A transparent look at how accurate MoneyBall Score AI is on MLB games — based on every verified game this season.
Verified
993
993 games
Overall accuracy
54.2%
538/993
Brier Score
0.249
Calibration gap
+3.2%p
Overconfident
When the AI says 60% confidence, does it actually win 60% of the time? Closer to the diagonal = better calibrated.
Average accuracy over the trailing 30 days for each date. Smooths out single-game luck to show the model's real trend. Above the 50% baseline means the model beats a coin flip.
Shows how closely the AI's predicted win probability buckets match actual outcomes.
Weekly Brier score over time. Lower is more accurate.
Accuracy matrix by day of week. Days with N<3 are grayed out as small samples.
| scoring_rule | 월 | 화 | 수 | 목 | 금 | 토 | 일 |
|---|---|---|---|---|---|---|---|
| 전체 | 55% 66/120 | 54% 89/164 | 56% 80/144 | 54% 55/101 | 51% 77/151 | 55% 88/159 | 54% 83/154 |
Accuracy matrix by scoring_rule (model weight version) × week. Time-axis sister view of the day-of-week matrix.
| scoring_rule | 8/17 주 | 8/24 주 | 8/31 주 | 9/14 주 |
|---|---|---|---|---|
| 전체 | 71% 67/95 | 50% 18/36 | 55% 17/31 | 70% 21/30 |
색상 — brand-700 ≥70% / brand-500 ≥60% / brand-300 ≥50% / brand-100 <50% / 회색 N<3 소표본. 최근 4주 cohort 시계열 비교.
Low confidence
~55%
58%
28/48
Medium confidence
55~65%
53%
451/857
High confidence
65%~
67%
59/88
Counted for any team involved in the game, home or away.
| Team | Games | Correct | Accuracy |
|---|---|---|---|
| Giants | 65 | 43 | 66.2% |
| Padres | 67 | 44 | 65.7% |
| Royals | 68 | 44 | 64.7% |
| Marlins | 64 | 41 | 64.1% |
| Rockies | 66 | 41 | 62.1% |
| Brewers | 68 | 40 | 58.8% |
| Mariners | 66 | 38 | 57.6% |
Shows how much the model over- or under-predicts each team's wins relative to their actual win rate. A large bias gap is a diagnostic signal for the model.
| Team | Predicted Win% | Actual Win% | Bias Gap | Accuracy |
|---|---|---|---|---|
| CubsUnder-predicted | 39.7% | 61.8% | -22.1%p | 45.6% |
| Cardinals | 60.3% | 44.1% | +16.2%p | 48.5% |
| Angels | 52.3% | 38.5% | +13.8%p | 55.4% |
| Astros | 41.5% | 55.4% | -13.8%p | 49.2% |
| Yankees | 63.1% | 49.2% |
Shows which opponents the AI predicts well or poorly for each team. n=1 results are dimmed (single-game sample). Home/away accuracy shown with each n.
How well each sabermetric factor predicted the actual game outcome — the rate at which the team a factor favored actually won.
Only games where home/away stat values differ are counted. Factors above baseline (54%) = model signal / below = possible noise. Out of n=993 verified games, valid-comparison count shown per factor.
| # | Factor | n (home/away) | Accuracy |
|---|---|---|---|
| 1 | Head-to-Headhead_to_head | 51 (29/22) | 59% |
| 2 | WARwar | 982 (507/475) |
색상 — 녹색 ≥60% / 노랑 ≥50% / 빨강 <50% / 회색 N<3 소표본. v1.6 anomaly + Sunday cap 효과 시각화.
| Phillies |
| 65 |
| 37 |
| 56.9% |
| Braves | 66 | 37 | 56.1% |
| Dodgers | 68 | 38 | 55.9% |
| Angels | 65 | 36 | 55.4% |
| Pirates | 67 | 37 | 55.2% |
| Rangers | 67 | 37 | 55.2% |
| Nationals | 67 | 37 | 55.2% |
| Guardians | 62 | 34 | 54.8% |
| Twins | 64 | 35 | 54.7% |
| Diamondbacks | 67 | 35 | 52.2% |
| Reds | 69 | 36 | 52.2% |
| Rays | 68 | 35 | 51.5% |
| Orioles | 63 | 32 | 50.8% |
| White Sox | 65 | 33 | 50.8% |
| Athletics | 67 | 34 | 50.7% |
| Red Sox | 67 | 34 | 50.7% |
| Mets | 67 | 33 | 49.3% |
| Astros | 65 | 32 | 49.2% |
| Cardinals | 68 | 33 | 48.5% |
| Yankees | 65 | 30 | 46.2% |
| Cubs | 68 | 31 | 45.6% |
| Tigers | 66 | 30 | 45.5% |
| Blue Jays | 64 | 29 | 45.3% |
| AL | 1 | 0 | (small sample) |
| NL | 1 | 0 | (small sample) |
| +13.8%p |
| 46.2% |
| Mariners | 56.1% | 43.9% | +12.1%p | 57.6% |
| Twins | 42.2% | 53.1% | -10.9%p | 54.7% |
| Athletics | 38.8% | 28.4% | +10.4%p | 50.7% |
| Reds | 36.2% | 46.4% | -10.1%p | 52.2% |
| Guardians | 56.5% | 46.8% | +9.7%p | 54.8% |
| Pirates | 41.8% | 50.7% | -9.0%p | 55.2% |
| Orioles | 42.9% | 50.8% | -7.9%p | 50.8% |
| Blue Jays | 39.1% | 46.9% | -7.8%p | 45.3% |
| Braves | 57.6% | 50.0% | +7.6%p | 56.1% |
| Red Sox | 61.2% | 68.7% | -7.5%p | 50.7% |
| Giants | 47.7% | 41.5% | +6.2%p | 66.2% |
| White Sox | 56.9% | 50.8% | +6.2%p | 50.8% |
| Rangers | 58.2% | 52.2% | +6.0%p | 55.2% |
| Padres | 50.7% | 55.2% | -4.5%p | 65.7% |
| Rays | 54.4% | 58.8% | -4.4%p | 51.5% |
| Phillies | 61.5% | 58.5% | +3.1%p | 56.9% |
| Nationals | 46.3% | 43.3% | +3.0%p | 55.2% |
| Diamondbacks | 50.7% | 53.7% | -3.0%p | 52.2% |
| Royals | 44.1% | 47.1% | -2.9%p | 64.7% |
| Marlins | 48.4% | 50.0% | -1.6%p | 64.1% |
| Rockies | 37.9% | 39.4% | -1.5%p | 62.1% |
| Mets | 43.3% | 43.3% | +0.0%p | 49.3% |
| Brewers | 60.3% | 60.3% | +0.0%p | 58.8% |
| Dodgers | 60.3% | 60.3% | +0.0%p | 55.9% |
| Tigers | 50.0% | 50.0% | +0.0%p | 45.5% |
Bias gap = predicted win rate − actual win rate. + means the model over-predicts wins, − means it under-predicts. Teams with n≥5 only. Actual win rate is based on completed games this season.
55% |
| 3 | Lineup wOBAlineup_woba | 966 (480/486) | 54% |
| 4 | Starter xFIPsp_xfip | 988 (510/478) | 54% |
| 5 | Recent Formrecent_form | 90 (38/52) | 53% |
| 6 | Starter FIPsp_fip | 989 (507/482) | 53% |
| 7 | Bullpen FIPbullpen_fip | 988 (506/482) | 53% |
| 8 | Lineup xwOBAlineup_xwoba | 935 (477/458) | 52% |
| 9 | Elo Ratingelo | 112 (56/56) | 52% |
| 10 | Barrel %lineup_barrel_pct | 971 (498/473) | 49% |
∣ Baseline = overall accuracy 54% (n=993) ∣ home/away = games where this factor favored the home/away team