The College Football Playoff Selection Committee released its final bracket on a Sunday afternoon in early December, and within forty minutes the Miami athletic department had issued a statement, the Notre Dame fan base had started a petition, and the BYU coaching staff had — in private, by the second-hand accounts that filtered through Salt Lake City reporters — described the experience as “watching your bowl placement get reorganized in real time.” The bracket itself: Indiana at the top, Ohio State at two, Georgia at three, Texas Tech at four. The controversy was several spots lower. Miami had been ranked twelfth in the December 2 release, behind Notre Dame at ten. Five days later, with BYU losing the Big 12 Championship and nothing else changing on the field, Miami was ranked tenth and into the bracket as the final at-large team. Notre Dame was on the outside.
That swap is the story this December produced. It is not unprecedented — the committee has reorganized teams between rankings before, especially after championship weekend rearranged the conferences — but it is the cleanest example we have had of the committee discretion overriding what the analytics community had been arguing about since November. SP+, the most widely-cited public CFB model, had been ahead of the committee on Miami the entire season. The committee got to the same answer in December. The path from one to the other is the part worth reading carefully.
What follows is what SP+ had been saying about Miami all season, where the committee’s bracket disagreed with model rankings beyond Miami, and what the gap between the two systems reveals about how college football evaluates teams now that the playoff is twelve teams and the marginal at-large decisions actually matter.
What SP+ had been saying about Miami since September
SP+, Bill Connelly’s opponent-adjusted efficiency rating, had Miami ranked between 7 and 10 for most of the second half of the season. The committee had Miami between 11 and 13 across the same window. The gap of roughly three places came down to one input the committee weighed and the model did not: Miami’s losses had been close, and the committee did not credit close losses the way close wins.
Miami finished the regular season 10-2. The two losses were to Louisville (24-21) and SMU (28-26). Both were single-score losses to ranked opponents on the road. SP+ treated both as inputs that adjusted Miami’s efficiency rating downward slightly but kept the team in the top ten on the strength of their offensive and defensive efficiency numbers. The committee weighted the losses as resume blemishes, and Miami stayed at 11 or 12 until BYU’s loss in the Big 12 title game reorganized the at-large pool.
The head-to-head argument with Notre Dame was the most visible piece of the disagreement. Miami had beaten Notre Dame 27-24 in Week 1. The committee’s December 2 ranking had Notre Dame at 10 and Miami at 12, with Notre Dame’s better resume against ranked opponents cited as the reason. SP+ had the two teams within 0.2 efficiency points of each other and gave Miami the edge based on head-to-head. The committee took until December 7 to land on the same answer, and only because BYU’s loss forced them to reorganize anyway.
Where SP+ and the committee disagreed beyond Miami
| Team | Committee final rank | SP+ rank | Gap | Likely reason |
|---|---|---|---|---|
| Indiana | 1 | 3 | -2 | Committee credits Big Ten title win heavily |
| Ohio State | 2 | 1 | +1 | Title-game loss penalized despite better metrics |
| Texas Tech | 4 | 7 | -3 | Big 12 champion bonus |
| Miami | 10 | 8 | -2 | Two close losses heavily weighted |
| BYU | 15 | 11 | -4 | Big 12 title loss collapsed ranking |
| Boise State | 13 | 16 | +3 | Group-of-Five bonus for SP+ |
The biggest gaps are at the top and the bubble. Ohio State by SP+ is the best team in the country; the committee places them second because they lost the Big Ten title game. Indiana wins the title game and gets the top seed; SP+ rates them three places lower because the underlying efficiency numbers do not actually catch the title-game victory the way the committee does. Texas Tech as a four-seed is the largest single gap on the bracket — SP+ has them outside the top five entirely.
None of these gaps are scandals. They are the structured difference between an opponent-adjusted efficiency rating and a process that weights conference championships, head-to-head results, and committee discretion. The model and the committee are answering different questions. The committee is answering “who should play for the title given conference structure.” The model is answering “who is the best team by underlying efficiency.” Both answers are defensible. Both produce different brackets.
Where this gets weird
The complicating factors that the SP+-vs-committee framing tends to flatten.
The first is that SP+ has a built-in conference-strength adjustment that the committee does not. SEC and Big Ten teams get a slight boost for the opponents they play. The committee evaluates each game on its own without an explicit conference-strength multiplier, which produces a different aggregate ranking when conference strengths diverge — which they did this year. The committee under-weights the SEC slightly compared to where SP+ places it, and the committee over-weights the Big Ten slightly relative to the same model.
The second is that the head-to-head argument the SP+ side likes to make — Miami beat Notre Dame in Week 1, so Miami should rank above — gets weaker the further the season runs from the head-to-head result. By December, Miami beating Notre Dame in the first week of September was four months old. Notre Dame’s resume across the back half of the season was, by both measures, stronger. The committee had a defensible argument for ranking Notre Dame above Miami. SP+ had a defensible argument for the reverse. Neither side was wrong; they were weighting recency differently.
The third is that the twelve-team format has changed what the marginal at-large decision actually means. Under the four-team format, the disagreements between SP+ and the committee at spots 5-12 were mostly cosmetic — they did not affect who played for the title. Under the twelve-team format, the difference between being the 11th-ranked team and the 13th-ranked team is the difference between playing for the title and watching from a bowl game. The stakes of the disagreement have risen by an order of magnitude.
What this should change about how the analytics community engages
- Stop treating SP+ as a referee on committee decisions. The model is answering a different question than the committee. The disagreement is not “who is right.” It is “which question should the playoff format be answering.”
- Track the gap closing or widening over the season. A team whose SP+ ranking improves more than their committee ranking has discovered something the committee has not yet rewarded. Miami in 2025 was a clean example.
- Watch how head-to-head gets weighted by date. The committee weights early-season head-to-head less than late-season head-to-head. SP+ treats both equivalently. The structure of the disagreement on Miami-Notre Dame was a textbook version of this asymmetry.
- Push for SP+ data to be visible in committee deliberations. The committee has access to advanced metrics. Whether they actually weight them is a separate question. Public pressure to surface the data in the rankings explanations is the cheapest fix available to the format.
The callback
The committee swapping Miami and Notre Dame between December 2 and December 7 was, on its surface, a recognition that the Big 12 title game had reorganized the bubble. Underneath, it was the moment the committee landed on a ranking SP+ had been arguing for since September. The fact that it took until BYU losing for the committee to get there is the part the analytics community should keep pointing at. Not as a complaint about the committee, which has a defensible process. As an observation about how slowly the system updates when the underlying data has already moved. The SEC vs Big Ten conference piece covers the broader version of how conference structure interacts with model output. Miami got the right ranking. They got it five days late. The model and the committee will keep disagreeing for as long as one of them is answering the wrong question on purpose.
SP+ ratings via Bill Connelly at ESPN; CFP committee rankings via CFP official site; bracket reorganization documented via NCAA.com.
