At 8:00 PM Eastern Time tonight on ESPN, the College Football Playoff committee will announce its first top 25 of the season. How much do these rankings matter? How will the top 25 line up? What should every team be looking for? We’ll touch it all, with plenty of help from our model.
How Much These Rankings Matter
The final rankings are the biggest deal, but these first rankings give us the most new information of any top 25 all year. The committee is offering its grade on the first eight or nine games for 25 different teams (and it’s really more like 30). The committee is saying what it makes of the first 70% of the college football regular season. That’s the majority.
These are also the only rankings the committee builds completely from scratch. Officially, the committee reevaluates every team every week. But we’ve been doing this for more than a decade now, and an unfortunate aspect of the AP and Coaches Polls has shown up in these CFP Rankings: They’re sticky. They don’t move around that much based on new results. The committee is more willing to reevaluate résumés than AP voters, but many of those unwritten AP Poll “rules” show up in the committee’s behavior. Teams rarely move downwards after wins. Teams rarely move based on changes in the quality of previous wins and losses (think: Oregon’s win over Penn State looking worse a few weeks later). Games in November don’t count the same amount or in the same way as games in September.
This makes these rankings more important than others in a weird way: This week is when the quality of previous wins and losses matters the most. If Oklahoma goes to Tuscaloosa in two weeks and gets blown to smithereens, Texas and Mississippi will be mostly unaffected. If Washington goes on a tear and finishes 10–2, Ohio State won’t get as much credit as if the Huskies had started 7–1 rather than 6–2.
Our model acknowledges a large amount of uncertainty around these rankings. There’s plenty of meat left on the bone, but in an average simulation, roughly twice as much FPA is handed out tonight as is handed out over the rest of the season combined.
What is FPA?
FPA is our model’s metric that accounts for differences between teams’ expected rankings (based on historic precedent) and their real rankings. The way our model ranks teams is to give everyone’s résumé a score, and to then calibrate those scores such that the top-ranked team in the country gets a 100.0 and the hypothetical bottom-ranked team (if the committee ranked teams 1 to 136 rather than 1 to 25) is at 0.0. FPA is added and subtracted such that the top 25 lines up in order.
How Will the Top 25 Line Up?
Here’s our model’s take. Plenty, plenty, plenty of notes in the next section:
1. Indiana (100.0)
2. Ohio State (96.8)
3. BYU (90.5)
4. Texas A&M (90.4)
5. Texas Tech (88.5)
6. Alabama (86.6)
7. Oregon (86.4)
8. Louisville (83.3)
9. Georgia (82.5)
10. Utah (81.9)
11. Mississippi (81.3)
12. Notre Dame (80.7)
13. Virginia (79.4)
14. Texas (78.4)
15. USC (78.0)
16. Miami (77.7)
17. Vanderbilt (76.6)
18. Michigan (75.5)
19. Georgia Tech (75.1)
20. Washington (74.9)
21. Oklahoma (74.8)
22. Iowa (73.2)
23. Missouri (72.5)
24. Cincinnati (72.1)
25. Illinois (71.9)
What Should Every Team Be Looking For?
We’re going to go through our model’s rankings in batches. For every team, we’ll explain what our model’s seeing, guess how correct or incorrect our model will be, and set a benchmark the team should be looking for tonight. We’ll include the first few teams on the outside looking in. Some of those might be in there tonight, and if they are, where they line up will matter.
1. Indiana (100.0)
2. Ohio State (96.8)
What our model sees:
A big piece of our model’s ranking formula is a metric called Adjusted Point Differential, or APD. All APD measures is how a team’s point differential in every game compares to their opponent’s average point differential on the season. Over the years, this on its own is fairly predictive of committee rankings. It’s especially a good proxy for how good the committee thinks teams are, independent of how accomplished they are. In APD, Indiana leads Ohio State and the rest of the nation by a touchdown. The Hoosiers are beating opponents by 36.8 more points per game than those opponents average. Ohio State is beating opponents by 30.6 more points per game than those opponents average. The third-place team comes in at 25.7. Indiana is lapping the field in blowout wins.
Add in that Indiana has one more win than Ohio State, and that one of their wins is the second-most impressive win in the country, at least the way our model measures it (beating Oregon by ten in Eugene), and the Hoosiers lead the Buckeyes in our model’s eyes.
Will our model be correct?
I don’t think so. The committee has an institutionalist streak. It doesn’t want to rock the boat. It’s pretty well-established that Ohio State’s a better team on paper than Indiana. Ohio State has plenty of good wins themselves (their second-best win—Illinois by 18 on the road—is better than Indiana’s in the eyes of our model). Ohio State is probably going to be #1.
What are Indiana and Ohio State’s benchmark rankings?
Ehh, who cares. If one is behind Texas A&M, that’ll be concerning for them, but that’s unlikely and not necessarily damning. Most likely, these guys are #1 and #2, and it doesn’t matter in what order.
3. BYU (90.5)
4. Texas A&M (90.4)
What our model sees:
BYU’s 8–0! As a power conference team! And while their wins aren’t bombshells…
- Utah grades out impressively, often blowing teams out.
- BYU’s 4–0 on the road.
- Texas A&M doesn’t get much credit from our model for anything besides the Notre Dame and LSU wins.
Will our model be correct?
I think not, and I think so. I’ll be shocked if Texas A&M is outside the top three—and to point something out, our model barely has Texas A&M outside the top three as it is, without any discount applied to the Big 12 and ACC relative to the SEC and Big Ten. (We still need more data to make sure that’s how the committee views it in this new Power Two/Four era. They were inconsistent on this last year.)
But BYU is one of four undefeated teams, and while they’ve played a lot of close games, their record in one-possession finishes is the same as Alabama’s: 3–0. I’m not saying BYU will be ranked 4th, but putting them behind all four of Alabama, Georgia, Oregon, and Mississippi like the AP Poll does is nonsense. Alabama lost to Florida State.
Maybe BYU has a public perception problem and that shows up tonight. But BYU is perfectly deserving of a top-four ranking for now.
What are BYU and Texas A&M’s benchmark rankings?
Texas A&M’s not a playoff lock, with three SEC games remaining plus potentially the SEC Championship. Anything worse than 3rd will be concerning, especially if the word “defense” gets thrown around when committee chair Mack Rhoades is asked about them. Texas A&M needs 10–2 to be enough. Texas A&M needs to be able to suffer two disappointing losses and stay inside the top ten. That probably means they need a top-3 ranking to begin with, which basically means they need to be treated like any other Big Ten or SEC unbeaten.
BYU’s also not a playoff lock, and they’re not relying on winning their conference to get in. They need an at-large bid to be an option. If they’re behind Texas Tech, whatever. They’ll play Texas Tech at least this weekend and maybe one more time. But if they’re behind Alabama, Oregon, and Georgia, the implication is that 10–2 from them won’t stand much of a chance without a lot of help. They can be ranked 8th tonight and still make the playoff. But they want to be ranked 4th. They want to be treated like an undefeated power conference team.
5. Texas Tech (88.5)
6. Alabama (86.6)
7. Oregon (86.4)
8. Louisville (83.3)
9. Georgia (82.5)
What our model sees:
This is a thick pack of one-loss teams. Louisville’s the one who sticks out, but Texas Tech is also worth examining.
Louisville doesn’t grade particularly highly in APD, but the Cards have some sneaky good wins. Everyone remembers the win over Miami on the road, but beating Pitt on the road and James Madison by two touchdowns at home is also important. One of those two teams might be ranked tonight. Pitt’s 7–2. Louisville’s only loss coming against Virginia on the road in overtime also doesn’t hurt. Virginia’s the current ACC favorite.
Texas Tech, meanwhile, is smoking teams, and they smoked one team who is also smoking teams. We mentioned that Indiana’s win at Oregon is, to our model, the second-most impressive of the season so far. What’s more impressive? Texas Tech beating Utah by 24 points in Salt Lake City. Maybe our model makes too much of margin when measuring those games, but that is an eye-popping result in hindsight.
Will our model be correct?
Oh I don’t think so. For Texas Tech, maybe. Our model does probably overrate the damage of the ASU loss, considering it came with ASU at fuller strength and Texas Tech playing what history will call its backup quarterback. I’d personally bet on the committee not rocking the boat and putting Alabama, Oregon, and Georgia all ahead of the Red Raiders.
As for Louisville:
This is a dumb unwritten rule, but it sounds good: “If two teams have the same number of losses and Team A beat Team B, Team A shall be ranked ahead of Team B.”
Louisville’s at risk of its whole résumé being reduced to the Virginia game. Louisville’s at risk of getting stuck behind Virginia like a tractor on a country road. It doesn’t help that Miami went and lost again. It doesn’t help that Florida State finally played good football for the first time since September, potentially saving Alabama from more serious examination.
What are Texas Tech, Alabama, Oregon, Louisville, and Georgia’s benchmark rankings?
Texas Tech needs the two-loss at-large path to stay open, but realistically, so much hinges on this BYU game on Saturday that their ranking this week doesn’t matter as much as it could. They’ll get reevaluated next week. Not everybody else will. They want to be ahead of Louisville and Virginia. It’d be good to be ahead of Mississippi too, but they won’t be sunk if they’re behind them.
Alabama would like to be ahead of Oregon, because Alabama’s got a bye to worry about in the event they win the SEC Championship. Even if they’re behind the Ducks, though, if Alabama takes care of business, Alabama will pass Oregon.
Oregon has enough good opportunities remaining that they’re probably not at risk of missing the playoff at 10–2. But, it’d be nice to at least be ahead of Georgia, who’s kind of a similar good–not–great team with one very fine loss and nothing outstanding in the win column.
For Louisville, it’s a question of how big the damage will be, at least watching through the lens of our model. Yesterday, our model had the Cardinals 37% playoff-likely. How much does that drop? The important thing for Louisville is that they line up ahead of all two-loss teams.
Georgia’s got meat left on the bone, similarly to Oregon. They’re also Georgia. 10–2 will get them in. Even being ranked 9th like our model says would be perfectly fine for the Dawgs.
10. Utah (81.9)
11. Mississippi (81.3)
12. Notre Dame (80.7)
13. Virginia (79.4)
What our model sees:
What a strange group.
Utah’s our first two-loss team, largely thanks to being the team with that third-best APD we mentioned above. Utah’s closest win this year was a 25-point victory over Wyoming. Utah has yet to win a game by three or fewer scores. They gave Cincinnati their only loss. The Utes’ own losses came to two teams who are a combined 16–1. It’s a great résumé for a two-loss team.
Mississippi and Virginia are two of our last three one-loss high-majors. Mississippi’s win in Norman is aging better than expected, but its win over LSU doesn’t look like much of a big deal anymore, and almost everything else is a red flag. Virginia has that nice win over Louisville, but the Hoos haven’t done much else. The NC State loss is bad, too.
With Notre Dame, we all know the story. They had two close losses the first two weeks. One of those came on the road against a team who’s now 6–2. The other came at home against a team expected to be ranked in the top three. Add in one good win over USC (15th-best in the country, per our model) and a good APD, and the résumé’s fine.
Will our model be correct?
I don’t know. The longer the committee looks at Utah and Mississippi, the more the committee will probably come in line with our model. But how long will they look?
Mississippi will probably be ahead of Utah, but they should be the last one-loss team aside from those from the ACC. There’s also been so much conventional wisdom pushed towards “Notre Dame’s the best two-loss team in the country” that it’s easy to see the Irish edging the Utes. Virginia could go either way, but the Louisville win should buoy them. Just as Louisville might get stuck behind Virginia, Virginia might get a little push from the more impressive Louisville and manage to stay ahead of this next batch of teams.
We’ll talk more about this when we get to Miami, but the Virginia/Louisville tractor/buoy thing is a possibility for Miami/Notre Dame as well.
What are Utah, Notre Dame, Mississippi, and Virginia’s benchmark rankings?
Utah would love to be the first two-loss team on the list. Since it probably won’t happen, the hope should be that Utah can at least stay ahead of everyone behind Notre Dame. You wouldn’t think that’d be that difficult, but the AP Poll currently has the Utes 17th, trailing all of Oklahoma, Texas, and Vanderbilt. Wacky.
Notre Dame would like to be that first two-loss team listed, but what Notre Dame really wants is to have some separation between themselves and Miami. Notre Dame wants to know that the committee views their résumé as better than that of a team who beat them by a field goal at that team’s home stadium. Our model has the Irish 88% playoff-likely if they win out despite expecting Utah to line up this week ahead of them. What would hurt Notre Dame a lot would be finding itself behind the Hurricanes.
Mississippi is in enough danger at 10–2 that they might have to get to 11–1 regardless. But for tonight, they should be relieved if they can slot in ahead of Utah, Virginia, and ideally Louisville.
Virginia’s fear is missing the playoff despite finishing 11–2, meaning missing the playoff despite losing the ACC Championship. Their fate in that scenario hinges on politicking, which is difficult for our model to estimate. For now, UVA should look for how many two-loss teams they lead. Ideally, they’ll lead all of them, something that would leave space for that 11–2 scenario to work out with especially chalky results elsewhere. But it’s an incremental situation. Every two-loss team ahead of them makes it a roughly equal amount worse.
14. Texas (78.4)
15. USC (78.0)
16. Miami (77.7)
17. Vanderbilt (76.6)
18. Michigan (75.5)
19. Georgia Tech (75.1)
What our model sees:
The first five of these are two-loss teams. Georgia Tech has one loss. Of the group’s combined eleven losses, Georgia Tech’s loss to NC State is more respectable than only Texas’s loss to Florida. Georgia Tech’s best win came against Duke, a 5–3 team our model would rank 47th if the rankings went that far. Georgia Tech has a bad APD.
In addition to having the worst loss of the group, Texas also has the most respectable loss in college football, the seven-point defeat in Columbus. Their APD isn’t amazing, and they would have benefitted from not making the Vanderbilt win so close, but their neutral-site win over Oklahoma looks decent enough to pass.
USC has a good APD (eighth-best in the country), respectable losses (Illinois and Notre Dame, the former close), and a nice win over Michigan at home by a lot of points.
Our model isn’t particularly worried about Miami’s losses, with both of them close, one of them on the road, and one of them at the hands of what our model thinks is a team currently in the committee’s top ten.
Vanderbilt has only respectable losses. Michigan’s aren’t bad either. Our model doesn’t directly consider head-to-head, so the transitive property intricacies of Michigan, Washington, Oklahoma, USC, Texas, and Vanderbilt go over its head.
Will our model be correct?
Probably not, just because it’s all so close. Although…
If playing the head-to-head game and playing it with zeal, Texas has to be ahead of Vanderbilt and Oklahoma (OU’s in the next section), USC and Oklahoma have to be ahead of Michigan, and Michigan has to be ahead of Washington. The only places that throws off our model are with Michigan vs. Oklahoma and Notre Dame vs. Miami. We’ll talk OU in a moment, but we do expect Michigan to be behind the Sooners.
USC’s been an AP Poll oddity all year, seemingly punished because last year they underperformed AP voters’ expectations. Is their narrative a burden like BYU’s might be?
Miami’s the other one I’m curious about. We’re seeing what might look like an ACC discount with Virginia and Georgia Tech. Will Miami get one, or will those wins over Notre Dame and USF help keep them upright? Could the win over Notre Dame even get that transitive property treatment?? They’re another one helped by Florida State winning so decisively on Saturday. That doesn’t make their win in Tallahassee look great, but it at least boosts some numbers the committee is shown.
What are Texas, USC, Miami, Vanderbilt, Michigan, and Georgia Tech’s benchmark rankings?
Texas would love to lead Utah and Virginia. Texas wants a shot at 9–3 if things get chaotic. (If Texas sweeps Georgia and Texas A&M, they’re almost definitely in. In that scenario, they also might win the SEC.)
USC’s path from here is Texas Lite. Win out, beating Oregon and thereby making noise along the way. Since Texas and Oklahoma will each probably lose again, leading the Longhorns doesn’t matter that much. But USC wants to be ahead of Miami and Michigan. Especially since Michigan could also make a whole lot of noise themselves against Ohio State.
Miami still has an at-large shot if they show up this high in the rankings. Miami is in a great spot if the committee decides they need to be ahead of Notre Dame. Drop behind all these teams and Oklahoma, though, and the Hurricanes might be out of it.
20. Washington (74.9)
21. Oklahoma (74.8)
22. Iowa (73.2)
23. Missouri (72.5)
24. Cincinnati (72.1)
What our model sees:
Five power conference teams with two losses. Oklahoma has head-to-head losses to Texas and Mississippi. Mizzou’s are to Vanderbilt and Alabama. Washington’s are to Ohio State and Michigan. Iowa’s are to Indiana and Iowa State. Cincinnati’s are to Utah and Nebraska.
Oklahoma has the best wins in the group, wins over Michigan (H) and Tennessee (A). Washington’s loss to Ohio State is the difference there, in our model’s eyes. Similarly, Iowa playing Indiana closer than anyone else has played them keeps the Hawkeyes ahead of Missouri for our model.
Will our model be correct?
I’d expect Oklahoma to be up in that next pack, thanks to a lot of different factors ranging from public narrative to the strength of the SEC to the head-to-head win over Michigan. I’d kind of expect Missouri to be ahead of Washington and Iowa, given they’ve enjoyed more respect so far this year, but Missouri’s best wins are Auburn and Kansas. That’s not inspiring.
I wouldn’t be that surprised if Cincinnati’s unranked, but I’d be a little surprised. We’ll get to a few alternatives in the final pack.
What are Washington, Oklahoma, Iowa, Missouri, and Cincinnati’s benchmark rankings?
Washington, Iowa, and Cincinnati all need to win out. If they’re ranked tonight and they win out, they probably have a shot. Hard to get too worked up about everything else just yet.
Oklahoma might have a chance at 9–3 in a chaotic scenario. They’ll want to lead USC, Miami, and Vanderbilt.
25. Illinois (71.9)
NR. James Madison (71.5)
NR. Pitt (70.1)
NR. North Texas (70.0)
NR. USF (69.6)
NR. TCU (66.3)
NR. Tennessee (66.3)
NR. Memphis (66.2)
NR. San Diego State (66.1)
What our model sees:
Our model gives 6–3 Illinois the last spot ahead of two-loss Pitt, two-loss TCU, two-loss USF, three-loss Tennessee, and one-loss JMU/UNT/Memphis/SDSU. Why? Tolerable losses (Ohio State, Indiana, and Washington on the road) and a nice win over USC.
Will our model be correct?
There are three ways the committee can go: First, they can stick their neck out for a three-loss team. Second, they can give a two-loss power conference team the spot and avoid anyone asking hard questions. Third, they can rank exactly one mid-major, to show their hand.
If it’s the first, Illinois is more deserving than Tennessee but Tennessee has been more prominent in the narrative. If it’s the second, Pitt and TCU would both be perfectly acceptable. Houston’s the other two-loss power conference team, but they’re a few more slots down on the list. If it’s the third, the options are either to rank JMU or to get into a head-to-head whirlpool in which Memphis beat USF who beat North Texas.
If I were the committee, I’d just rank JMU. But none of these outside of San Diego State would surprise me. The committee often includes someone who doesn’t belong by any objective measurement. Remember UNLV last year?
What are Illinois, James Madison, Pitt, North Texas, USF, TCU, Tennessee, Memphis, and San Diego State’s benchmark rankings?
They don’t really have them. For those who still have a playoff chance, being unranked isn’t a death sentence. Most of those playoff chances are through automatic bids anyway.
One team, though, should be on guard.
If James Madison is ranked behind a team from the American *right now*, that is a terrible sign for JMU. Those teams get to keep playing each other. JMU doesn’t have as many opportunities remaining. JMU doesn’t need to be ranked. But assuming they aren’t, they need no one from the American to show up in that top 25.
What Comes Next
We’ll update our model after tonight’s rankings and games. We should have a new post up tomorrow with bracketology, playoff probabilities, and how far our model’s estimated top 25 deviated from the real thing. Expect uncertainty to shrink a lot in those simulations. We’re getting a lot of information tonight.
**
