Back to board
2026-W32 · Aug 3 – Aug 9, 2026
Weekly accuracy report
Every one of the 36 games that finished this week was re-graded, scoring the 2367 probabilities that had been published against actual outcomes. Lower Brier is better; 0.25 is coin-flip level.
Brier (model)
0.1931
raw 0.1970
Calibration gain
−0.0039
lower is better
Hit rate
71.1%
base 32.7%
Calibration error
3.9%
raw 7.8%
Brier by day (lower is better)
By league
| League | Games | Scored | Brier |
|---|---|---|---|
| MLB | 26 | 2309 | 0.1917/ 0.1957 |
| WNBA | 9 | 45 | 0.2476/ 0.2463 |
| MLS | 1 | 13 | 0.2624/ 0.2624 |
What changed this week
- • Graded 1362 markets from yesterday. Shipped probabilities scored Brier 0.1869 (raw 0.1934; a coin flip is 0.25).
- • Graded 990 markets from yesterday. Shipped probabilities scored Brier 0.2009 (raw 0.2013; a coin flip is 0.25).
- • league:mlb:pitcher_k calibration trailed raw output on live games (Brier 0.1888 vs 0.1842, 257 samples). Pulled and quarantined.
- • Adopted a isotonic calibrator for league:mlb:batter_contact (hold-out Brier 0.2183 → 0.2012, 6122 samples).
- • Adopted a isotonic calibrator for baseball:batter_contact (hold-out Brier 0.2183 → 0.2012, 6122 samples).
- • Adopted a isotonic calibrator for *:batter_contact (hold-out Brier 0.2183 → 0.2012, 6122 samples).
- • Adopted a beta calibrator for league:mlb:batter_contact (hold-out Brier 0.2203 → 0.2050, 6310 samples).
- • Adopted a beta calibrator for baseball:batter_contact (hold-out Brier 0.2203 → 0.2050, 6310 samples).