A useful rating system should be surprised sometimes. When it is wrong in the right places, it tells us where the season is teaching us something new. The table below lists games where the pre-snapshot model direction and the actual margin disagreed most sharply among rated completed games.
| Date | Result | Model home edge | Actual home margin |
|---|---|---|---|
| 2026-09-08 | John Carroll Catholic High School 20, Springville High School 6 | -7.2 | +14 |
| 2026-08-25 | B.C. Rain High School 0, McGill-Toolen Catholic High School 13 | +4.1 | -13 |
| 2026-08-25 | Satsuma High School 19, St. Michael Catholic High School 12 | -5.9 | +7 |
| 2026-08-25 | Spanish Fort High School 13, Fairhope High School 19 | +5.2 | -6 |
| 2026-09-08 | Northside Methodist Academy 18, Charles Henderson High School 12 | -4.9 | +6 |
| 2026-08-25 | J.B. Pennington High School 6, Springville High School 0 | -3.5 | +6 |
| 2026-08-22 | Auburn High School 6, Albertville High School 12 | +3.2 | -6 |
| 2026-09-01 | Brewbaker Tech Magnet High School 7, Vestavia Hills High School 13 | +3.1 | -6 |
Why surprises matter
One upset does not erase all prior information, especially early in the season. But repeated model misses in the same direction are exactly how a team climbs or falls. Coaches should look for whether the result reflects a repeatable matchup problem—speed, rush timing, red-zone efficiency—or a one-night event such as turnovers and missing players.
What the numbers mean
The model home edge above is the home team's rating plus the current home-field estimate (2.12 points) minus the opponent rating. The actual home margin is the scoreboard difference. Opposite signs create the surprise list.
Data and methods
This article was generated from data/games_2026.csv and public/data/rankings.json. It uses only completed, rated games in the repository snapshot and does not infer injuries, roster availability, or coaching decisions that are not present in the data.