Who Controls Their College Football Playoff Fate

Three main questions preceded last night’s update to the College Football Playoff rankings:

1. Would Washington jump Florida State?
2. Would Louisville jump Missouri?
3. What would we see of the Group of Five picture?

The answers:

1. Washington did jump Florida State, with committee chair (and NC State athletic director) Boo Corrigan stressing that this was about what each has accomplished through this season so far, not about any inference from the committee on how good FSU will be without Jordan Travis.

2. Louisville did not jump Missouri. The Cardinals lack any realistic path to the playoff, though they can still play spoiler and the Orange Bowl remains on the table.

3. Tulane still leads, but Liberty is ahead of the rest of the pack right now. This doesn’t mean another team can’t jump Liberty—we didn’t see how close SMU, Toledo, UNLV, and Troy are or aren’t—but we finally learned where the committee has Liberty in the pecking order.

We’re going to get into a lot more detail down below, but if you’re curious what each team needs to make the playoff, the most likely case is the following. Few are guaranteed anything, but as of right now, it would be surprising if the following didn’t get the job done.

• Georgia: Win out or get help from others losing.
• Ohio State: Win out.
• Michigan: Win out.
• Washington: Win out.
• Florida State: Win out.
• Oregon: Win out.
• Texas: Win out and/or get help from Alabama, Washington, and/or Florida State losing, with some possible benefit from Michigan losing as well.
• Alabama: Win out and/or get help from Texas, Washington, and/or Florida State losing, with some possible benefit from Michigan losing as well.
• Louisville: Invent a time machine and wipe the University of Pittsburgh from the face of the earth.

Our college football model has a variable named FPA (Forgiveness/Punishment Adjustment) which measures how far from precedent the committee is deviating with regards to each team. Here are those numbers for this week’s rankings, scaled such that a ranking score of 100.0 is that of the top-ranked team in the FBS and a score of 0.0 is that of the worst-ranked team in the FBS, if these rankings went all the way to 133.

One important note is that we don’t know that Toledo is right on Liberty’s heels. We know that 26^th is the highest they can be, but they don’t show up on this list because we know or even think that they’re 26^th. They’re simply the only unranked team currently receiving any FPA. Our real guess is that Toledo is receiving FPA similar to that of Liberty, with Conference USA and the MAC both far below the AAC, where Tulane is receiving no FPA anymore and is therefore being treated perfectly in line with precedent. This Tulane piece is illustrative. The committee gives different discounts to different Group of Five conferences, but because such a high percentage of good Group of Five teams in the playoff era have come from the AAC, most Group of Five precedent—precedent is the foundation of our model—comes from how the committee treats the AAC. Justifiably, the committee doesn’t treat Liberty like they’re playing an AAC schedule, and that probably goes for Toledo as well. We need to adjust this over the offseason. Helpfully, Liberty has given us a lot more data to work with as we try to shape the model in this way.

There’s consequence here with the New Year’s Six picture—Missouri is now likelier to hold their ranking than Mississippi is to hold its own, making Missouri now the likely at-large bid (full updated bowl projections are included below, and can always be found here)—but the bigger question is the one ESPN loves to ask:

What happens if these teams all keep winning?

Our model is not infallible. It has a perfect track record on predicting playoff teams, but some of that is luck. It hasn’t been hard to figure out who the four teams will be each year, for human. computer, or—importantly—committee. This year might be harder. We don’t know that it will be—the doomsday scenario where Florida State, the Ohio State/Michigan winner, Texas, and Alabama all win out (leaving only two or three clear playoff teams and a four or five-team mess for the final one or two spots) is only 13% likely—but the probability of this all working itself out is getting lower. If we take a big shortcut and say that Louisville is out, that Ohio State and Michigan are each out with a loss (this is the shortcut piece), and that Washington and Florida State aren’t necessarily in if they win out (more on this in a moment), there’s only a 55% chance of this working out easily for the committee, per our model’s simulations, meaning in only 55% of simulations are there four or fewer strong contenders for the four spots. In the other 45%, the committee has to make someone very, very mad.

What would the committee do in those 45% of simulations? We aren’t sure. We personally doubt the committee would leave a 13–0 Florida State or Washington out of the playoff (these committees don’t tend to do things this audacious), and it’s hard for our model to know how much of a lead Georgia’s 11–0 has over that of Ohio State’s. Georgia’s FPA, in our model’s eyes, is still what it was before the Mississippi game, meaning: The committee was giving Georgia a lot of love before they beat up Mississippi and Tennessee, and our model doesn’t know whether those results confirmed the committee’s priors, catching Georgia’s résumé up to where the committee had them (this would erase much or all of Georgia’s FPA, pulling Georgia back closer to Ohio State) or whether those results further impressed the committee (the model errs on this side, rightly or wrongly). With that caveat, we checked our model’s simulations to find each team’s playoff probability if they win out, and this is what we found:

What to make of this, most simply, is that Alabama and Texas are the teams in danger of finding themselves on the bubble. Oregon is more than 90% likely to make it at 12–1. Florida State and Washington are almost certain to make it at 13–0. Ohio State and Michigan are locks at 13–0. Georgia is a lock at 13–0 and has a chance at 12–1. It’s a coin flip whether this works itself out or not, but if it doesn’t, the likeliest outcome is that the committee has to choose between Texas’s head-to-head win or Alabama’s overall body of work and on-paper ability (Alabama would be about a 10-point favorite right now in a neutral-site rematch). Conference titles count for enough that Oregon can play its way in. Undefeated Power Five champions are unthinkable to leave out. Alabama and Texas? They need help, and the committee is probably hoping they get that help, because if this comes down to Alabama and Texas, the committee is between a rock and a hard place.

What help do Alabama and Texas need? Well, here’s how much their playoff probabilities increase and decrease based on the presence of the other playoff teams (Louisville excluded due to small sample):

In short, if Oregon makes the playoff, Alabama is 12% more likely to have also made the playoff. If Michigan makes the playoff, Texas is 33% less likely to have also made it. According to our model. In reality, Michigan and Ohio State are probably closer to one another than our model implies, but the overall message is likely true: Alabama and Texas want Washington and Florida State to lose. They prefer Michigan losing to Ohio State losing. And, although this isn’t shown in the table above, they each want each other to lose. The days of their fates being linked have passed. It no longer helps either to have the other in the picture. Alabama is 25% less likely to make the playoff if Texas has made it, and Texas is 25% less likely if Alabama’s within the four. Neither wants the other to lose as badly as they want Washington and Florida State to lose, but each wants the other to lose.

There are some dormant volcanoes which could really change things, and there’s a decent probability one of those erupts. The probability of at least one of Georgia (at Georgia Tech), Washington (vs. Washington State), Florida State (at Florida), Texas (vs. Texas Tech), Oregon (vs. Oregon State), or Alabama (at Auburn) losing this weekend is close to 40%, or 2-in-5. The corollary to that, though, is that the 55% of scenarios in which this all works out smoothly rely, in a lot of cases, on one or more of those upsets happening. If this weekend goes chalkily, the probability of a committee headache is going to eclipse 50%, and by a lot.

*

*

*

*

Bowl projections; updated ratings and probabilities (for archival purposes):