Model status
Every week, how the model's picks did: wins and losses, against the 52.4% that breaks even at standard prices. Every pick was published before the games and graded against the box score. The whole record.
Above the 52.4% that breaks even at standard prices.
Week by week
Win rate; the line is the break-evenIs the confidence honest?
The chartThe confidence holds up: picks win about as often as the model says.
Since Tue 20 Oct: picks grouped by the chance the model gave them, against how often each group won. This week alone: cautious by 6.7 points, on 315 picks.
| By stat | W–L | Won | Units |
|---|---|---|---|
| Points | 38–7 | 84.4% | +27.5 |
| Rebounds | 85–31 | 73.3% | +46.3 |
| Assists | 99–63 | 61.1% | +27.0 |
| Strong picks | 18–4 | 81.8% | +12.4 |
| Solid picks | 68–17 | 80.0% | +44.8 |
| Lean picks | 136–80 | 63.0% | +43.6 |
| Overs | 64–22 | 74.4% | +36.2 |
| Unders | 158–79 | 66.7% | +64.6 |
Best calls
The widest winsWorst misses
The widest losses, in public tooThe automated bankrolls this week
All bankrollsVirtual bankrolls that bet the model's picks by themselves, before the games. Past results are no promise.