Massachusetts Over Rutgers: The Opening-Week Earthquake
Massachusetts entered Week 1 with a 1% pregame win probability against Rutgers. The Minutemen left the field with a 37-21 victory, the most improbable result of an opening week that produced three outright upses. In raw terms, this was the largest gap between expectation and outcome: a team given virtually no path to victory simply took one, methodically, across sixty minutes. What happened in that game remains visible in the film, even if the pregame models were wrong.
The broader pattern is worth examining. Three games in which the underdog carried a 40% or lower pregame probability of winning all broke toward the underdog. Massachusetts at 1%. Oklahoma State at Tulsa at 16%. Colorado at Georgia Tech at 29%. That's not variance; that's a systematic mismatch between what the models expected and what the field produced.
The Thinnest Margins
Four of the five closest finishes came down to either one or three points. Louisville at Ole Miss went 38-41, a three-point decision in the most electrically entertaining game of the day—the excitement index of 8.34 suggests a back-and-forth throughout. Auburn over Baylor by one. Western Michigan at Michigan by one, though that one carried a pregame win probability of 94% for Michigan, meaning the favorite still won; just barely. Troy over Sam Houston, also by three.
One-score football is indifferent to seed or ranking. It's a coin-flip environment in which execution, turnover luck, and the referee's whistle matter more than anything the pregame models can divine. But the sheer frequency of it—five games decided by a possession or fewer out of fifty played—suggests the opening week produced tighter competition than expected. Whether that's because teams are closer in talent this year or because Week 1 simply tends to be a high-variance event remains unanswered.
The Excitement and the Question
Louisville-Ole Miss registered the highest excitement index (8.34), followed by Boise State-Oregon (7.78), a 27-34 game that stayed close through the second half. The model's excitement metric captures volatility and momentum swings—not just closeness, but the rhythm of how a game unfolded. These two games delivered the most dramatic sequences.
What's unsettled is whether this Week 1 chaos predicts anything about the season ahead. Do three upset-adjacent results suggest the talent distribution is genuinely more level than preseason expectations implied, or are we simply seeing the natural noise of a single weekend in which fifty games play out simultaneously. The models will absorb the results, adjust for information, and Week 2 will test whether these trends hold or dissolve into the usual hierarchy.
For now: Massachusetts, Oklahoma State, and Colorado won games they had almost no business winning. Auburn and Troy survived one-possession tests. The favorites who held on did so by the slimmest of margins.