Buy High Sell Low

Inspiration

We were inspired by the challenge sponsor, the interesting concepts you can learn about in this area, and our interest for NLP

What it does

Market maker trading bot reacting to news

How we built it

• Starting with the simplest basic case, then learning about concepts step by step and improving.
• The news sentiment model is a FinBERT model with a replaced classification layer that we trained on the provided news data and some synthetic data.

Challenges we ran into

We had to learn everything on the go

Accomplishments that we're proud of

We got a quite well-working model although it was the first time we ever tried creating a market making trading bot

What we learned

market making, arbitrage, retraining a language classification model

What's next for Buy High Sell Low

• exploring if we can train the classification model to output an intensity instead of only the movement direction and if we can make use of that

Built With

• python