Every call, graded in public — sliced every way.
The published-call record: hits, misses, and voids, with the one number that matters — how often our stated probabilities actually came true. Daily Edition · refreshed nightly · data through Jun 17
All-time: our calls stated an average 29% chance and hit 32% of the time — across 560 graded calls.
On the diagonal = honest. Above = under-confident; below = over-confident. Point size = sample.
Stated vs. actual on the calls we published — reconciles to this ledger. (The model-wide calibration scoreboard, over every prediction, is on the Track Record page.)
| Bucket | Said | Hit | Δ | n | P&L |
|---|---|---|---|---|---|
| 0–10% | 8% | 24% (12/51) | +15.2 | 51 | -3.80u |
| 10–20% | 18% | 27% (41/152) | +9.4 | 152 | +6.20u |
| 20–30% | 24% | 27% (39/147) | +2.6 | 147 | +2.93u |
| 30–40% | 35% | 33% (31/94) | -1.6 | 94 | — |
| 40–50% | 44% | 38% (16/42) | -5.9 | 42 | -1.07u |
| 50–60% | 55% | 55% (11/20) | +0.3 | 20 | +0.20u |
| 60–70% | 65% | 57% (26/46) | -8.1 | 46 | -0.76u |
| 70–80% | 74% | 57% (4/7) | -17.3 | 7 | — |
| 90–100% | 92% | 0% (0/1) | -91.7 | 1 | — |
| Market | Hit rate | Record | Said | Calib Δ | Voids | P&L |
|---|---|---|---|---|---|---|
| Total bases — 4+ | 29% | 17–42 | 19% | +10.0 | 1 | — |
| K — 6+ | 43% | 25–33 | 46% | -2.9 | 2 | -0.87u |
| K — 8+ | 14% | 8–50 | 23% | -9.2 | 2 | — |
| ER — 2 or fewer | 53% | 30–27 | 66% | -13.0 | 3 | -0.76u |
| Win | 30% | 17–40 | 35% | -5.4 | 3 | — |
| Quality start | 40% | 23–34 | 31% | +8.9 | 3 | — |
| Hits — 2+ | 30% | 17–40 | 25% | +5.1 | 3 | -4.00u |
| SB — 1+ | 20% | 11–43 | 16% | +4.2 | 6 | +9.12u |
| HR — 1+ | 25% | 13–39 | 8% | +16.7 | 8 | -3.80u |
| RBI — 1+ | 37% | 19–32 | 20% | +17.3 | 9 | +4.01u |
Profit-and-loss vs. the median captured book line, 1-unit flat stakes — All-time. Calls with no captured line are excluded (not counted as a loss).