College Football Playoff Rankings, Week 12: Does Georgia Pass Ohio State?

Georgia pounded Mississippi, and Michigan passed its first test with an A. Where does that leave the top 25?

Every week, our college football model projects the College Football Playoff rankings. Every week, our model then reacts to the rankings, changing its forecasts for future rankings based on what the committee has shown so far. We have a detailed explanation of how the model approaches this estimate, but the short version is that the model starts by ranking the teams according to demonstrated committee precedent over the years and then moves teams from their forecasted position to their real position following each set of Tuesday rankings. Then, it carries this change forward in its simulations of future rankings, including—ultimately—the final CFP rankings upon which the playoff field and New Year’s Six bowl matchups are selected.

This approach doesn’t always give us the most accurate prediction of the Tuesday rankings. It’s great at the end of the season, especially at the top, but our model sometimes expects the committee to react to games in ways that are consistent with the committee’s own logic but inconsistent with the way college football rankings are approached. We do think the committee is willing to shuffle rankings, moving teams down after wins, but there’s still an instinctive horse–race nature to the rankings. So, after we share the model’s forecast for tonight’s rankings, we’ll talk through what we think it’s getting right and what we think it might get wrong below.

Here are those rankings.

Starting from the top:

We’ll see about Georgia. The committee was giving them the benefit of the doubt before this week, ranking them above their accomplishments and therefore probably pricing in some future victories like Saturday’s. Still, Saturday’s performance was impressive. They smoked those guys. If you asked a large sample of college football watchers who the best team is in the country, many would say Georgia. That probably applies to the committee as well.

Michigan doesn’t have much of a case to jump Ohio State. They didn’t beat Penn State more emphatically than the Buckeyes did, and Ohio State still has the road win over Notre Dame, a top-20 team. Michigan won’t slip past Florida State or Washington, but it’s hard to see them passing Ohio State, and passing Georgia would be a surprise.

Our model has the gap very narrow between Florida State and Washington, but if it took away what else the committee has shown us, it would have Florida State ahead by virtue of having won by wider margins, something the committee tends to value in roundabout ways. That said, FSU did not win by a wide margin this weekend, and while Washington didn’t look sensational, they did beat a ranked opponent, which could pull them upwards. This race probably doesn’t matter all that much—for one thing, Washington’s an underdog to win the Pac-12; for another, each team is in the playoff if they win out and will need help making it with a loss—but it will get attention, and understandably so.

Oregon is in a straightforward position.

Our model doesn’t expect Alabama to pass Texas, even with Texas winning underwhelmingly again and Alabama dominating. Alabama’s domination wasn’t a big departure from their other performances this year. It’s a little noteworthy that our model thinks this, because our model doesn’t consider direct head-to-head results (those are generally weighted more by the public than the committee, though this year has been an aberration). It’s something to watch, but we’d be stunned, and our model would be surprised.

Penn State’s loss was objectively a better performance than Mississippi’s, even if it was uglier. We don’t really think Penn State will pass Mississippi, but precedent would suggest the Nittany Lions should be ranked higher, as has been the case all year so far. Some of this is our model always assuming the minimum amount of departure from precedent.

What’s more interesting is whether Penn State and Mississippi fall past Louisville or not. Louisville has a path to 12­–1 and a Power Five championship, traditionally a playoff résumé, but they also have the one thing that’s doomed 12–1 P5 champs: A terrible, horrible, spectacularly disappointing loss. Only one or two other ranked teams have a loss as bad as Louisville’s loss to Pitt. That’s before we get to how the committee views Louisville’s escape from Pitt. We’d personally guess the committee puts the Cards ahead of all the two-loss teams, but they’d be justified in holding Louisville back, and it would align with precedent.

Missouri smoking Tennessee might push them past Oregon State, and that could theoretically matter in some New Year’s Six considerations.

After Oregon State, there’s a whole mess of teams, and it’s fairly inconsequential where most of them land. We doubt Notre Dame will pass LSU after LSU won by 17 against Florida (rough look for Florida’s résumé, though some of this is Notre Dame getting help from Ohio State, Clemson, and NC State). We don’t know how far Utah and Tennessee will fall, though our model’s estimate seems like a fine guess. We don’t know how high Iowa will rise, but they will probably rise. People seem to be noticing the defensive piece of their identity, and what a piece that is. Arizona is unlikely to fall, even after making the win over Colorado a tough one.

Our model keeps pushing Liberty higher, as Liberty keeps winning by wide margins, but as we’ve mentioned previously, our model treats all Group of Five conferences the same and that’s looking more and more like a systematic flaw. Conference USA is not the same as the AAC, Sun Belt, or Mountain West. The MAC isn’t the same as those leagues either, which means our model is probably overestimating the committee’s perception of 9–1 Toledo. Troy continues to be a significant character in the race for a New Year’s Six spot, as does SMU, whom our model has 35^th but 30^th if you give them the same FPA as fellow AAC member Tulane (FPA is Forgiveness/Punishment Adjustment, the number our model assigns to account for the committee’s deviations from expectations). Our model would have James Madison 13^th if the Dukes were fully bowl-eligible. That might be a stretch, but only if the committee hasn’t caught on to how good the Sun Belt has become. (Movelor, our model’s rating system, has the average Sun Belt team nearly a full point better than the average AAC or Mountain West team.)

Personally, I’d guess Kansas and Oklahoma State remain ranked in place of Liberty and Toledo, and that if it isn’t those two, it’s SMU or NC State (whom our model has 34^th). Kansas is especially likely to receive a kind look given Jason Bean exited their loss with an apparent head injury. I don’t think Duke will enter after a loss, and I don’t think Texas A&M will enter after firing their coach. I do think Liberty might be ranked—at some point, our model has to catch up to the committee on them—but I would be very surprised if they’re higher than 25^th. I believe our model that Liberty is the likeliest Group of Five NY6 representative, but I think it’s going to take a while to get there and require an additional loss each from Tulane and SMU.