Harpoon

• 

  UI

• 

  UI 2

Inspiration

We are heavily inspired by quantitative analysis on the current US stock market and wanted to make something that has a similar effect on this emerging market space. We also are amazed by previous cases of insider trading like the bet of "Will there be a super bowl streaker?" placed by the streaker themselves.

What it does

Pulls real time data from Polymarket's API and performs high level analysis upon it. Gives users the largest trades of the data, analytically sorted markets, and AI suggested trades with the best chance of cashing out.

How we built it

We built it using Node.js, Supabase, and Cursor. We split our project into two repos. The first of which holds all of the frontend code and small calls to our API. The second one is our backend heavy code for our artificial intelligence training and Railway deployed market collector.

Challenges we ran into

The API that Polymarket gives is very difficult to work around. It took many hours of scanning through syntax to understand the correct parsing and calls needed to make our application run. We also had trouble getting backtesting information and had to compile this information through previous trade history ourselves as this is all Polymarket stores on previous markets with the Gamma API.

Accomplishments that we're proud of

The ML agent that can give accurate information on undervalued trades. It does have a positive return on investment when ran through many markets. We are also proud of the UI and design choices we employed.

What we learned

We learned how to create an independent AI agent backtested upon information.

What's next for Harpoon

The future of our product would be to integrate more prediction markets beyond Polymarket. The obvious one would be Kalshi but more are beginning to enter the scene. We want to provide analysis and find differences in trades between these sites and provide users with a one stop terminal for informed prediction market trading. The main feature and future success of this site as a startup would be to add a trade along feature, that connects to user wallets and executes instantaneous trades alongside designated users, markets, or our AI prediction agent.

Built With

• nextjs
• polymarket
• postgresql
• python
• railway
• supabase
• typescript
• vercel