Week 3 in review

Flattered and undersold

Flattered · the score ran ahead of the play by more than 7 points

• Texas vs UTSA, 30–6 W+24 vs +3 implied · +21
  The scoreboard flattered Texas: the +24 margin ran 21 points ahead of the +3 implied by per-play quality (PPA (expected points added per play) -0.06 vs -0.17, success rate 45% vs 27%) (UTSA turned it over once).
• Texas A&M vs Kentucky, 21–31 L-10 vs -29 implied · +19
  The scoreboard flattered Texas A&M: the -10 margin ran 19 points ahead of the -29 implied by per-play quality (PPA (expected points added per play) -0.05 vs +0.45, success rate 42% vs 53%).
• Ohio State vs Kent State, 59–3 W+56 vs +40 implied · +16
  The scoreboard flattered Ohio State: the +56 margin ran 16 points ahead of the +40 implied by per-play quality (PPA (expected points added per play) +0.38 vs -0.32, success rate 57% vs 19%) (Kent State turned it over 3 times).
• Tennessee vs Kennesaw State, 42–9 W+33 vs +22 implied · +11
  The scoreboard flattered Tennessee: the +33 margin ran 11 points ahead of the +22 implied by per-play quality (PPA (expected points added per play) +0.38 vs +0.06, success rate 54% vs 36%) (Kennesaw State turned it over once).
• Georgia at Arkansas, 45–17 W+28 vs +18 implied · +10
  The scoreboard flattered Georgia: the +28 margin ran 10 points ahead of the +18 implied by per-play quality (PPA (expected points added per play) +0.33 vs +0.06, success rate 52% vs 43%) (Arkansas turned it over once).

Undersold · the score ran behind the play by more than 7 points

• Washington vs Eastern Washington, 48–16 W+32 vs +49 implied · -17
  The scoreboard undersold Washington: the +32 margin ran 17 points behind the +49 implied by per-play quality (PPA (expected points added per play) +0.48 vs -0.26, success rate 52% vs 34%) (Washington turned it over once), against an FCS opponent.
• Louisville vs SMU, 41–31 W+10 vs +25 implied · -15
  The scoreboard undersold Louisville: the +10 margin ran 15 points behind the +25 implied by per-play quality (PPA (expected points added per play) +0.41 vs -0.00, success rate 49% vs 43%).
• Iowa vs Northern Iowa, 55–0 W+55 vs +66 implied · -11
  The scoreboard undersold Iowa: the +55 margin ran 11 points behind the +66 implied by per-play quality (PPA (expected points added per play) +0.67 vs -0.37, success rate 62% vs 26%), against an FCS opponent.
• Houston at Texas Tech, 26–28 L-2 vs +6 implied · -8
  The scoreboard undersold Houston: the -2 margin ran 8 points behind the +6 implied by per-play quality (PPA (expected points added per play) +0.28 vs +0.26, success rate 51% vs 48%) (Houston turned it over once).
• Florida at Auburn, 44–39 W+5 vs +13 implied · -8
  The scoreboard undersold Florida: the +5 margin ran 8 points behind the +13 implied by per-play quality (PPA (expected points added per play) +0.31 vs +0.17, success rate 39% vs 43%) (Florida turned it over once).

Closest projection

Over the 21 games this week where SP+, Elo and the line all existed before kickoff, the line as a benchmark (its miss only, never closest) (6 more shown on the Projections page but not in the record: no page before kickoff).

Line only (no SP+ or Elo rating for an FCS opponent): 3 games, avg miss 17.2; not in the table.

Upsets

Ranked teams that lost to an unranked or lower-ranked opponent, by the AP Top 25 in effect for week 3, with each model's pregame margin from the ranked side and the line beside them for comparison only; ✓ when a model had the winner, ✗ when it did not.

• #7 LSU lost at #8 Ole MissOle Miss 32, LSU 24
  SP+ ✗

  +6.4

  Elo ✓

  -4.8

  Line

  +2.5

  Our model

  –
• #9 Texas A&M lost vs KentuckyKentucky 31, Texas A&M 21
  SP+ ✗

  +20.5

  Elo ✗

  +20.5

  Line

  +15.5

  Our model

  –
• #16 SMU lost at #23 LouisvilleLouisville 41, SMU 31
  SP+ ✓

  -1.3

  Elo ✗

  +0.1

  Line

  -1.5

  Our model

  –
• #25 Virginia lost vs West VirginiaWest Virginia 38, Virginia 27
  SP+ ✗

  +9.7

  Elo ✗

  +11.5

  Line

  +9.5

  Our model

  –

Called it: SP+ 1 of 4 · Elo 1 of 4 · Our model 0 of 0.

Biggest swings

The plays that moved win probability most, across the 29 games with win-probability data, from the home side.

• Houston at Texas TechTexas Tech 91% → 38%
  Q4 0:50, 1st & 10: (00:50) No Huddle-Shotgun #1 C.Weigman pass complete short left to #9 P.Overmyer caught at TTU00, for 10 yards to the TTU00 TOUCHDOWN, clock 00:49, 1ST DOWN rush attempt failed rush attempt failed. The previous play is under automatic review - "" #1 C.Weigman rush attempt failed. The previous play is under automatic review - "" #1 C.Weigman rush attempt failed. The previous play is under automatic review - "". CALL UPHELD #1 C.Weigman rush attempt failed. The previous play is under automatic review - "Short of the goal line". CALL UPHELD · final Texas Tech 28, Houston 26
• Houston at Texas TechTexas Tech 38% → 90%
  Q4 0:41, 1st & 10: (00:41) Shotgun #8 C.Dickey rush middle for 5 yards gain to the HOU42 (#90 D.Thomas) · final Texas Tech 28, Houston 26
• UTEP at MichiganMichigan 61% → 19%
  Q2 5:58, 1st & 10: (05:58) Shotgun #1 E.Colson Jr. pass complete short left to #32 L.Sperling caught at U-M15, for 18 yards to the U-M00 TOUCHDOWN, clock 05:58, 1ST DOWN #91 C.Arreola kick attempt good (H: #47 B.Braun, LS: #46 C.Loeb) · final Michigan 52, UTEP 17

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.

• Utah · flattered +18 vs Arkansasvs Utah State 33–0 W · -9.5 vs form (-2.2 vs the field)
• Texas Tech · flattered +16 vs Oregon Statevs Houston 28–26 W · -17.5 vs form (-10.2 vs the field)
• Penn State · flattered +15 vs Templevs Buffalo 55–13 W · +8.0 vs form (+15.3 vs the field)
• Houston · flattered +15 vs Southernat Texas Tech 26–28 L · -39.0 vs form (-31.7 vs the field)
• Notre Dame · flattered +13 vs Ricevs Michigan State 27–10 W · -24.2 vs form (-16.9 vs the field)
• Miami · flattered +13 vs Florida A&Mat Wake Forest 33–20 W · -39.0 vs form (-31.7 vs the field)
• Texas A&M · flattered +13 vs Arizona Statevs Kentucky 21–31 L · -49.0 vs form (-41.7 vs the field)
• Missouri · flattered +12 vs Kansasvs Troy 27–17 W · -21.0 vs form (-13.7 vs the field)
• Ole Miss · flattered +11 vs Charlottevs LSU 32–24 W · -10.8 vs form (-3.4 vs the field)
• Virginia · flattered +11 vs Norfolk Statevs West Virginia 27–38 L · -49.5 vs form (-42.2 vs the field)
• LSU · flattered +10 vs Louisiana Techat Ole Miss 24–32 L · -39.0 vs form (-31.7 vs the field)
• Oregon · undersold -14 vs Oklahoma Statevs Portland State 84–0 W · +82.0 vs form (+89.3 vs the field)

Baseline: all 36 page teams that played ran -7.3 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: -26.4 vs form over 11, -19.1 against the baseline (-3.0 in the study) · Undersold: +82.0 vs form over 1, +89.3 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.