Hackalytics

What it is

Hackalytics looks for price differences across sportsbooks and prediction markets. Games, odds, predictions, and stake sizing all run locally. You do not need paid data services or cloud ML to try it, and it can fall back to sample data if the live APIs are down.

How it works

1. 1

   The backend pulls games from public sports APIs like ESPN, MLB, and NHL. If those are unavailable, it generates odds from sample data so the app still runs.

2. 2

   A trained PyTorch model scores opportunities on CPU from a saved checkpoint.

3. 3

   Another pipeline checks Polymarket and Kalshi for gaps between Yes and No prices.

4. 4

   The React frontend on Vercel talks to a Python API. A monthly cron can refresh predictions when the backend is online.