‹ Week 10ProjectionsMetrics explained
Flattered and undersold
Flattered · the score ran ahead of the play by more than 7 points
- BYU vs Baylor, 28–7 W+21 vs +6 implied · +15The scoreboard flattered BYU: the +21 margin ran 15 points ahead of the +6 implied by per-play quality (PPA (expected points added per play) +0.15 vs +0.07, success rate 45% vs 37%) (Baylor turned it over once).
- Wisconsin at Maryland, 21–24 L-3 vs -17 implied · +14The scoreboard flattered Wisconsin: the -3 margin ran 14 points ahead of the -17 implied by per-play quality (PPA (expected points added per play) -0.25 vs +0.06, success rate 32% vs 38%) (Maryland turned it over once).
- Boise State vs Oregon State, 46–41 W+5 vs -4 implied · +9The scoreboard flattered Boise State: the +5 margin ran 9 points ahead of the -4 implied by per-play quality (PPA (expected points added per play) +0.40 vs +0.45, success rate 47% vs 59%) (Oregon State turned it over once).
- Virginia vs California, 36–7 W+29 vs +21 implied · +8The scoreboard flattered Virginia: the +29 margin ran 8 points ahead of the +21 implied by per-play quality (PPA (expected points added per play) +0.22 vs -0.10, success rate 44% vs 35%) (California turned it over 3 times).
Undersold · the score ran behind the play by more than 7 points
- LSU vs Texas, 21–22 L-1 vs +23 implied · -24The scoreboard undersold LSU: the -1 margin ran 24 points behind the +23 implied by per-play quality (PPA (expected points added per play) +0.24 vs -0.10, success rate 35% vs 36%) (LSU turned it over 3 times).
- Notre Dame vs Boston College, 41–7 W+34 vs +43 implied · -9The scoreboard undersold Notre Dame: the +34 margin ran 9 points behind the +43 implied by per-play quality (PPA (expected points added per play) +0.30 vs -0.37, success rate 52% vs 24%) (Notre Dame turned it over once).
- Tennessee at Texas A&M, 22–7 W+15 vs +24 implied · -9The scoreboard undersold Tennessee: the +15 margin ran 9 points behind the +24 implied by per-play quality (PPA (expected points added per play) +0.08 vs -0.40, success rate 37% vs 27%).
- Oklahoma vs Ole Miss, 24–17 W+7 vs +15 implied · -8The scoreboard undersold Oklahoma: the +7 margin ran 8 points behind the +15 implied by per-play quality (PPA (expected points added per play) +0.16 vs -0.07, success rate 43% vs 45%).
- Miami vs Duke, 36–25 W+11 vs +19 implied · -8The scoreboard undersold Miami: the +11 margin ran 8 points behind the +19 implied by per-play quality (PPA (expected points added per play) +0.41 vs +0.19, success rate 48% vs 44%).
Closest projection
Over the 28 games this week where SP+, Elo and the market all existed before kickoff (2 more shown on the Projections page but not in the record: no page before kickoff).
| model | avg miss | closest |
|---|---|---|
| SP+ | 13.0 | 5 of 28 |
| Elo | 11.3 | 10 of 28 |
| Market | 11.8 | 13 of 28 |
Our model experimental · 28 eligible games · vs SP+: 11.5 against 13.0, closest 17 / 11 · vs Elo: 11.5 against 11.3, closest 12 / 16 · vs Market: 11.5 against 11.8, closest 16 / 12 · all four: our model closest 10 of 28
Our call · 28 games where SP+, Elo and Our model all existed before kickoff, and each model on those same games, the market among them:
| model | games | avg miss | named the winner |
|---|---|---|---|
| Our call | 28 | 11.8 | 19 of 28 |
| SP+ | 28 | 13.0 | 19 of 28 |
| Elo | 28 | 11.3 | 20 of 28 |
| Market | 28 | 11.8 | 20 of 28 |
| Our model | 28 | 11.5 | 22 of 28 |
Our call against the market, head to head: Our call was at least as close in 15 of 28, the market in 13 (a tie counts for both).
Who's been right: every model
The scorecard: every model by week, with Our call's biggest misses and best calls ›
Each model is graded on every game it predicted before kickoff, from the first game of the season. The market is graded like every model.
| model | named the winner | avg miss | from |
|---|---|---|---|
| Market | 217 of 276 · 79% | 12.0 | Wk 1 |
| Elo | 190 of 253 · 75% | 12.6 | Wk 1 |
| Our model | 137 of 185 · 74% | 12.6 | Wk 4 |
| SP+ | 160 of 220 · 73% | 11.8 | Wk 2 |
| The Margin | 71 of 99 · 72% | 12.6 | Wk 8 |
| ESPN FPI | 70 of 99 · 71% | 12.5 | Wk 8 |
| Sagarin | 70 of 99 · 71% | 12.3 | Wk 8 |
| Our call | 127 of 181 · 70% | 12.2 | Wk 4 |
| Record | 161 of 232 · 69% | – | Wk 2 |
| SRS | 69 of 100 · 69% | 12.9 | Wk 8 |
| Sonny Moore | 65 of 99 · 66% | 12.2 | Wk 8 |
Market for Weeks 1, 2, 4, 6 (58 games) and Elo for Week 1 (33 games) come from their published pre-game numbers (CollegeFootballData), read after the fact: reconstructed, never a number published after kickoff.
Record (ours) names a winner only: the better record before the week, the home team on equal records. ESPN FPI and SRS come from CollegeFootballData's ratings; Sagarin (his PREDICTOR rating), Sonny Moore and The Margin from their own pages, credited by name on our models page. A game counts only when the model's number was saved before its kickoff.
Season to date through Week 11. The table above is the head to head: the models on the same games this week.
Upsets
Ranked teams that lost to an unranked or lower-ranked opponent, by the AP Top 25 in effect for week 11, with each model's pregame margin from the ranked side, the market among them; ✓ when a model had the winner, ✗ when it did not.
-
SP+ ✗+4.9Elo ✗+2.0Market ✗+2.5Our model ✓-2.8
- #14 Houston lost at ColoradoColorado 42, Houston 32SP+ ✗+16.8Elo ✗+11.5Market ✗+10.5Our model ✗+7.2
-
SP+ ✗+3.2Elo ✗+4.3Market ✗+2.5Our model ✓-2.1
- #20 South Carolina lost at ArkansasArkansas 34, South Carolina 6SP+ ✗+14.8Elo ✗+8.6Market ✗+8.5Our model ✗+7.5
- #23 New Mexico lost vs UNLVUNLV 21, New Mexico 17SP+ ✗+20.6Elo ✗+12.7Market ✗+13.5Our model ✗+10.6
Called it: SP+ 0 of 5 · Elo 0 of 5 · Market 0 of 5 · Our model 2 of 5.
Biggest swings
The plays that moved win probability most, across the 28 games with win-probability data, from the home side.
-
Q4 0:04, 2nd & 4: (00:04) Shotgun #16 A.Manning pass complete short left to #0 R.Brown caught at LSU03, for 4 yards to the LSU00 TOUCHDOWN, clock 00:03 rush attempt failed · final LSU 21, Texas 22
- Wisconsin at MarylandMaryland 16% → 70%Q4 2:18, 1st & 10: (02:18) No Huddle-Shotgun #6 C.Williams pass complete short right to #8 P.Howard caught at WIS25, for 39 yards to the WIS00 TOUCHDOWN, clock 02:11, 1ST DOWN #36 S.O'Haire kick attempt good (H: #34 B.McFerson, LS: #97 L.O'Hare) · final Maryland 24, Wisconsin 21
-
Q4 2:46, 2nd & 5: (02:46) No Huddle-Shotgun #4 D.Dampier pass complete deep middle to #2 C.Whittemore caught at ARIZ45, for 70 yards to the ARIZ00 TOUCHDOWN, clock 02:35, 1ST DOWN #17 D.Curtis kick attempt good (H: #92 O.Phillips, LS: #69 G.Victor) · final Arizona 39, Utah 40
Regression watch, checked
Last week's teams whose score ran more than 7 points from the play, and what they did this week against their season-to-date form (average margin before the game, home field removed). The 2022–25 study found flattered teams fell -3.0 short of form the next game and undersold teams beat it by +1.7.
- UCLA · flattered +11 vs Minnesotavs Illinois 22–14 W · -13.7 vs form (-5.4 vs the field)
- SMU · flattered +11 vs Virginia Techvs Wake Forest 7–49 L · -57.8 vs form (-49.6 vs the field)
- Texas · flattered +9 vs Missouriat LSU 22–21 W · -9.1 vs form (-0.8 vs the field)
- Miami · flattered +9 vs Notre Damevs Duke 36–25 W · -17.6 vs form (-9.3 vs the field)
- Wisconsin · flattered +9 vs Rutgersat Maryland 21–24 L · -8.1 vs form (+0.2 vs the field)
- South Carolina · flattered +8 vs Texas A&Mat Arkansas 6–34 L · -35.9 vs form (-27.7 vs the field)
- Missouri · undersold -9 vs Texasat Georgia 21–48 L · -34.3 vs form (-26.1 vs the field)
- Notre Dame · undersold -9 vs Miamivs Boston College 41–7 W · +7.0 vs form (+15.2 vs the field)
- Texas A&M · undersold -8 vs South Carolinavs Tennessee 7–22 L · -24.7 vs form (-16.4 vs the field)
- Texas Tech · undersold -8 vs West Virginiaat Oklahoma State 52–27 W · +14.9 vs form (+23.1 vs the field)
- Louisville · undersold -7 vs Georgia Techat North Carolina 16–14 W · -9.8 vs form (-1.6 vs the field)
Baseline: all 45 page teams that played ran -8.2 vs form this week. The baseline matters because early-season form comes mostly from easier non-conference games, so the whole field tends to fall short of it once conference play starts; a flagged group is read against the field, not against zero.
Flattered: -23.7 vs form over 6, -15.4 against the baseline (-3.0 in the study) · Undersold: -9.4 vs form over 5, -1.2 against the baseline (+1.7 in the study). One week is a few teams and proves nothing either way; the point is that the check is visible every week.
How these are computed
Verdicts are the team pages' own: the score against the margin the plays earned, recomputed here from the box score and drives, so any past week rebuilds the same way. The closest-projection table is the accuracy report's block for the week; a game counts only if one of its teams had a page before kickoff, and our model is kept on its own line. SP+ is graded only in weeks where its own rating was posted before kickoff; an earlier snapshot shown before a game is never scored. Upsets use the AP Top 25 in effect for the week's games; every ranked team has a page, so no upset is missed. Pregame margins are the same ones the season record scores. Swings are the win-probability moves in the play-by-play. The regression check uses the study's own definition of form. Nothing on this page is written by AI, and a section with nothing to show is left out.