodle code" that I was just fooling around with. It's probably buggy, and it's a bad idea to copy paste any of it with the assumption it does what I say. However, it is a good idea to try for yourself, fix it up and make pull requests.