Week 23 Trading Results
Our ML models traded 175 positions across 7 sports this week, with a 45.7% win rate and -$16.22 net P&L.
Every trade is executed on Polymarket and verified on-chain.
Performance by Sport
| Sport | Trades | Record | Win Rate | c/Trade | P&L |
|---|---|---|---|---|---|
| NHL | 5 | 5W-0L | 100.0% | 37.6c | +$5.62 |
| CS2 | 15 | 9W-6L | 60.0% | 15.6c | +$4.39 |
| WNBA | 16 | 7W-9L | 43.8% | 2.0c | +$2.49 |
| LOL | 18 | 8W-10L | 44.4% | 1.4c | +$0.02 |
| WTA | 4 | 0W-4L | 0.0% | -32.0c | -$4.57 |
| MLB | 55 | 26W-29L | 47.3% | -4.5c | -$7.80 |
| ATP | 62 | 25W-37L | 40.3% | -11.8c | -$16.37 |
Notable Wins
- Jakub Mensik (ATP): Bought at 24c, fair value 29c, edge +4c
- Moise Kouame (ATP): Bought at 25c, fair value 25c, edge +0c
- portlandfire (WNBA): Bought at 26c, fair value 32c, edge +6c
How It Works
Our models are trained on 60,000+ historical games and predict fair win probability from game state alone — score, time, period, Elo ratings. When our fair probability exceeds the market ask by 8c+, we buy and hold to settlement.