Receipts
Calibration
When the model predicts an X% HR probability, do those batters actually hit home runs X% of the time? This page is updated nightly from boxscore outcomes. No cherry-picking, no hindsight edits — every forecast we publish ends up here.
Forecasts scored
814
Actual HRs
83
Overall predicted
12.15%
Overall actual
10.20%
By Probability Bucket
Predicted
N
Predicted avg
Actual HR rate
Grade
< 3%
0
—
—
—
3-5%
0
—
—
—
5-7%
0
—
—
—
7-10%
98
9.4%
3.1%
over-predicting
10-13%
486
11.5%
10.5%
calibrated
13-16%
189
14.1%
12.2%
close
16%+
41
17.4%
14.6%
close
Buckets with fewer than 30 forecasts are too small to judge. A calibrated bucket means the predicted average closely matches the observed HR rate.
Top 10/day Performance
Top picks scored
00/0
Top hit rate
0.0%
Slate baseline
10.2%
Lift vs baseline
0.00x
Of the top 10 ranked picks per game day, how often does the batter hit a home run? Lift >1.0 means the top slots are genuinely better than randomly picking any forecast.
Brier score: 0.0916(lower is better; 0 is perfect)Last outcome: 2026-04-20Models: hr-v1.5.3hr-v1.6.0hr-v1.7.0