Four weeks left. A big Saturday ahead. Plenty of bumps and bruises, but still a lot of hope. Before this weekend kicks off, let’s look at the playoff picture.
We’re not going to do all the graphics this time (that time of year), but if you’re looking for probabilities and average final rankings, you can always find those here. Use that page on your desktop. The search function is your friend.
Now. Here’s what our model’s seeing after Tuesday’s rankings and the weeknight games:
Utah Out, Texas In
The biggest change in our bracket post-rankings is the inclusion of Texas instead of Utah. Our model thinks Utah (43%) is likelier than Texas (34%) to make the playoff, but Texas’s upside is so enormous that it has the better average final ranking by fractions of a position. Texas’s average season is better than Utah’s. Utah has a better chance to make the playoff. Our bracketology works off the average season thing.
We’ll get into this a little more later (“Next in Line”), but the way the picture shapes up, there’s currently one “open” at-large spot. Nine of these teams—ten if you count USF—are safely in if they handle business. The ACC and James Madison will combine to contribute one automatic bid. That leaves an open spot. Right now, our model has Utah as the likeliest to get it but Texas in the bracket. Neither is likelier than a coin toss.
The Georgia–Bama Shuffle
On Sunday, our model expected Georgia to be ranked ahead of Alabama. Obviously, it shouldn’t have expected that, as we said at the time. Why did it?
Ostensibly, the committee is ranking how good every team is. What this is sometimes said to mean, and what’s partly true, is that the committee is ranking how good every team’s résumé is. What’s really happening is that the committee is ranking a combination of those two things—how good a team is, how accomplished a team is—with two serious caveats:
- If teams are close to one another and one beat the other, the head-to-head winner will be ranked ahead of the head-to-head loser.
- As those teams’ seasons go on, the head-to-head loser can almost never jump the head-to-head winner unless the head-to-head winner loses.
Those two caveats are tough to include in a model. Yes, we need to figure out how to get them in there, but they’re tough, and the second is a lot tougher than the first. Here’s what I mean:
Prior to the first rankings, our model said, “Alabama’s a lot better than Georgia. They’ve been winning more convincingly, their best wins are more impressive, and their loss is bad but not bad enough to outweigh those other three things.” Our model expected Alabama to be ranked 6th and Georgia to be ranked 9th.
The first rankings came out, and Alabama was ranked 4th while Georgia was ranked 5th. “No problem,” our model said. “The committee sees these two teams as decently close together. Georgia’s better than Texas Tech and Oregon. The committee thinks highly of them. The first eight games of Alabama’s schedule are, right now, only one spot better than the first eight games of Georgia’s. Got it.”
Last weekend, Alabama played fine but wasn’t impressive. Georgia, meanwhile, beat the pants off of Mississippi State on the road. What did our model say? “Wow. Georgia just showed us something new. That should change the committee’s perception of them by more than one spot. And since Alabama played about as expected and saw their loss get a little worse, Alabama should move down if they move at all.” Up went Georgia in our model’s projections. Past Alabama.
Then, on Tuesday, Alabama was ranked ahead of Georgia. “No problem,” said our model. “I guess the first eight games of Alabama’s schedule are worth more than they were worth a week ago.” Next week, if Alabama and Georgia both win, we’re probably going to do the same thing again.
This is, again, a flaw with our model. The CFP rankings are a partial horse-race. They’re not as bad as the AP Poll in this respect, but prior rankings are an important variable in future rankings, and not only because they reflect what the committee thought of a team a week ago. The CFP committee ranks teams in part based on how it ranked teams a week ago, independent of what that ranking meant. Alabama was better than Georgia, so they must still be better than Georgia in the eyes of the committee. No matter what else happened, shy of Alabama losing or Georgia doing something titanic.
We improved this piece of our model this year, but we still don’t have enough of it built in.
Mississippi Isn’t Sacred
We were curious whether Texas Tech might jump Mississippi, and we’re relieved to see it did. Mississippi beat Oklahoma on the road and blew out Tulane. Every other one of its results is either fine or concerning. It’s good to see the committee acknowledge that. Very strange that they’re still ahead of Oregon, but Oregon at least has chances left to do what Texas Tech did.
Miami Up, BYU Down
Outside the top 25, Miami moved up a bit and BYU dropped. One of the funny things from this set of rankings: They implied that the ACC is the second-best conference in college football. What do I mean?
If you don’t include any conference-level adjustments in our model (and for the most part, we don’t), only two conferences are currently outperforming their expected rankings. The first is the SEC, which makes sense. That league is deep, and it’s fair for the committee to respect that. The second is the ACC, which doesn’t make sense. The ACC is not deep, and it’s not strong at the top. But the committee didn’t want to move Virginia out of the top 25 after it lost with its starting quarterback hurt, and the committee’s tied to putting Louisville behind Virginia because of those two weird pseudo-rules we mentioned above.
I don’t think this is what happened (I don’t think the committee thinks about it on this level), but you could interpret this move as the committee planning ahead. There was a news cycle early this week centered on the possibility of Duke winning the ACC but finishing ranked behind both James Madison and the AAC champion. We asked: Could the whole ACC miss the playoff? Thanks to this week’s committee forgiveness towards Miami (and Virginia, etc.), that’s a little less likely. Not because Duke will magically get in if they win the ACC, but because Miami’s a serious contender for an at-large bid now.
Four Kinds of Teams: Bye-Curious
Moving on to the playoff picture as a whole: We’ve got, more or less, four types of teams. The first is the ones who are pretty safely making the playoff, but who are chasing a bye. Ohio State. Indiana. Texas A&M. Georgia. Alabama. Maybe Texas Tech, depending how anxious you are about their chances at 11–2.
Four Kinds of Teams: No Room for Error
After those guys, you have the teams who are safe for now but won’t be safe if they lose again. Because of the bizarre weakness of their résumé, Mississippi is in this group. Because of the bizarre treatment of their résumé, Oregon is in this group. Notre Dame is in this group too, unless Texas and Oklahoma both win this weekend (that’s very unlikely, and even then, Notre Dame would have some outs). If you don’t have Texas Tech in the group above, that means they’re here instead. Also here: USF, who will get an automatic bid if they take care of business.
Four Kinds of Teams: Next in Line
If you’ve been keeping track, that’s ten teams, with one automatic bid unclaimed that will probably go to the ACC. As we mentioned above, that means there’s at least one open spot. But the words “at least” are important. If Mississippi, Texas Tech, or Oregon loses, there might be another open spot. If Notre Dame loses, there’ll be another open spot. If USF loses, it doesn’t quite work that way, but there’ll at least be another open automatic bid, with JMU, the ACC champ, and the AAC champ fighting over two automatic slots instead of JMU and the ACC champ fighting over one.
Who’s in this next group? Right now, it’s a big group. Texas. Oklahoma. BYU. Utah. Miami. Vanderbilt. USC. Michigan. Those are the eight. Important notes:
If Oklahoma beats Alabama, they move up to the “No Room for Error” category, meeting the Crimson Tide in the middle. If Oklahoma beats Alabama, there are no open spots and everyone is hoping on someone in the top ten to lose.
If Texas beats Georgia, they’re still on the bubble. They might be favored against Texas A&M, but it’s harder to say, “Texas just needs to take care of business and they’re in.”
If BYU wins out and loses again to Texas Tech, they’re probably in. They’d finish the regular season 11–1 with their only loss coming against what’ll probably be a top-five team. In this sense, BYU’s the leader among this group, at least if you have them favored against Cincinnati. BYU can take care of business and probably make the playoff. It isn’t as simple as it is for Notre Dame and teams above them, but BYU’s in the driver’s seat more than Utah, Vanderbilt, Miami, Michigan, or USC.
Four Kinds of Teams: Not Dead Yet
After that group, it’s a lot of starving coyotes, a group of teams either hoping for total chaos (Iowa, Tennessee) or grappling for an automatic bid (North Texas, Tulane; James Madison; Georgia Tech, SMU, Pitt, Virginia, Duke, Louisville; Cincinnati, Arizona State, Houston). Some are in worse shape than others (JMU’s almost 50% likely to get in, with a losing hand but a couple different outs). But none of them control their fate, and unlike that group above, none can really hope to finish ranked in the top nine or ten.
FPA’s
FPA, or Forgiveness/Punishment Adjustment, measures how far each team’s ranking is from our model’s expectations. The scale is such that the top résumé in the country is at a ranking score of 100.0 and the worst in the FBS is at 0.0.
In the list below, each team is listed with three numbers next to their names. The first is their ranking score, that number between 0.0 and 100.0. The second is the total FPA a team is receiving—how far from their expected ranking score the committee is implying they land. The third is this week’s FPA—how much the committee seemed to add or subtract this week beyond what we’d expect based on the results on the field.
What our model made of this week’s rankings, in order:
1. Ohio State (100.0, +2.0, –1.2)
2. Indiana (97.9, –3.0, +1.2)
3. Texas A&M (95.8, +1.0, –0.2)
Our model doesn’t actually know how close the committee has these three together, so it assumes they’re evenly spaced. This week, that led to our model dialing back on how much credit it thinks Ohio State is getting and ramping up how much credit it thinks Indiana is getting. This is one of my least favorite aspects of our model, but the method we used to use (moving teams the minimum amount possible from where precedent suggested they should be ranked) wasn’t working in testing for the 12-team era. It was the Alabama/Georgia phenomenon we talked about above, but on steroids.
4. Alabama (89.6, +2.0, +0.7)
5. Georgia (89.1, +3.0, –1.7)
6. Texas Tech (88.3, –5.3, –1.5)
7. Mississippi (87.4, +2.5, –2.5)
8. Oregon (86.7, –2.2, +0.5)
In order: Alabama got more credit than before. The committee dialed it back on its Georgia love in order to keep them behind Alabama. The committee increased its doubt of Texas Tech relative to our expectations, part of a broader trend where the committee doesn’t seem to respect Big 12 résumés very much right now. Mississippi’s FPA was mercifully cut in half. Oregon got a little bit of a reevaluation after beating Iowa, but they’re still significantly lower than precedent would suggest they’d be.
9. Notre Dame (85.0, +0.2, –0.2)
10. Texas (84.1, +4.0, +2.1)
11. Oklahoma (83.2, +5.0, 0.0)
12. BYU (82.4, –5.8, –0.8)
13. Utah (81.6, –2.5, +0.1)
14. Vanderbilt (80.9, +2.5, +1.7)
15. Miami (79.9, –0.2, +1.3)
16. Georgia Tech (79.1, +1.5, –0.3)
17. USC (78.3, –4.2, –1.9)
18. Michigan (77.5, –0.5, +0.4)
Again starting from the top, but only mentioning teams whose FPA changed: Texas got a lot more credit than before. This was done in order to keep it ahead of Oklahoma despite Florida’s blowout loss to Kentucky worsening Texas’s résumé. I doubt the committee realized it did it. BYU got a little extra punishment—more of that Big 12 disrespect from the committee. Vanderbilt got more FPA in order to keep the order the same despite a gross showing from the Commodores against Auburn. Miami got a lot of love and USC got some punishment, again in order to keep the order constant after USC’s relatively impressive win against Northwestern. Had the committee started this week instead of last week, maybe USC would be ranked ahead of Miami. But the committee makes itself beholden to its prior rankings.
19. Virginia (75.2, +2.2, +3.0)
20. Louisville (74.7, –1.8, +3.4)
21. Iowa (74.2, +2.0, +0.2)
22. Pitt (73.6, +2.0, +0.5)
23. Tennessee (73.1, +3.7, –1.2)
24. USF (72.7, +0.5, +1.2)
25. Cincinnati (72.2, –2.7, –1.2)
NR. Illinois (71.7, –2.2, –0.3)
NR. North Texas (71.2, –1.0, –0.4)
NR. James Madison (70.7, –2.7, +0.4)
Our model is guessing on the unranked teams, but it thinks all of Illinois, North Texas, and James Madison are in the picture. Other movement:
Virginia’s getting some loss forgiveness, probably because Chandler Morris got hurt early during the loss to Wake Forest. This allowed the committee to release some of Louisville’s punishment (Louisville’s only behind Virginia because of the head-to-head loss). Tennessee should have moved up based on what the committee said about their résumé last week, but the committee dialed it back. Liberated from their head-to-head loss to Memphis, USF moved up to a more reasonable ranking. Cincinnati took on a little more Big 12 punishment. I don’t know why the committee likes the ACC, dislikes the Big Ten, and despises the Big 12. But that’s what’s happening. At least for now.
**
