Week 11 College Football Playoff Rankings: Where Our Model Thinks They’ll Land

We’ve had one set of College Football Playoff rankings, and we’ve had one week of games on the back of them. Tonight, we’ll get the committee’s updated top 25. Here’s where our model thinks they’ll lie.

The core piece of our college football model is how it estimates the committee ranks teams. We have a detailed explanation of how the model approaches this estimate, but the two high-level things to know are these:

• It’s 36-for-36 in ultimately predicting the playoff teams.
• It adjusts each week, reacting to the committee’s rankings.

The way our model reacts to the committee’s rankings is to assume they’re hewing as closely to precedent as possible. In effect, it assumes the committee is as consistent as it can possibly be, relative to its previous rankings. This can sometimes lead our model to expect changes in the rankings that humans would not expect. An example: Penn State was ranked lower last week than our model expected. Our model reacted by moving them to only inches behind Mississippi, rather than some uniform distance. The consequence? It now has Penn State rising all the way up to 7^th after their road domination of Maryland. We do not personally expect that.

This is still useful to see. It’s a reminder that Penn State dramatically improved its résumé this weekend, and that if the committee continues to hold them back, it values them even less—relative to precedent—than we previously might have guessed. We continue to learn what the committee thinks of these teams, even if the committee can be known to waffle.

We’ll have more on specific rankings below, and thoughts on what to watch for during this week’s reveal, but we wanted to lead with this anecdote to say one important thing: This isn’t where we think the rankings will land. This is where our model thinks the rankings will land. The model is a good tool, but this is an area where it is decidedly imperfect.

Thoughts, starting at the top:

We’d be surprised if Ohio State and Georgia don’t stay 1^st and 2^nd, in that order, though the narrative about Ohio State’s game is very different from the game’s result, and you could make a case that the committee might really value Georgia’s win over Missouri, far and away its best of the year as measured by their rankings.

We’d be surprised if Florida State jumped Michigan, but we wouldn’t be all that shocked to see Washington jump the Seminoles, if not Michigan as well. Washington went on the road and beat a team the committee had ranked, looking spectacular offensively in the process. As Oregon continues to dominate, Washington’s win over the Ducks looks better and better as well, and the Arizona win’s value is similarly on the rise. It’s easy to make a case for a strong Washington ranking, even if you don’t believe in them as a team.

We don’t expect Penn State to jump all those teams, but it’s interesting that our model is giving them such a boost for beating Maryland and not giving as much of a boost to Alabama or Texas for beating LSU or Kansas State, who are better teams. What happened here is partly that Penn State won by a ton of points and did it on the road, but also: Alabama already had three wins very similar in quality, considering margin and location, to its win over LSU. Alabama beat Tennessee by the exact same margin its last time out. Alabama beat Mississippi by the exact same margin in September. Alabama beat Texas A&M by only six, but it was on the road. In short, Alabama’s win shouldn’t change anyone’s perception all that much, unless you’re leaning heavily into the eye test. But if you’re leaning heavily into the eye test, shouldn’t that pounding in College Park matter too? The committee is unlikely to elevate Penn State past anyone not named Oklahoma, and it isn’t going to get much attention, but there isn’t a great explanation as to why not.

The defining characteristic of the first set of rankings is how closely they resembled standings. The undefeated Power Five teams came first, then the one-loss Power Fives, then as many two-loss Power Fives as could fit, with Tulane and Air Force thrown in at the end. This makes LSU an interesting figure in these rankings. They were the top-ranked two-loss team last week, and they lost on the road against Alabama, hardly something to punish. Will they stay ahead of any two-loss Power Five teams? We’d assume they will (Iowa and UNC were unranked last week), which makes the real question how many.

Among teams who both won, our model has Kansas passing both Oregon State and Utah after the Jayhawks beat Iowa State, the Beavers beat Colorado, and the Utes beat Arizona State. It could happen, but I’d guess it doesn’t. The CFP rankings aren’t as much of a horse race as the AP Poll, but there are definitely horse race elements, especially further away from the top, where the rankings matter less and seem to get less attention from the committee itself.

Liberty remains a tough one for our model, which treats all Group of Five conferences alike in certain ways. Conference USA is a major step down from the AAC, Mountain West, and Sun Belt, and the committee treats it as such. We don’t have much of an opinion on how fair that is, but I don’t think this is the week Liberty climbs in. We think they can eventually, and we understand our model’s case that they’re the likeliest Group of Five representative in a New Year’s Six bowl, but we don’t think the ranking comes this week.

Kansas State took a top-ten team to overtime on the road, which shouldn’t hurt perceptions of them, but it wouldn’t be surprising if they fall out. The problem is who to add. Iowa and Duke weren’t all that inspiring on Saturday and Thursday, with Riley Leonard’s injury potentially especially hurting the Blue Devils. For what it’s worth, the next few teams in line, in our model’s eyes, are Arizona, USC, UNC, UCLA, Toledo, and Troy. Maybe someone like Kentucky gets in (the model has them 39^th), but whoever takes those last two spots, they won’t be inspiring. Our personal guess is that Iowa and UNC get in, with the committee deferring to overall win-loss record as it did last week, and that Arizona climbs in. So, if you’re looking for our confidence level in our own model’s perception of the top 25, we’d guess it’ll get 23 of the 25 teams correct this week, matching last week’s mark.