Optiver challenge - sentiment analysis algo trading

Optibook

Inspiration

We had fun participating in the Optiver Challenge. We have no experience with algorithmic trading and we found the topic very interesting to learn.

What it does

We bult Ann agent that can trade in an exchange using Optibook API. The agent listens for posts (tweets), and tries to detect to its relevance to the stocks and sentiment, and make smart decisions based on this information.

How we built it

We participated in the workshop by Optivert and mainly used their basic agent ad we had no background in this domain at all. We used GPT for the classification of the relevant stock, and facebook/bart-large-mnli for the sentiment analysis.

Challenges we ran into

We have no knowledge in also trading, and understanding how to make profit was not easy and still is a challenge. Also one of us (half of the team) got sick in the middle of the hackathon which imposed a challenge.

Accomplishments that we're proud of

We learned about also trading and market making, and had some fun trying our skills in sentiment analysis. Even though our agent is not the best and loses, we still had fun and learned a lot.