← back
Live
CS2 Prop Model
Quantitative model for esports player-prop betting.
- Status
- Live — forward-tested every day
- Stack
- Python · SQLite · NegBin regression · OCR vision pipeline
A full pipeline that imports HLTV match data, fits a hand-coded negative-binomial projection model, and evaluates daily prop slates from screenshots. Outputs a ranked CSV, builds parlay tickets, and auto-settles results from screenshots of finished bets.
The current Formula v1 era hits 69.6% win rate and +32.9% ROI on 79 out-of-sample walk-forward bets. The two validated UNDER rules (line ≥ 30 kills, shading > +3) hold 65.5% / +25% across three eras on 438 bets.
Adjacent to this is a Kalshi auto-bet runner that places small flat bets on the match-outcome model when edge crosses 5%.