Washington beat Boise State by 28 while gaining 0.7 more yards per drive

Washington led Boise State 10-3 with two minutes left in the first half. Thirty-nine seconds later it led 24-3. Demond Williams Jr. threw six yards to Dezmen Roebuck at Q2 1:56. At 1:39, Rahshawn Clark intercepted Maddux Madsen at the Boise State 39 — Isaiah Ward is credited with hurrying the throw — and returned it four yards to the 35. At 1:17, Williams found Raiden Vines-Bright from three yards out. Two touchdowns and an interception inside a single 39-second window, and Boise State never again played a snap with the deficit under 21 points.

That window is the mechanism, not the whole margin. It supplied 14 of the 28 points. Colton Boomer's 52-yard field goal at Q1 11:35 had opened the scoring, Grady Gross answered from 33 at 4:20, and Denzel Boston's 78-yard catch from Williams at Q2 9:54 made it 10-3. Everything after Vines-Bright arrived with the result settled: Quentin Moore's 32-yard touchdown catch at Q3 12:05 for 31-3, Jonah Coleman's six-yard run at Q4 10:46 for 38-3, and Matt Lauter's eight-yard catch from Max Cutforth at Q4 1:18 for the final.

The 28 points of separation are not visible in the per-drive production. Washington gained 25.0 yards per drive across 14 possessions; Boise State gained 24.3 across 13. Seven-tenths of a yard. The separation is in what those drives were worth and where they began: Washington scored on 42.9 percent of its possessions to Boise State's 7.7, and started its average drive 59.9 yards from the end zone against Boise State's 71.0. Per play, Washington was worth 0.128 predicted points to Boise State's -0.163, and succeeded on 36.9 percent of its snaps to 31.0 percent — the gap of a comfortable afternoon, not a 28-point one.

That 11.1-yard field-position edge was bought by the Washington secondary. Boise State threw five interceptions off two quarterbacks, and three of them handed Washington the ball inside Boise territory. Xe'ree Alexander took Madsen at the Washington 37 at Q1 1:09 and returned it 29 yards to the Boise State 34. Clark's pick at the Boise State 39 set up the Vines-Bright touchdown 22 seconds later. Deshawn Lynch intercepted Cutforth at the Washington 34 at Q4 11:35 and returned it 57 yards to the Boise State 9. Leroy Bryant's two came at the Washington 40 and in the end zone for a touchback. Four of the five throws credit a separate hurry, each to a different Washington defender — Clark, Anterio Thompson, Isaiah Ward and Ta'ita'i Uiagalelei. Washington gave the ball away once, on Coleman's fumble after a 10-yard catch.

Williams turned 215 passing yards into four touchdowns thrown to four different receivers, which is the finishing edge stated in personnel terms. Boston's 78-yard catch was 36 percent of Williams' total yardage by itself and left him at 126 receiving yards, 57 percent above his 80.1-yard average across 11 games. Roebuck, Vines-Bright and Moore took the other three scores from 6, 3 and 32 yards. Williams' own yardage was 9 percent below his 235.8-yard average over 13 games — an ordinary passing night that produced four touchdowns because three interceptions kept handing him the ball inside Boise State territory.

Boise State's passing and receiving lines are not the lines of a team beaten by 28. Cutforth threw for 202 yards, 72 percent above his 117.7-yard average over nine games. Chris Marshall caught 97, 69 percent above his 57.4 across ten. The failure was on the ground, and the run-blocking numbers read backwards: Washington's carries were stopped at or behind the line 20 percent of the time, five times Boise State's 4 percent, and Washington still produced 3.43 line yards a carry to Boise State's 2.59. Dylan Riley, whose season average is 80.4 yards over 14 games, finished with 34. Coleman went for 85 against a 63.2-yard average over 12.

Boise State's offense graded at -0.163 predicted points added per play, the lowest of the 14 games in its season and the 0th percentile of its own distribution — its worst day, not a bad stretch. The drives that died explain where the yardage went. A 12-play, 50-yard possession ended on downs 17 yards from the end zone when Cutforth's throw to Marshall fell incomplete at Q3 0:59. A six-play, 47-yard possession reached the Washington 6 before Bryant intercepted in the end zone at Q4 12:38, with Uiagalelei hurrying. The only long drive that finished — 13 plays, 94 yards to Lauter's touchdown — came when it was already 38-3.

Boise State's one fourth-quarter break lasted 13 seconds. Fifty seconds after Bryant's end-zone interception, at Q4 11:48, Coleman caught a Williams pass at the Washington 30 for 10 yards and fumbled it away. At 11:35, Cutforth was intercepted at the Washington 34 by Lynch, who returned it 57 yards to the Boise State 9. Coleman ran it in from six at 10:46. Boise State's recovery at the Washington 30 became a Washington touchdown 62 seconds later.

What the model had, and what it missed

Our drive-sim made Boise State the favorite by 8.2 points with a 69.7 percent win probability at a neutral SoFi Stadium. The market had Washington at roughly -10 — an 18.2-point disagreement before kickoff. The game landed on neither side of it: Washington won by 28, clearing the market's number by 18.2 points and missing the projection by 36.2, with the model's pick beaten outright.

The dominant pregame driver was last season's body of work, worth 4.2 points to Boise State — the largest of five factors, ahead of quarterback (+2.3), returning production (+1.9) and transfer portal (+1.1). Roster talent was the only factor that leaned Washington, and only by 1.1. The trouble is that last season's body of work pushes the other way on nearly every line of the two résumés.

Washington walked in ranked 15th at 8-4, scoring 33.8 points a game and allowing 19.4. Boise State walked in ranked 40th at 9-4, scoring 31.4 and allowing 23.0. Boise State's advantages were a better record and a three-game winning streak against Washington's one-game skid. The engine ran on a preseason prior with no 2025 games behind it, and still published a 69.7 percent number. The result fell outside its own stated uncertainty: the 80 percent margin band ran from -28 to +11 from Washington's side, and Washington won by 28, clearing the top of the band by 17 points.

The total was the one correct call, and it was correct for a reason the model did not hold. The projection of 49.6 landed 1.6 points from the actual 48 against a market total of 55.5. But it got under the number while grading both offenses as strengths — Boise State's offensive edge at 0.81, Washington's at 0.19 — and neither edge showed up: Washington's offense produced 0.128 predicted points per play, Boise State's -0.163. The under cashed because both offenses were throttled, not because either was expected to be.

Washington posted 0.128 predicted points added per play against a 0.292 season mark; Boise State posted -0.163 against 0.192. That is 56 percent and 185 percent below each offense's own standard. Both teams were worse than themselves. Only one of them was intercepted five times, and that is the entire distance between 24.3 yards per drive and 10 points.

For a bettor, this game paid on conversion rate rather than yardage — Washington -10 covered by 18.2 without the offense out-producing Boise State's by a full yard per possession, which is not a repeatable profile. For the model, it is a cleaner lesson: the factor carrying the most weight, last season's body of work, pointed at the team ranked 25 spots lower nationally, scoring 2.4 fewer points a game and allowing 3.6 more.