trAIders

Inspiration

Applying RL algos models to build innovative mehtods

What it does

Uses PPO to learn how to make free money

How we built it

Using the available code base from Optiver, OpenAI gym and Stablebaselines3 We implement a custom environment using the Optiver commands

Challenges we ran into

RL takes long to learn and is doing incredibly silly things at the start. Things like self trades, selling things for like 3 cents when the price is 45.00. Definitely no free money...

Accomplishments that we're proud of

The team spirit. Getting the RL algo learning. Eventhough the model is terrible we are able to use benefit of the doubt as an excuse. The model 'totally' would've been amazing if it had a little more time to learn.