The Committee’s About to Contradict Itself, and That’s (Mostly) Ok

Every week, our college football model predicts the College Football Playoff committee’s rankings, and this week, it’s going to be wrong.

Every week, to be fair, our model is wrong. There’s uncertainty within the model, it expects to be wrong, that’s why the page we link you to shows various probabilities between 0.0% and 100.0%. But, we know a specific thing it’s going to be wrong about this week, and this is a problem. It’s a problem with our model, yes, but it’s also a problem with the committee. Because the model follows the committee’s precedent, as one would hope the committee would, as would make the rubric the same every year—the closest thing to fairness we can insert into this subjective process—when the model is wrong, it means the committee is deviating from precedent. And in this case, it’s not even because of something reasonable. It’s purely situational.

What’s going on? Well, let’s share where our model sees tonight’s top 25 coming out. The numbers next to each team here are a score on a 0.0-100.0 score in which the top-ranked team nationally is at 100.0 and the theoretical bottom-ranked team nationally (UMass, this week) is at 0.0.

1. Georgia (100.0)
2. Ohio State (97.5)
3. Tennessee (93.4)
4. Michigan (92.3)
5. TCU (89.0)
6. Alabama (85.9)
7. Clemson (84.2)
8. Oregon (83.6)
9. USC (81.8)
10. LSU (80.3)
11. Mississippi (79.5)
12. UCLA (79.2)
13. Utah (78.6)
14. Penn State (77.1)
15. UNC (75.5)
16. NC State (71.5)
17. Texas (71.1)
18. Tulane (70.8)
19. Kansas State (69.2)
20. Notre Dame (69.0)
21. Florida State (68.6)
22. UCF (67.9)
23. Oklahoma State (66.8)
24. Illinois (66.8)
25. Liberty (66.4)

Notice anything? UNC as the 15th-best team in the country? No, that’s just where the committee will probably have UNC. Yes, we’re talking about Alabama and LSU. Alabama and LSU, who our model believes will not move from last week’s positions.

I’m not going to argue that Alabama should get more credit than LSU for Saturday’s respective performances. That isn’t what this is. LSU beat Alabama, it wasn’t by as much as home-field advantage is worth, but it was a win. That is a reasonable way for sports to work. We need home teams. We need final scores, not comprehensive metrics grading each team’s backwards-looking win probability (though those are quite useful for other purposes).

I’m also not going to argue that Alabama should be ranked ahead of LSU. That is also not what this is. I have opinions on how these rankings should be handled, but I’m not here to talk about my opinions (in these few paragraphs, anyway).

What is true, though, is that on the committee’s traditional grading scale, Alabama has a better résumé than LSU. Who’s the better team? Put the two on a neutral field next week and Bama’s a ten-point favorite. Who has the better losses? Alabama lost by a combined four points on the road against two top-ten teams. LSU lost to a team just now cracking the top 25 and got blown out at home by a team Alabama nearly beat on the road. LSU does have the better best win, at least for now (Alabama gets to play Mississippi on the road on Saturday, and winning by 20 or more would be more impressive than anything LSU has done, at least by our model’s estimation scale), but call it a split decision. What the committee is likely to do is fair, but it’s selective. You consider head-to-head results and results against common opponents? How do you choose which?

I don’t think this is necessarily a bad thing. It’s a fine way for the sport to work, really, and no Alabama fan should be reasonably outraged if the Tigers jump them tonight. This is, after all, how the SEC West standings will now stand, and the SEC West tiebreakers are just and well-reasoned. But it gets at an important point about the committee, and a disappointing one for those of us who like to put a percentage on everything: These things are situational in nature. And there are more situations out there, in all the parallel universes, than we could ever both imagine and accurately predict the committee’s actions within. What we do isn’t fruitless—there’s about a 70% chance Georgia or Ohio State wins the national championship, that’s pretty accurate—but it has its limitations. Because the committee is inherently inconsistent. Because the committee is inherently unreasonable. Because that’s how we want the committee to be.

Let’s talk about questions for tonight:

How High Does LSU Rise?

LSU is in a strange place in our model at the moment, because with how much of a boost our model knows the committee gave them last week, and with its estimation of possibilities for committee forgiveness bestowed from here onward, they’re still not quite tall enough to reach the swing set. They win the SEC in 10.8% of simulations. They make the playoff in 0.6%. That’s roughly a 1-in-20 conditional probability. Where do they cross the threshold where the playoff becomes plausible? I’m not sure. USC’s at around a 1-in-4 conditional probability, calculated the same way. Oregon’s at about a 50/50 situation. Clemson’s at about a 1-in-6 shot. In all of these, though, there are scenarios where the playoff case is undone but the conference championship is still won, and the probabilities we end up in that boat with various teams vary from team to team. We’re not really comparing apples to apples. So, we can’t tell you exactly how much each additional spot in the rankings helps LSU, but each spot does help. Each spot better than 10th that the Tigers are ranked, the playoff probability rises that much more.

How Fall Does Alabama Fall?

In a similar vein, Alabama will likely be ranked worse than 6th, and how much worse will deflate their own playoff chance in reality and in our model’s eyes. Alabama still has some sort of chance to win the SEC. They need LSU to lose to both Arkansas and Texas A&M, but that’s possible. They also might not need to win the SEC to make the playoff, though. Should LSU lose exactly one of those, LSU is toast, and Alabama is still finishing 10-2. Should two-loss teams win the Big 12 and Pac-12—possible, if not likely—Alabama is a better team on paper with the same number of losses. It turns into a 10-2 Alabama vs. 11-1 Michigan vs. 12-1 Clemson situation, or something like that, and the committee would probably just choose Clemson, but they’d be clearly disregarding their instructions to choose the best four teams if they did that.

Clemson? Michigan?

We assume Michigan will be fourth, but there’s a chance they’ll be third. Is their 9-0 better than Tennessee’s 8-1? The answer matters, but it doesn’t matter as much as it could. At this point, we’re pretty convinced Tennessee’s 11-1 will be better than Michigan’s 11-1 in scenarios (not all that unlikely of scenarios) where that becomes a decision point.

Clemson, meanwhile, is very much alive, but if they drop past Oregon, that liveliness drops off by a lot, and if they drop past USC, they’ll be getting down close to Utah and Alabama territory on the believability index. A one-loss Power Five champion has a chance until they don’t, and last week the committee was more than happy to give the ACC its roses (for unknown reasons), but the way Clemson lost and the narrative coming out of that loss could persuade the committee to do some reevaluating of that entire league. Notre Dame is the ACC Champion right now and Notre Dame doesn’t play in the ACC. Notre Dame also isn’t likely to be in the top 15.

Does Illinois Stay Ranked? Does Wake Forest?

Illinois’s ranking is potentially a big deal for Michigan, though the Illini could always slide back in next week when one of UCF/Tulane and presumably another few ranked teams lose. Even though Michigan State beating Illinois makes Michigan’s win over MSU all the better, Michigan wants Illinois to be thought of highly, not MSU. Because: The committee weights the extreme ends of the single-game performance list more heavily than the middle. Your three best wins matter, and your losses matter, and beyond that you can kind of do whatever the hell you want as long as you aren’t making too big a habit of flirting with disaster.

Wake Forest’s ranking is potentially a big deal for Clemson, who could easily technically finish with five top 25 wins (UNC, Florida State, NC State, Wake, and potentially a re-entering Syracuse, who’s almost definitely dropping out this week, or an emerging Louisville, who could do just about anything these last three games and still not surprise) and give the committee quite the talking point to shout over those of us wailing, “But they’ll be 20-point underdogs against Georgia!”

**

There are other questions, but these are the big ones. My personal guess as to where we’re headed? I think I listed this on Sunday, but it’s probably 1) Georgia, 2) Ohio State, 3) Tennessee, 4) Michigan, 5) TCU, 6) Oregon, 7) LSU, 8) Clemson, 9) Alabama, 10) USC, or something in that vein. I don’t see these guys keeping LSU above Alabama, but they seemed to like Oregon last week, and distinguishing between the one-loss Power Five leader who lost on effectively the road to Georgia and the one-loss Power Five leader who got stomped by the team that lost to Marshall is something the committee probably wants to do. They didn’t seem to believe in USC last week, and USC didn’t give much cause for belief this Saturday night.

At the bottom, I do think Liberty comes in after beating Arkansas. Oklahoma State’s trend of collapse might pull the Pokes out (Washington could come in instead, or a half-dozen other teams, but I’d guess Washington), Illinois’s loss could pull them out but probably won’t, who really knows the committee clearly spends way less time on the back end of the top 25 than the top end, which is reasonable but twists into unreasonability when they then use “top 25 wins” as a talking point between the guys at the top.

The show’s at 7:00 PM EST on ESPN. Reactions from us sometime in the 24-ish hours afterwards.

The Barking Crow's resident numbers man. Was asked to do NIT Bracketology in 2018 and never looked back. Fields inquiries on Twitter: @joestunardi.
Posts created 3304

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts

Begin typing your search term above and press enter to search. Press ESC to cancel.