Relative to most who talk about the College Football Playoff rankings, I don’t often find myself criticizing the committee. Relative to most who talk about power dynamics around the playoff, I don’t often find myself criticizing ESPN for pro-SEC bias. The committee usually does the commonsense thing. At ESPN, sincerer voices usually carry the day.
Last night was a little different.
First, Tulane shouldn’t be ranked. They lost to UTSA by a lot of points. Their best win came against a 5–5 team from the ACC. The committee has a tendency to do wild things in the 21st to 25th range, and the committee has a tendency to do wild things with mid-majors, but this is baseless. This is, “We’ve heard of Tulane and we haven’t heard of North Texas or James Madison.” The committee needs to spend more time on the mid-majors. It’s one thing to include someone like Arizona State at 25th. Weird choice, but whatever, it doesn’t affect anything. It’s another to ignore UNT and JMU’s better bodies of work than Tulane’s. That could change who makes the playoff, which could change history at those three schools.
Second, Andrea Adelson’s “report” on the rankings—ESPN’s Top Headline last night on the website and the app—was transparently PR. The discussion of Miami and Notre Dame was reasonable enough and, thanks to the Hunter Yurachek quotes, informative. But the discussion of Alabama and Notre Dame was a clumsy attempt to set a pro-Alabama narrative. It focused on each team’s ranked wins (four for Alabama, one for Notre Dame) but buried how Notre Dame beat a ranked team this weekend despite Yurachek mentioning that as a factor in the rankings. It said an SEC Championship appearance for Alabama “would all but guarantee a spot in the CFP no matter the result,” which isn’t true, as we’ll explain below. Most disingenuously, nowhere in the column did Adelson mention that Alabama lost to Florida State, again despite Yurachek specifically naming that loss as a contributing factor in the ranking decision. Plenty of attention paid to Notre Dame losing to Miami in Week 1. No mention of Alabama suffering a much worse loss two days earlier.
For those of you who remember the time we talked about the danger of journalists interweaving analysis with reporting, this is what we’re talking about. Ostensibly, Adelson’s including analysis to give more context to the facts. Really, she’s using it as a Trojan horse to carry opinion—hers or her bosses’, I don’t know—into what should be a straightforward piece sharing Yurachek’s comments with those who didn’t watch the show. Couple that with the omission of Yurachek’s comments on the Bama–FSU game, and you’ve got a news article that’s intentionally misleading. If ESPN reporters are going to carry water for the SEC (and ACC, where Miami would also benefit from Notre Dame slipping), they could at least show fans respect and try to be subtle about it.
With that out of the way, let’s go through the new Top 25 then take a look at our model’s updated bracketology. As always, the number in parentheses is each team’s “ranking score,” our model’s estimate of where they line up between the best résumé in the FBS (100.0) and the worst (0.0). As always, FPA stands for “Forgiveness/Punishment Adjustment,” and it’s a measure of how far each team’s ranking score is from our model’s baseline, precedent-based expectation. Total FPA is what it sounds like. New FPA is what our model estimates the committee added or took away this week. We’ll add commentary where it’s relevant.
Week 12 FPA’s
1. Ohio State (100.0)
Total FPA: +1.9
New FPA: –0.1
2. Indiana (98.0)
Total FPA: –2.6
New FPA: +0.4
3. Texas A&M (96.2)
Total FPA: +1.1
New FPA: +0.1
4. Georgia (91.4)
Total FPA: +2.9
New FPA: –0.1
5. Texas Tech (90.3)
Total FPA: –5.3
New FPA: 0.0
6. Mississippi (89.2)
Total FPA: +2.4
New FPA: –0.1
7. Oregon (87.5)
Total FPA: –1.9
New FPA: +0.2
8. Oklahoma (86.7)
Total FPA: +4.8
New FPA: –0.2
9. Notre Dame (85.9)
Total FPA: –2.3
New FPA: –2.4
10. Alabama (85.2)
Total FPA: +1.3
New FPA: –0.7
Notre Dame and Alabama got a lot of headlines, and what the committee effectively said with those two was: Notre Dame’s win over Pitt wasn’t anything to write home about, and Alabama’s loss to Oklahoma was a little more concerning than the final score alone would indicate.
We mentioned yesterday that the committee could give itself a pathway out of selecting a hypothetical 10–3 Alabama—an Alabama who makes the SEC Championship and loses it. We said the committee should put the Tide behind Notre Dame if they wanted that option. To be clear, I don’t think anyone would say that part out loud in the committee room. I don’t think there’s any direct discussion of shaping rankings in order to make future rankings easier. But I do wonder if that goes through committee members’ minds.
In a scenario where the final spot comes down to 10–3 Alabama vs. 11–2 Texas Tech vs. 10–2 Miami, the committee has an easy way now to make whatever choice it wants. Prefer Miami over Alabama? Say that Miami’s head-to-head over Notre Dame is too much to ignore. Prefer Alabama over Miami? Keep things the same. Prefer to leave Texas Tech out in the event where they lose to BYU? Call the offense concerning. Intentionally or not, the committee gave itself options here. Even if Alabama does make the SEC Championship.
In our model’s simulations, Alabama misses the playoff 60% of the time when they make and lose the SEC Championship. But. In most simulations, our model expects that final loss to count. Limit the sample to simulations where teams can’t be punished for losing their conference championships, and the situation gets better for the Tide. Maybe that’s what happens. Maybe it isn’t. We only have one year of precedent there, so we didn’t make our model very confident. In most simulations, it splits the middle.
11. BYU (84.4)
Total FPA: –6.1
New FPA: –0.3
12. Utah (83.6)
Total FPA: –2.7
New FPA: –0.2
13. Miami (82.7)
Total FPA: +1.1
New FPA: +1.3
Credit to the committee. Our model expected Miami to jump Vanderbilt and USC to jump Georgia Tech. The résumés changed enough to justify both those moves. Usually, the committee doesn’t bother noticing that, which can create some unfortunate ripple effects (teams can get stuck even if they’re playing great football). Thankfully, the committee did seem to reevaluate both Miami and USC objectively in light of last week’s results.
Two notes on Miami and Notre Dame, since that keeps coming up:
First, hopefully these numbers help show how ridiculous it would be to put Miami ahead of Notre Dame. Without FPA, Notre Dame’s résumé is a full 6.5 points of ranking score better than Miami’s. Even with some negative FPA for Notre Dame and some positive FPA for Miami, the two are far apart. Moving them would require screwing one, undeservedly rewarding the other, and either screwing or undeservedly rewarding all the teams between them. You can’t put Miami ahead of Notre Dame without either putting Utah ahead of Notre Dame or Miami ahead of Utah.
Second, that was a three-point game. Miami was the home team. Home-field advantage is worth a few points. There are no ties, and that was not a tie, but it’s not like Miami smoked Notre Dame. Like it or not, margin of victory heavily factors into how the committee treats different teams. It’s mattering here, and I haven’t heard anyone mention it outside of saying things like, “Both Notre Dame’s losses were close.”
Again, the Miami–Notre Dame thing is disingenuous, unless the people yelling about it spent the same amount of time last week yelling about Miami and Louisville, who at the time had the same record as the Hurricanes and beat them head-to-head. If they did yell about that, more power to them. They’re being dumb, but at least that’s consistent.
14. Vanderbilt (81.9)
Total FPA: +2.9
New FPA: +0.4
15. USC (81.1)
Total FPA: –3.1
New FPA: +1.1
16. Georgia Tech (80.4)
Total FPA: +2.4
New FPA: +0.9
17. Texas (79.5)
Total FPA: +4.0
New FPA: 0.0
18. Michigan (78.7)
Total FPA: –0.6
New FPA: –0.1
No surprises in this section, but a couple modeling notes: In past seasons, we used to minimize how much FPA our model put in, meaning we assumed the committee was deviating from our model’s expectations by the smallest amount possible. This year, we changed FPA so that it lines teams up evenly within designated segments of the rankings curve. The goal here was to avoid situations like last year’s where Miami’s résumé kept marginally improving, our model kept expecting Miami to rise in the rankings, and Miami kept not rising in the rankings. We knew the committee wasn’t putting Miami into the playoff on the day of the selection show. Our model didn’t.
Don’t, then, read too much into Georgia Tech getting extra forgiveness for the near-loss to Boston College. That’s only there to keep them halfway between USC and Texas. Maybe it exists, maybe it doesn’t. Our model put it there just to keep the gaps constant and minimize late-season errors.
19. Virginia (77.0)
Total FPA: +1.6
New FPA: –0.5
20. Tennessee (76.0)
Total FPA: +4.2
New FPA: +0.5
21. Illinois (75.0)
Total FPA: –0.5
New FPA: +1.7
22. Missouri (74.1)
Total FPA: +2.4
New FPA: +2.4
23. Houston (73.1)
Total FPA: +2.1
New FPA: +2.1
24. Tulane (72.2)
Total FPA: +6.1
New FPA: +6.1
25. Arizona State (71.2)
Total FPA: +1.9
New FPA: +1.9
26? James Madison (70.2)
Total FPA: –6.6
New FPA: –3.9
27? North Texas (70.0)
Total FPA: –4.8
New FPA: –3.8
We don’t know that JMU and UNT are 26th and 27th, so we ended up lumping them close together. Lumped in right behind them are Washington, Louisville, Iowa, Arizona, and Cincinnati, each of whom also needed negative FPA in order for our model to rank Tulane. Tulane’s ranking, again, is crazy. Insane. Ludicrous. The most shocking thing I can remember the committee doing after seven years of running this model and plenty more spent following this stuff closely as only a fan. The best explanation I can find is that Tulane scheduled three power conference teams and went 2–1 against them. Those two wins were against Duke and Northwestern, and Tulane did lose to UTSA, but maybe the committee thought that deserved credit. Wild.
CFP Bracketology
How this all affected the bracket:
Ten Teams, Ten Spots?
There were two changes to our model’s bracket after last night’s rankings: Oklahoma moved in at Utah’s expense while Oregon moved back ahead of Notre Dame. Neither of these is a huge difference-maker. Our model ranks teams’ average finishes. Oregon’s is seventh-best, Notre Dame’s is eighth-best, Alabama’s is ninth-best, and Oklahoma’s is tenth-best. Those are the averages. If they all win out, it looks a little different.
Overall, we’re still looking at ten teams in ten spots, with two pure automatic bids and then a comfortable top ten holding down the space above them. At this point, the playoff race is a question of how many teams will finish the job and who those teams will be. Ohio State, Indiana, and Texas A&M are in. Georgia’s in if they don’t do something unthinkable in their last two or three games. Texas Tech’s in if they win the Big 12. Mississippi’s in if they beat Mississippi State. Alabama’s in if they win the SEC, and probably in if they just beat Auburn. Oklahoma’s almost definitely in if they beat both Missouri and LSU. Oregon’s in if they beat USC and Washington. Notre Dame’s most likely in if they beat both Syracuse and Stanford comfortably.
The probability of all those things happening, though, is weak. That’s why our model has all of Miami, BYU, and Utah so close to a 50/50 playoff shot. For BYU, there’s some automatic bid included in that number. BYU could still win the Big 12 themselves. But Utah and Miami are longshots to win their conference titles. More realistically, they’re the next teams up if Oklahoma, Alabama, or Notre Dame slips, or maybe even Mississippi or Oregon (especially if Oregon’s loss comes against Washington). We’re not assured of chaos. Miami and Utah aren’t up there with Notre Dame in playoff probability. But we’ll probably get at least an upset or two. Our model thinks there’s better than a 50/50 shot that enough goes down for those teams to get consideration.
Where’s Vanderbilt?
Currently ranked 14th, Vanderbilt seems like they should be a candidate for an at-large bid. Our model has them only 1% likely to pull it off, and only 3% likely even if they beat both Kentucky and Tennessee.
What’s happening here ties back to who could go down in front of the Commodores, and who’s likely to jump them. None of the thirteen teams in front of Vanderbilt are particularly likely to drop out of the top thirteen. Oregon could lose to USC, but they’d still have a good chance of lining up ahead of Vanderbilt. Teams behind Vanderbilt, meanwhile, have good opportunities to rise past them. USC could topple Oregon. Texas is currently a 1-point underdog against Texas A&M on Black Friday. Michigan and Georgia Tech each get a shot against Ohio State and Georgia.
Winning at Tennessee would be impressive, and our model admittedly might be undervaluing the upside of that win. But Vanderbilt’s best win right now came against Missouri, who’s barely ranked. Another win like that isn’t going to get Nashville’s finest past Alabama.
The ACC Is Finally Fine
After a couple weeks of flirting with disaster, we’re up to a 97% chance that the ACC champion makes the College Football Playoff. In simulations where they don’t, Miami gets an at-large bid 53% of the time. Add in the few simulations where, say, Georgia Tech beats Georgia but loses the ACC Championship, and we’re extremely unlikely to see zero ACC teams in this year’s College Football Playoff.
Good for them. The ACC is not a villainous conference. Dunderheaded at times, sure. But Tony Petitti does not lead the ACC.
Why JMU’s Still the Mid-Major Favorite
Our model still has JMU ahead of Tulane and North Texas in probability, though it no longer has JMU’s average finish placing them inside the top 25. What’s happening here is that Tulane—fifth-best in the American by Movelor power rating—isn’t very good at football, which means there’s a good chance that North Texas beats them in the AAC Championship if North Texas gets in. Right now, we’d have the Mean Green favored by 2.5 in that game, even in New Orleans. Would winning that get UNT into the CFP? Probably, but it’s hard to know. Much like JMU, the committee doesn’t seem to respect UNT.
In short: If Tulane wins the American, they’re most likely in. If North Texas wins the American, they’re probably in but it isn’t guaranteed. That opens the door for JMU, who’s almost definitely in if neither Tulane nor North Texas wins the American, provided JMU takes care of business.
Of the seven mid-majors with at least a 0.5% playoff chance, here’s how the conditional probabilities break down:
- Tulane: 88% playoff-likely if they win the American
- North Texas: 67% playoff-likely if they win the American
- James Madison: 42% playoff-likely if they win the Sun Belt
- Navy: 34% playoff-likely if they win the American
- USF: 33% playoff-likely if they win the American
- East Carolina: 15% playoff-likely if they win the American
- San Diego State: 5% playoff-likely if they win the Mountain West
Even if you assume that Tulane number should be 100%, and even if you assume North Texas will be ranked ahead of James Madison if they themselves take care of business, JMU’s number still doesn’t dip below 1-in-4 or 1-in-5.
**
