2026 Season · Utah Utes Football
Honest Scorecard
Each forecast is a model trained only on 2020–2025 play-by-play, then graded on the real 2026 snaps. 2 of 8 graded forecasts were locked (frozen) before kickoff; the rest were locked after the game from the same frozen training data — leak-free but not pre-registered, and labelled RETROSPECTIVE. 2 more are frozen and awaiting kickoff. Nothing is re-fit after the fact, and bad weeks stay on the page.
The three models
- B1 — league situation table: down × distance × score × field position. No coach in it.
- B5 — B1 plus a shrunk coach adjustment (the coach’s habits, pulled toward the league average).
- B3 — the career lookup this site has shown: what that coach did in the same situation over 2020–2025.
Score = Brier (lower is better). Δ vs B1 below zero means the model beat the league table. Panels use competitive snaps only. Opponent and Utah offenses are always scored separately.
Season to date
Opponent offense
PROSPECTIVEB5 − B1
−0.003
beat B1 in 1 of 1
B3 − B1
+0.012
beat B1 in 0 of 1
1 game · 57 competitive snaps · snap-weighted
Opponent offense
RETROSPECTIVEB5 − B1
+0.002
beat B1 in 1 of 3
B3 − B1
+0.024
beat B1 in 0 of 3
3 games · 165 competitive snaps · snap-weighted
Utah offense
PROSPECTIVEB5 − B1
−0.006
beat B1 in 1 of 1
B3 − B1
−0.009
beat B1 in 1 of 1
1 game · 45 competitive snaps · snap-weighted
Utah offense
RETROSPECTIVEB5 − B1
−0.003
beat B1 in 3 of 3
B3 − B1
−0.006
beat B1 in 2 of 3
3 games · 170 competitive snaps · snap-weighted
Noise note: rule of thumb from the preseason study's walk-forward splits, not an interval computed for these games — a single-game contrast carries roughly ±0.02–0.03 Brier of noise, and we will not call a verdict before about 6–8 games per side. Prospective = frozen before kickoff. Retrospective = frozen after kickoff (leak-free, but not pre-registered), so the two are never added together.
Game by game
| Wk | Opponent | Offense | Coach | Snaps | B1 | B5 | B3 | B5−B1 | B3−B1 | Status | Corpus |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Idaho | Opponent offense | Ian Shoemaker | 46 | 0.207 | 0.215 | 0.220 | +0.008 | +0.013 | RETROSPECTIVE | v1 |
| 1 | Idaho | Utah offense | Kevin Mcgiven | 58 | 0.246 | 0.246 | 0.262 | −0.000 | +0.016 | RETROSPECTIVE | v1 |
| 2 | Arkansas | Opponent offense | Tim Cramsey | 57 | 0.227 | 0.224 | 0.239 | −0.003 | +0.012 | PROSPECTIVE | v1 |
| 2 | Arkansas | Utah offense | Kevin Mcgiven | 45 | 0.271 | 0.265 | 0.262 | −0.006 | −0.009 | PROSPECTIVE | v1 |
| 3 | Utah State | Opponent offense | Robert Anae | 46 | 0.254 | 0.252 | 0.254 | −0.002 | +0.000 | RETROSPECTIVE | v1 |
| 3 | Utah State | Utah offense | Kevin Mcgiven | 54 | 0.262 | 0.256 | 0.229 | −0.006 | −0.032 | RETROSPECTIVE | v1 |
| 4 | Iowa State | Opponent offense | Tyler Roehl | 73 | 0.238 | 0.239 | 0.284 | +0.001 | +0.046 | RETROSPECTIVE | v1 |
| 4 | Iowa State | Utah offense | Kevin Mcgiven | 58 | 0.236 | 0.234 | 0.234 | −0.003 | −0.003 | RETROSPECTIVE | v1 |
Frozen, awaiting kickoff
Wk 6 · Kansas · Opponent offense
Coach: Andy Kotelnicki · corpus v1
Frozen Sep 30, 3:57 PM MT · kickoff Oct 10, 8:15 PM MT
Wk 6 · Kansas · Utah offense
Coach: Kevin Mcgiven · corpus v1
Frozen Sep 30, 2:37 PM MT · kickoff Oct 10, 8:15 PM MT
Superseded forecasts (kept on the record)
Wk 6 · Kansas · Opponent offense
First frozen Sep 30, 2:37 PM MT naming Jim Zebrowski; replaced by a forecast naming Andy Kotelnicki. Excluded from season totals.
Why: wrong play-caller: Andy Kotelnicki returned as Kansas OC/play-caller for 2026 (kuathletics.com staff directory); Jim Zebrowski is passing-game coordinator. Caught by the 2026-09-30 data scan, before kickoff.
What this teaches about football
- 1
Down, distance, score and field position predict run/pass better than knowing the coach.
The league situation table (B1) has been hard to beat by adding the coach.
Evidence: preseason study + 2026 ledger
- 2
Coaches have real habits, but small ones.
About a 6-point lean off the league average in run rate, and stable enough to measure (split-half reliability r≈0.6–0.7).
Evidence: preseason study
- 3
The scoreboard changes the offense.
Utah’s opponents trailed by 8+ points on 70–94% of their snaps, so the game situation is doing much of the talking (Utah’s first four 2026 opponents, CFBD play-by-play).
Evidence: 2026 games
- 4
Offensive coordinators are more alike than their reputations.
The middle half of FBS OCs sit within about 0.065 EPA/play of each other.
Evidence: career EPA spread, 2020–2025
- 5
Habits don’t clearly follow a coach to a new job.
0 of 13 job changes beat the league table in the preseason transfer test, and 8+ of Utah’s 12 play-callers in 2026 are in new jobs.
Evidence: preseason transfer test + 2026 schedule
Sources: Preseason validation study, 2026-09-05 (u-of-u data-pipeline docs/validation-phase1-4); 2026 ledger results on this page.