What Happens if Alabama Beats Georgia? The Doomsday Scenario for the College Football Playoff

It’s possible—likely, even—that Georgia beats Alabama on Saturday and renders all of this moot. If Georgia wins, there’s no controversy. Georgia makes the playoff, the Pac-12 champion makes the playoff, and for the final two spots the priority list goes something like: 13–0 Michigan; 13–0 Florida State; 12–1 Texas (we think); 12–1 Michigan (maybe); 11–1 Ohio State; 12–1 Washington (we think). It would be easy for the committee to insert 13–0 power conference teams. It would be easy for the committee to eliminate hypothetical 11–2 Oregon. It would be easy for the committee to keep hypothetical 12–1 Oregon ahead of Texas, whose signature win in Tuscaloosa would now trail Oregon’s in the Washington rematch (with the rest of the Longhorn résumé also worse across virtually every other meaningful category). Disagreement could arise with an Iowa upset of Michigan, or with losses by both Florida State and Texas, but the disagreement would be over which traditionally undeserving teams to include, not which traditionally deserving team to exclude. If Georgia beats Alabama, the committee’s job is simple.

If Alabama beats Georgia, the committee’s job is going to be hard, barring a whole lot of help. Alabama will be a 12–1 SEC champion with a head-to-head win over Georgia. Georgia will be 12–1, their only loss to Alabama. Texas may be a 12–1 Big 12 champion with a worse loss than Georgia’s or Alabama’s but a decisive head-to-head win over the Tide in Tuscaloosa. Oregon may be a 12–1 Pac-12 champion with a better loss than Texas’s and a rematch win over Washington, the team who beat them. If Oregon isn’t 12–1, Washington will be 13–0. Florida State could be 13–0. Michigan could conceivably be 12–1 with a loss comparable in quality to Texas’s and wins over Ohio State and Penn State. The committee will have to do at least one of the following things:

1. Exclude the SEC from the playoff entirely.
2. Exclude a 13–0 Power Five champion from the playoff.
3. Include a 12–1 team (Georgia or Alabama) ahead of a different 12–1 team who beat them head-to-head (Alabama or Texas).
4. Exclude a 12–1 Oregon in favor of one or more teams the committee has consistently said are worse than Oregon.

Complicating matters further, the committee’s nominal mandate is not to choose the most deserving teams. It’s to choose the best teams. If Alabama beats Georgia, the committee’s job is going to be hard, barring a whole lot of help.

We have our college football model, which simulates this process (and has a great track record—it’s never missed a playoff team), and we’ll get to our model. But our model is trained on precedent. This hypothetical is not one for which we have precedent. So, first, we’re going to talk about this from a human lens, because the decision is going to be made by humans.

The core of the potential disagreement here is Texas vs. Alabama, and it’s whether one head-to-head result should matter more than an entire season’s worth of games. If both Alabama and Texas were to win on Saturday and you were to then line up their résumés head-to-head, the comparison would look something like this:

Texas has a better best win than Alabama in this scenario. Alabama has a better loss, and it has better second through fourth-best wins. Neither has been playing spectacularly well of late, on the aggregate, but Texas played their last good game more recently than Alabama played theirs. Alabama won the better conference (and Texas fans have little argument against this, on their way to the SEC themselves). Alabama would be something like a 3-point favorite in a hypothetical neutral-field rematch.

Overall, Alabama has the better body of work. Alabama is more accomplished, in addition to being a better team. Alabama is more deserving than Texas of a chance at a national championship, and more likely to win if they do make the playoff. But Texas won in Tuscaloosa, and there’s a time-honored tradition in sports—especially American sports—in which we decide that one game is going to trump whatever happened over the entire rest of the season. That’s how playoffs work. That’s what consistently produces the dramatic moments so many sports fans love. The difference here is that instead of this single-elimination game happening at the end of the season, like Game 7 of the World Series or Butler vs. Duke, the decisive game in this instance will have come a long time ago, in the second week of the season, back in September.

The arguments, then?

• Alabama: Better team; Better résumé.
• Texas: Beat Alabama in their own stadium.

It’s pretty hard to argue against Texas.

The bright spot here—and this has always been a bright spot with the four-team playoff—is that this is a mess entirely of Alabama’s and Texas’s making. If Alabama wanted to spare themselves a scenario in which one game outweighed their other twelve, they should have figured out an offense that worked for Jalen Milroe in April. If Texas wanted to avoid an outcome in which a fellow 12–1 team they beat head-to-head made the playoff ahead of them, they shouldn’t have turned the ball over on both their opening possessions against Oklahoma and then allowed the Sooners to waltz down the field on their game-clinching one-minute drill. The committee isn’t going to screw anybody over on Sunday. If the controversy happens, it’s going to happen because neither of these teams did what they had to do to win a national championship.

There’s a debate going on over whether this would be a good season or a bad season for the impending 12-team playoff. My personal view on the matter is well-established and unrealistic, and it’s that the playoff should be of variable size and only include teams with a legitimate claim towards a national championship. This year, that would be 13–0 Power Five champions, assuming at least one exists come Saturday night (sorry, Liberty, but every opponent has its price, and you chose MAC teams and UMass to fill up your nonconference rather than shelling out to compete against quality). I do think that the stakes of this week’s games would be lesser if they were mostly deciding who gets playoff byes, who gets playoff home games, and whether any wildcards sneak in. But I also think that the first round matchups would be strong, with potential games between Texas/Tulane, Ohio State/Mississippi, Washington/Penn State, and Alabama/Mizzou; and that the second round would also be strong, with potential games between Georgia/Alabama (Round 2!), Florida State/Ohio State, Oregon/Texas, and Michigan/Penn State (ok, that might actually be too much Penn State for Penn State’s own sanity). We’re going to get a larger quantity of marquee games thanks to the 12-team playoff, but we’re going to reduce the stakes of the other games, sometimes by a little and sometimes by a lot. It’ll be like drinking more Coca-Cola but never getting to drink it from McDonald’s.

Overall, the four-team playoff has served us well, and Georgia vs. Alabama is going to be more exciting on Saturday because of how much it affects not just those two teams, but everyone in the playoff picture, and therefore every college football fan in the country. Georgia vs. Alabama could be a de facto national title game. It could be an SEC kamikaze mission. It’s going to be something.

Now, let’s get to our model.

We wrote a lot on Monday about who controls their fate, in the eyes of our model. Since then, our only new data point is that the committee takes a dimmer view of Ohio State than precedent would suggest. If you want to think conspiratorial, you could say that the committee intentionally put Ohio State between Oregon and Texas to signal that there’s a gap, and that they’re not too high on either Texas or Alabama. More likely, that gap just existed in the committee’s eyes already and Ohio State filled it. The bottom line, for our model, is that Ohio State’s playoff probability was roughly halved by the committee ranking the Buckeyes where they did. Most committees, historically, would have ranked a team 3^rd with a current résumé like Ohio State’s. This committee is unusually deferential to overall win-loss record, though, and it has been since its first rankings were released. Previous committees seemed to want to show the world how smart they were, how willing they were to reconsider positions upon the presentation of new evidence, how unlike the regrettably thoughtless AP Poll their first-of-their-kind rankings would be. The current committee seems to want to avoid death threats to whatever extent possible. Texas and Alabama are making that difficult.

The bottom line is that the committee’s most recent rankings took Ohio State from a 1-in-5 chance to a 1-in-10 chance, in the eyes of our model. There’s something, however, which we need to address about our model.

It doesn’t know what the committee thinks about Georgia.

Early in the season, Georgia was playing mediocre football (by title contender standards). They were struggling to beat South Carolina, and struggling to beat Auburn, and even in games like the one against UAB they won handily but didn’t win like a great team would win. Still, they were always ranked ahead of Michigan.

There were explanations as to why. When the first rankings came out, Georgia had just bullied Florida and Connor Stalions was becoming a household name. When the first rankings came out, Michigan’s signature victory was over Rutgers. Subconsciously, the committee was likely still impacted by the memory of J.J. McCarthy throwing multiple touchdowns to TCU defenders one week before Georgia treated TCU like they were Cornell. To our model, though, this was surprising. By the rubric used by every previous committee, Michigan should have been ahead of Georgia. They were winning more impressively, and Georgia’s victories were nothing to write home about either.

So, our model reacted, effectively saying, “This committee loves Georgia, and this committee hates Michigan.” It filed that away, keeping it in mind over the rest of the season.

This happens every week with our model. A team is ranked somewhere surprising and our model reacts. Oklahoma State? This committee views them like a favorite child. Notre Dame? Don’t bring that name up to this committee again. The variable we use to measure this forgiveness or punishment is called FPA, which stands for “Forgiveness/Punishment Adjustment.” Here’s how much of it each team has, and how much each was handed this week:

Usually, FPA wavers over the course of the year. Big wins and big losses have a way to convince committees to reconsider different teams, and reconsideration lands them somewhere new. Oregon State was once a favored son, but this last loss dropped them more in line with traditional measurement. Kansas State was once an outcast, but the committee relented this week.

With Georgia and Michigan, though, there was never any reckoning. When Georgia beat Mississippi, our model believed Georgia would be the new #1 team, and our model was correct, so it didn’t change Georgia’s FPA at all. It assumed the committee still viewed Georgia’s first eight games with the same kind eye it had turned on them weeks earlier. We don’t know if this is correct. We theorized then that the win over Mississippi might have caught Georgia up to the committee’s perception, erasing the need for the previous FPA. We still wonder about that theory.

One month later, our model still has Georgia with the massive FPA it enjoyed when the rankings first came out, and as a consequence, it thinks it would be very hard for Georgia to fall back beyond fourth place in the rankings, even with a loss to Alabama. In simulations in which Georgia loses to Alabama, our model still has the Dawgs in the playoff 79% of the time.

We looked into this, removing Georgia’s FPA but leaving Michigan’s negative FPA in place, something which left Georgia still currently ranked 1^st. The results were striking. In these simulations, an Alabama upset left Georgia making the playoff only 8% of the time. Naturally, this reduction in probability had to correspond with increases elsewhere. Here’s where it went:

For some teams, Georgia isn’t a factor. Michigan faces a similar situation regardless of the committee’s opinion of the Dawgs. Oregon, Florida State, and Washington are in the same boat. For Texas and Alabama, though, and for Ohio State to an extent, Georgia is a major player. When we leave Georgia’s FPA in our model and ask it to decide between 12–1 Texas and 12–1 Alabama, it says, “Why? They’re ranked fifth and sixth unless someone else goes down.” When we take Georgia’s FPA out of our model and ask the same question, the model starts overheating and flipping coins, trying to unbalance a painfully balanced scale.

You may notice that Georgia is less than 50% likely to make the field in our model’s simulations without its FPA, and you may know that Georgia is favored over Alabama, and you may wonder what the hell is going on. It’s not that our model thinks Georgia has any chance of missing the playoff with a win. In all our model’s simulations in which Georgia wins on Saturday, leaving FPA or removing it, Georgia makes the playoff. What our model thinks is that Alabama should be favored over Georgia. Relative to other ratings systems, this isn’t all that outlandish of a view. Betting markets are much more confident in Georgia than ratings systems. They’re more accurate than ratings systems, so we should trust them, but that’s what’s going on. The important takeaway here is that Georgia might be 79% likely to make the playoff with a loss, and it might be 8% likely, and where it falls between those poles depends on what the committee really thinks of this team, a team who’s spent preciously little time under the committee’s microscope.

We’ll include scenarios in Friday’s post, with a few different ways of calculating them, and we’ll do the same on Saturday morning after Oregon and Washington play Friday night. We’ll also have a post up tonight or tomorrow about the Group of Five picture, with some bowl talk thrown in. At a high level, though, the playoff message, from our model and from our minds, is the following:

• Georgia: In with a win, needs help with a loss.
• Michigan: In with a win, needs help with a loss.
• Washington: 95% in with a win, 99% out with a loss.
• Florida State: 95% in with a win, out with a loss.
• Oregon: 95% in with a win, out with a loss.
• Ohio State: Needs help (specifically: Florida State, Alabama, probably Texas, and maybe Oregon to lose), but might get it (10% chance?).
• Texas: Needs to win, needs one or more of Florida State, Michigan, and Georgia to lose, might need to win an argument over Alabama.
• Alabama: Needs to win, needs to either get some help or win an argument over Texas.

Four teams remain undefeated. The stakes are high in all five power conference championships. Six of college football’s twelve marquee programs are involved in the discussion, with two fun outsiders from the Pac-12 thrown in. It’s hard to ask for more than that from the weekend ahead.