The penultimate College Football Rankings were handed down from on high last night, and things were learned. Let’s go through those things.
First off, let’s take a look at how our model interpreted the rankings. The full explanation of how our model works can be found here, but the big thing we’re about to look at is FPA, or Forgiveness/Punishment Adjustment, which is our model’s metric measuring how the committee’s rankings deviate from precedent. A positive FPA means the committee is giving a team more credit than precedent would suggest they’d receive. A negative FPA means the committee is giving a team less credit than precedent would suggest they’d receive. FPA accumulates and shifts over the course of the season as rankings are released, reflecting both what the committee seems to think this week and what the committee seems to have thought in all the weeks prior.
Here are this week’s FPA’s (the formatting is best viewed on a computer, though a sideways phone is better than a vertical one):
| Ranking | Team | Ranking w/o FPA | Est. Ranking Score | FPA | LW FPA | New FPA |
| 1 | Georgia | 1 | 100.0 | 0.0 | 0.0 | 0.0 |
| 2 | Michigan | 2 | 92.9 | 1.6 | 1.5 | 0.1 |
| 3 | Alabama | 5 | 92.2 | 2.5 | 1.9 | 0.6 |
| 4 | Cincinnati | 3 | 92.1 | 1.5 | 2.2 | -0.7 |
| 5 | Oklahoma State | 7 | 88.1 | 1.5 | 0.9 | 0.6 |
| 6 | Notre Dame | 4 | 88.0 | -1.9 | -1.3 | -0.6 |
| 7 | Ohio State | 6 | 86.7 | -0.3 | -0.3 | 0.0 |
| 8 | Mississippi | 15 | 82.2 | 3.8 | 3.8 | 0.0 |
| 9 | Baylor | 8 | 82.2 | 0.6 | 0.6 | 0.0 |
| 10 | Oregon | 19 | 80.7 | 3.6 | 3.6 | 0.0 |
| 11 | Michigan State | 9 | 80.6 | -0.4 | -0.4 | 0.0 |
| 12 | Brigham Young | 13 | 79.4 | 0.7 | 0.7 | 0.0 |
| 13 | Iowa | 12 | 77.9 | -0.9 | -1.9 | 1.0 |
| 14 | Oklahoma | 10 | 77.6 | -1.7 | -2.8 | 1.0 |
| 15 | Pittsburgh | 16 | 77.6 | -0.7 | 0.0 | -0.7 |
| 16 | Wake Forest | 11 | 77.4 | -1.6 | -1.0 | -0.6 |
| 17 | Utah | 21 | 76.7 | 0.4 | -1.3 | 1.7 |
| 18 | NC State | 25 | 76.2 | 2.2 | 1.2 | 1.0 |
| 19 | San Diego State | 14 | 76.0 | -2.5 | -0.4 | -2.0 |
| 20 | Clemson | 17 | 76.0 | -2.2 | -0.9 | -1.3 |
| 21 | Houston | 18 | 75.9 | -1.7 | -1.6 | -0.1 |
| 22 | Arkansas | 29 | 74.2 | 6.2 | 5.9 | 0.3 |
| 23 | Kentucky | 27 | 73.2 | 1.5 | 0.0 | 1.5 |
| 24 | Louisiana-Lafayette | 20 | 73.2 | -3.6 | -3.6 | 0.0 |
| 25 | Texas A&M | 30 | 73.1 | 5.2 | 5.1 | 0.1 |
| NR | UTSA | 22 | 71.7 | -4.4 | -9.3 | 4.9 |
| NR | Wisconsin | 23 | 71.7 | -4.1 | -1.5 | -2.6 |
| NR | Purdue | 24 | 71.6 | -2.6 | -1.5 | -1.2 |
| NR | Appalachian State | 26 | 71.2 | -2.6 | -0.4 | -2.2 |
Now, what we learned:
Cincinnati Might Not Be Safe
Were Cincinnati ahead of Alabama, we could prepare to lock them in should they win on Saturday. Instead, we need to wait. Our model does think Cincinnati will be ahead of both the Tide and Oklahoma State with a win and an Alabama loss, and by means of the data point of Cincinnati’s former lead over Michigan, it’s doing some triangulating to get to that projection, but we still don’t fully know what the committee thinks of these three teams and what it might think of various results this weekend. There’s a chance the traditional one-loss Power Five champion, Oklahoma State, ends up ahead of Cincinnati. If that’s the case, it means Cincinnati’s out with Alabama, Michigan, and Oklahoma State wins. There’s also the chance a two-loss SEC team trumps an undefeated Cincinnati. If that’s the case, it means Cincinnati’s out with Michigan and Oklahoma State wins.
Notre Dame Is Behind a One-Loss Oklahoma State
We thought this was the case, but it’s confirmed: If Oklahoma State wins, they’re ahead of the Irish. It’s not that you can’t fall with a win, but there’s no way a winning data point hurts the Cowboys in this context, unless something bizarre happens like Mike Gundy jumping ship for LSU, at which point…
In all seriousness, though, I think the only way Notre Dame could pass an Oklahoma State which wins is if the game’s as ugly as possible and Oklahoma State has a ton of players get hurt. Even then, it doesn’t seem like it’ll happen, but we’ve never seen this scenario play out, so we technically don’t know for sure.
Notre Dame Is Ahead of Ohio State
This isn’t the absolute floor on who can make the playoff, but Notre Dame staying ahead of the Buckeyes confirms that Notre Dame’s next in line behind the five who can seemingly play their way in (including the one who already has).
Baylor and Iowa Have Cause to Recklessly Hope
For Baylor, it’s because they’re the top two-loss Power Five conference championship participant. Our model doesn’t see a route for them to make the field, but our model could be wrong, and if the committee decides to resort to two-loss Power Five champions, should the need arise, Baylor stands to be the first one called.
For Iowa, it’s because they have the highest-ranked opponent of the two-loss Power Five conference championship participants. As with Baylor (and Oregon, and Pitt, and Wake Forest), our model doesn’t see a path for the Hawkeyes to make the field, but again, our model could be wrong, and we don’t definitively know how much beating Michigan would move the needle. We tested all the possible two-loss Power Five champions winning by 40 points in the context of other games going as expected, and none of them jumped Ohio State, but it’s possible there’s some combination we didn’t test or, again, that our model could be wrong.
***
That’s what we learned. Here’s what else we know.
We said above that our model thinks Cincinnati will stay ahead of Oklahoma State with a win and leapfrog Alabama with a win and an Alabama loss. Here’s what exactly our model thinks of that situation, having plugged in each game going roughly according to the current Vegas spread:
| Expected Ranking | Team | Est. Ranking Score |
| 1 | Georgia | 100.0 |
| 2 | Michigan | 93.0 |
| 3 | Cincinnati | 90.6 |
| 4 | Oklahoma State | 87.4 |
| 5 | Alabama | 86.4 |
| 6 | Notre Dame | 84.7 |
And here’s what the model thinks if we take away FPA:
| Expected Ranking | Team | Est. Ranking Score |
| 1 | Georgia | 100.0 |
| 2 | Michigan | 91.4 |
| 3 | Cincinnati | 89.2 |
| 4 | Notre Dame | 86.6 |
| 5 | Oklahoma State | 86.0 |
| 6 | Alabama | 83.9 |
We’ve talked before about our model being a perfect 28-for-28 in back testing both with and without FPA when it comes to predicting playoff qualifiers. Clearly, we’re looking at a situation where that might change. If the games go according to script and Alabama makes the field, we’ll probably have to expand our model’s foci to include a team’s four best wins, rather than just their three best, or to incorporate a forward-looking, predictive metric to reflect who the committee thinks are the “best” teams. If the games go according to script and Oklahoma State makes the field, the first place we’ll look will be the weight of a Power Five conference championship, which might be higher than we’ve realized.
The games, though, might not go according to script. According to SP+, there’s an 80% that even just the four most important games—the Big Ten, Big 12, AAC, and SEC Championships—don’t all see the favorite win. Let’s walk through some scenarios.
Here’s what the model expects if all games go according to script except Iowa beats Michigan by a field goal:
| Expected Ranking | Team | Est. Ranking Score |
| 1 | Georgia | 100.0 |
| 2 | Cincinnati | 90.6 |
| 3 | Oklahoma State | 87.4 |
| 4 | Alabama | 86.4 |
| 5 | Notre Dame | 84.7 |
| 6 | Michigan | 84.6 |
Here’s what the model expects if all games go according to script except Houston beats Cincinnati by a field goal:
| Expected Ranking | Team | Est. Ranking Score |
| 1 | Georgia | 100.0 |
| 2 | Michigan | 93.0 |
| 3 | Oklahoma State | 87.4 |
| 4 | Alabama | 86.4 |
| 5 | Notre Dame | 84.3 |
| 6 | Cincinnati | 83.7 |
Here’s what the model expects if all games go according to script except Alabama beats Georgia by a field goal:
| Expected Ranking | Team | Est. Ranking Score |
| 1 | Alabama | 100.0 |
| 2 | Michigan | 99.7 |
| 3 | Georgia | 99.0 |
| 4 | Cincinnati | 96.3 |
| 5 | Oklahoma State | 93.6 |
| 6 | Notre Dame | 90.7 |
As we said above: If Baylor beats Oklahoma State, our model still expects Ohio State to be ahead of the Bears.
What does this all tell us?
Michigan or Ohio State Has the Second-Best Two-Loss Case
This could be important. I’m not extraordinarily confident in this, because our model was built off of much simpler situations, but in the three percent of cases in which Michigan, Oklahoma State, and Cincinnati all lose, the committee will be forced to expand their circle of consideration beyond those three teams, Georgia, Alabama, and Notre Dame. Our model, looking at this all, thinks Michigan would be the team chosen, so long as the loss comes by five points or fewer. If we expand Iowa’s win to a six-point victory, in the context of all other games going as expected, Ohio State moves ahead of the Wolverines. This “in the context of all other games going as expected” thing is key: Other games being played will impact our model’s perception. We’re not guaranteeing that in this fringe case in which Baylor wins and Houston wins and Iowa wins by six, you’ll see Ohio State in the playoff. But that’s where it tips, and that’s who it tips to, and until it tips to Ohio State, it’s Michigan’s spot in that instance.
Cincinnati Should Stay Ahead of Oklahoma State
Our model does think the Bearcats best the Pokes in scenarios in which each team wins. This is massive for Cincinnati.
Cincinnati Should Pass an 11-2 Alabama
Our model also does think the Bearcats should best the Tide in scenarios in which Cincinnati wins and Alabama loses. This, too, is massive for Cincinnati.
Notre Dame Can’t Rely on Alabama Losing
Our model does not think the Fighting Irish best the Tide in scenarios in which Alabama loses. This is, of course, quite bad for Notre Dame.
A two-loss Alabama, losing by a touchdown, is still ranked ahead of Notre Dame in our model’s expectations, and by a decent enough margin. Flip it up to a 40-point loss, and Alabama is still ahead of Notre Dame.
I do have some questions about this. I don’t know exactly how the voting order goes—whether the committee will decide the top four, then place them in order; or if they’ll simply decide the top four in order—but I doubt the committee would want to make Alabama play a rematch with Georgia in the semifinal, regardless of the final score but especially if it’s a blowout. The problem for Notre Dame (and Oklahoma State, and Cincinnati) is that the committee’s solution to this might not be to push Alabama out of the field. It might be to keep them ahead of the Cowboys and the Bearcats, and by extension, the Irish.
I’d imagine that out of everything our model’s showing, besides that three-percent scenario in which we’re looking at 11-2 Michigan and 10-2 Ohio State and maybe 11-2 Baylor and Iowa, this is the most controversial piece, especially because we have a decent share of Notre Dame readers. Let’s give it a closer look:
The Losses
Alabama would have a loss to Georgia, which would set them back in our model’s eyes, but would be tied with Clemson as the least damaging loss of the season were they to keep the margin to a touchdown, and wouldn’t be extremely damaging in our model in any scenario. Alabama would also have that three-point loss to Texas A&M, which—importantly—came in College Station, not Tuscaloosa. Notre Dame has an eleven-point loss to Cincinnati, at home. Cincinnati is better than Texas A&M, but not by leaps and bounds. The spread in a hypothetical matchup between the two would be less than a touchdown, if not less than a field goal. Notre Dame’s loss to Cincinnati is arguably worse than Alabama’s loss to Texas A&M. Our model views the two as roughly equivalent.
The Wins
The model is still impressed by Notre Dame’s margin of victory against Wisconsin, especially because it believes the committee would have Wisconsin ranked right now were they following precedent to a tee. It also believes Purdue would be ranked, and Notre Dame beat Purdue by two scores, albeit at home. Beyond that, though, Notre Dame has nothing. The next-best win came at Virginia, and while it was decisive, Virginia’s an ACC team who went 6-6. Our model has them 53rd in hypothetical 130-team rankings.
Alabama, meanwhile, has beaten Mississippi by 21 at home, Mississippi State by 40 on the road, and Miami by 31 on a neutral field. They’ve also beaten Tennessee by 28 at home, which would be their fourth-best victory in our model’s eyes. Virginia, you’ll recall, is 53rd in our model’s lineup. Mississippi State, Miami, and Tennessee are 32nd, 35th, and 36th respectively, and Mississippi State and Tennessee are probably actually higher, because the SEC is receiving uniformly positive FPA this year (as we’ve discussed before, this is probably a fair adjustment that reflects strength of schedule in a way our model can miss at times). The losses might tilt ever-so-slightly in Notre Dame’s favor, but the wins are decidedly supportive of this hypothetical Alabama.
On-Field Performance
Our model actually views the Irish as the more impressive on-field team, using its adjusted point differential metric. This is, however, outweighted by FPA, and in the real world, Alabama would be favored against Notre Dame, which the committee undoubtedly knows.
It’s easy to get hung up on the 11-2 vs. 11-1 thing, and it’s fair to not want an Alabama/Georgia rematch, but Notre Dame does not have a single ranked win, or a single top-twenty win if you love Wisconsin and Purdue, and they muffed a wonderful chance to prove themselves against Cincinnati at home. Independence absolutely works for the Irish, especially in terms of program prestige. (You want to become Penn State? Join the Big Ten.) In terms of playoff possibility, though, it might not work this year. This isn’t one of those stupid Join a conference! arguments. It’s just an acknowledgment that while it works in a lot of areas, a scenario in which Notre Dame goes 11-1 against a schedule lacking much weight towards the top is a decidedly bad one for independent Notre Dame.
What happens, then? Let’s go through the sixteen ways the four biggest games can play out, using SP+’s probabilities (Bill Connelly has noted that SP+ does not factor injuries into account, so it probably underestimates Oklahoma State’s chances against a hobbled or inactive Gerry Bohanon, meaning these likely overstate the probability of a Cowboy loss, ever so slightly).
1. Georgia, Michigan, Cincinnati, and Oklahoma State Win – 20% Likely
These are the four you should expect in the playoff, in this order, with a small chance of Alabama but man, that would be surprising.
2. Georgia, Michigan, Cincinnati, and Baylor Win – 17% Likely
Our model expects Alabama to be the fourth team in this scenario, though I personally think they might be ranked third, ahead of Cincinnati. Notre Dame has a chance here, but it’s not a good one. Our model does not think Baylor has a chance.
3. Alabama, Michigan, Cincinnati, and Oklahoma State Win – 10% Likely
Alabama jumps to the top spot here, followed by Michigan, Georgia, and Cincinnati. This is according to our model. It’s possible Oklahoma State would edge the Bearcats. This is a nightmare scenario for Cincinnati for that reason.
4. Alabama, Michigan, Cincinnati, and Baylor Win – 9% Likely
Same story as above, except Oklahoma State is no longer a threat. This isn’t a dream scenario for Cincy—they’d probably rather play Michigan than one of the SEC teams—but it’s a good one for their security, as are any in which Oklahoma State and/or Michigan lose.
5. Georgia, Iowa, Cincinnati, and Oklahoma State Win – 7% Likely
The Big Ten is out in this scenario, with us back to the question of whether Alabama would be the fourth team in, or if they’d be the third team in, or if they’d be the second team in, or if they’d fall behind Notre Dame and land in fifth.
6. Georgia, Michigan, Houston, and Oklahoma State Win – 6% Likely
Again, we’re in Oklahoma State vs. Alabama vs. Notre Dame territory, but this time, Cincinnati’s out of the mix.
7. Georgia, Iowa, Cincinnati, and Baylor Win – 6% Likely
This is straightforward for the committee, who we assume would let Notre Dame in to play Georgia while Alabama and Cincinnati go to the other semifinal.
8. Georgia, Michigan, Houston, and Baylor Win – 5% Likely
Again, Notre Dame’s in, with Alabama and Michigan in the second semifinal this time.
9. Alabama, Iowa, Cincinnati, and Oklahoma State Win – 4% Likely
Alabama takes over the top spot, Georgia likely stays in the top two, and then it’s Cincinnati and Oklahoma State in some order.
10. Alabama, Michigan, Houston, and Oklahoma State Win – 3% Likely
Straightforward again. Alabama, Michigan, Georgia, Oklahoma State.
11. Alabama, Iowa, Cincinnati, and Baylor Win – 3% Likely
Notre Dame is in every scenario from here on out, illustrating that the Irish, as you’d expect, are the beneficiaries of improbable scenarios should they occur. One note here, since we haven’t addressed it yet: I really doubt the committee would demote the Irish past anyone currently behind them on the basis of Brian Kelly having left. That would really, really surprise me. That’s just my opinion, but for whatever it’s worth.
In this scenario, it’d be Alabama/Georgia/Cincinnati/Notre Dame.
12. Alabama, Michigan, Houston, and Baylor Win – 2% Likely
Alabama/Michigan/Georgia/Notre Dame.
13. Georgia, Iowa, Houston, and Oklahoma State Win – 2% Likely
Here, we get Oklahoma State up to second overall, the only scenario where that happens. They play Alabama, Georgia plays Notre Dame.
14. Georgia, Iowa, Houston, and Baylor Win – 2% Likely
This is where we finally have to look hard for the fourth. Our model thinks it would be Michigan or Ohio State, but it’s really uncharted. Georgia, Alabama, and Notre Dame are the top three.
15. Alabama, Iowa, Houston, and Oklahoma State Win – 1% Likely
Alabama plays Notre Dame, Georgia plays Oklahoma State.
16. Alabama, Iowa, Houston, and Baylor Win – 1% Likely
Again, we’re looking hard for the fourth in this scenario, but our model thinks it’s Michigan or Ohio State. Alabama, Georgia, and Notre Dame are the top three.
***
Overall, this leaves us with the following playoff probabilities:
Georgia: 100%. In. If there is a playoff this year, Georgia will be in it.
Alabama: Somewhere a little bit below 80%. Win and in. See one of Michigan/Oklahoma State/Cincinnati lose, probably in. It’s possible they’ll drop behind Notre Dame, but it’s not likely.
Michigan: Somewhere just above 75%. Win and in. Get total chaos, maybe in.
Cincinnati: Somewhere just below 76%. Win and probably in. It’s possible Alabama and Oklahoma State would both edge them, but it’s not likely.
Oklahoma State: A little above 44%. Win and probably in. Just need one of Alabama, Cincinnati, and Michigan to lose.
Notre Dame: Around 24%. Need two or three of Michigan, Oklahoma State, and Cincinnati to lose, though there could be a path for them to pass Alabama that our model doesn’t see.
Ohio State: Roughly 1%. Need all three of Michigan, Cincinnati, and Oklahoma State to lose, and probably need Michigan to lose convincingly, and even then might need some luck.
Baylor: Less than 1%. Maybe our model’s missing this, but it doesn’t seem them passing the Buckeyes, let alone the Irish.
Iowa: Less than 1%. Again, this would be shocking.
Oregon: Less than 1%. I don’t know what the justification would be for this, but I don’t know what the justification is for having them in the top ten after those particular losses.
After those ten…nobody. There is no way for Pitt or Wake Forest or Mississippi to make the playoff that I or our model can fathom. These are the only ten. Really, it’s probably only six or seven.
***
Come Friday, we’ll have the preview post, in which I think we’ll go through the weekend chronologically and say what every team’s fans should be pulling for in every game. Including potentially that Cal/USC makeup, where Notre Dame and Oregon will be hoping for Lincoln Riley to debut with a 1,000-point beatdown of the Hippies.
Until then, happy fuming, Notre Dame fans, and happy hoping to the rest of you.
