The first College Football Playoff Rankings of 2019 were released last night, giving our model an important look into the thinking of the selection committee.
We built the model such that it could be easily adjusted in response to each round of rankings. While this process isn’t automated, it’s a simple fix: we have a variable in the model that starts at zero for each team and, after each round of rankings is released, is either boosted or dropped until the to-date rankings our model pumps out match those produced by the committee. The goal of this variable is to capture the differences between how the committee is perceiving a team and its résumé thus far and how the average prior committee would have perceived the same team and its résumé (the model is built on data from the five prior years of the College Football Playoff). We’ll adjust this variable each week when rankings are released, but this week’s adjustments will most likely produce the biggest shuffle, considering there are ten weeks of impressions to be adjusted for, while future weeks will only have to adjust for one week of impressions.
As one would anticipate, this process created some changes in our model’s view of the playoff picture. Here’s what happened:
Losing to South Carolina Is Allowed
Georgia’s home loss to South Carolina appears to have already been almost entirely forgiven. The Dawgs were narrowly behind Boise State for the second-largest positive adjustment from the rankings, with the result being that their playoff probability is now more than three times what it was at this time yesterday. The adjustment is nearly half of the forgiveness Clemson received in 2017 for losing at Syracuse in a game Kelly Bryant exited early with a concussion. It is, to say the least, surprising.
It’s possible the committee will backtrack on this, but for the time being, our model assumes it won’t. For now, Georgia gets a good deal of that loss wiped clean.
It’s hard to say whether this is an extension of Georgia’s favorable ranking or a cause of it, but Florida also received a noticeable boost from the committee. The Gators are such a longshot that the adjustment doesn’t affect the playoff picture, but it definitely affects the New Year’s Six landscape, as if the season ended today, Florida would receive the Orange Bowl nod.
Clemson Gets a Gift
Clemson has had some bad luck this year. Despite a non-conference schedule featuring two respectable SEC teams, they’re in danger of finishing the year having played zero teams in the final top 25, which is why, while Clemson didn’t themselves receive a positive adjustment from the rankings, they could be the biggest beneficiary of the largest adjustment the committee handed our model.
Wake Forest has one loss, a three-point home affair with Louisville in which they allowed 62 points. Their most impressive wins, according to our model’s estimations, are a home triumph over North Carolina and a road victory over Boston College. The two came by a combined nine points, against teams with a combined 9-9 record. The aggregate ratings our model uses to predict outcomes thinks Wake Forest is, at this point in time, the 42nd-best team in the FBS.
The committee ranked them 19th.
This isn’t to criticize Wake Forest. They haven’t done anything wrong. It isn’t even to criticize the committee—they might be right about the Deac’s. It is, though, very surprising to our model—the most surprising thing these rankings produced. And the end result is not that Wake Forest is suddenly a playoff contender or anything that concerns them directly. It’s that Clemson now has a much better chance of having at least one top 25 win, which could make the difference in an eventual comparison with a one-loss contender such as Oregon, Utah, Oklahoma, or a non-conference champion from the Big Ten or SEC.
Clemson could also get that top 25 victory (or a second top 25 win, if things really break their way) from Texas A&M, who our model thinks was one of the closest unranked teams to make the top 25 last night. Regardless, the Wake Forest ranking could end up being a life-saver for Clemson, who, it should be noted, probably won’t need it (a Power Five championship counts for a lot in our model, even if the conference isn’t one of the five best), but would nonetheless like to have it in their back pocket. Wake Forest’s trip to Blacksburg this weekend has suddenly become a whole lot more interesting.
No Love for Baylor or Minnesota
Our model recognizes that Baylor and Minnesota each lack particularly impressive victories. But it was surprised the committee placed the Bears and Gophers as low as they did.
Baylor’s beaten Kansas State—whom the committee ranked 16th—by three scores on the road. Baylor’s beaten Oklahoma State—whom the committee ranked 23rd—by three scores on the road. Those are solid results, but evidently the Bears are lacking something major in the committee’s eyes, as they and Minnesota shared the most significant negative adjustment this week. Stepping outside of the model (and it’s possible the committee chair addressed this), I’d guess Baylor is getting dinged for their nonconference schedule, something that’s inconsistently been a sticking point for the committee in years past. That’s just a personal guess, though, and something we’ll research in the future.
Our model didn’t like Baylor’s playoff chances yesterday, mostly because it assumed they’d lose at least once to Oklahoma before all was said and done. It really doesn’t like them now, because going undefeated might not be enough to become the eleventh program to appear in a playoff game. Their chances are less than half what they were yesterday.
Minnesota doesn’t share Baylor’s strength of best victories. Their two most impressive wins, not adjusting for score, were road escapes over Fresno State and Purdue. Those are fine wins, but not exactly playoff stuff. Still, our model was surprised by just how much the committee punished the Gophers. As with Baylor, it’s possible nonconference scheduling did them in.
Utah’s Gain Is Oklahoma’s Pain
The two largest climbers in playoff probability, in terms of sheer distance from where they were yesterday to where they are now, were Georgia and LSU. But Utah was notably next on the list.
The Utes’ playoff chances jumped noticeably when our model learned the committee ranked them eighth this week. Similarly to Georgia, Utah seems to be being forgiven a rather unsightly loss to USC, perhaps because it was on the road, or because it happened in September. In very related news, the biggest downward mover in playoff probability following the release of the rankings was Oklahoma, whose chances were cut by roughly a quarter. It’s possible that the road loss to the 16th-ranked team is being punished because of how much Kansas State controlled that game. It’s also possible the Sooners will swing back into favor soon. Either way, Utah’s in a much better place now than they were yesterday.
Other Notes
Auburn did not receive the love our model expected for having defeated Oregon and played one of the toughest schedules in the country so far. They were behind Minnesota and Baylor for the largest negative adjustment. Michigan also received a negative adjustment, perhaps for the way in which that loss to Wisconsin occurred, while Wisconsin, for their part, received a slight negative adjustment, perhaps because of the nature of their own loss (which would, it should be noted, not be out of line with past precedent).
Kansas State, Memphis, and—most importantly—LSU all received positive adjustments. The narrative has LSU’s schedule pegged to be stronger than it actually has been (Texas, unfortunately for the Tigers, isn’t all that great), but it’s still a reputable schedule, and it wasn’t the largest surprise for our model to see LSU ranked so highly. It does hurt Penn State’s chances, but our model was right in anticipating the Nittany Lions to be ranked in the top four, even if they were not ahead of Alabama and LSU as our model narrowly projected.
Our model correctly predicted that Ohio State would be the top-ranked team, but with LSU and Alabama squaring off, that status could be short-lived. One thing that makes evaluating the rankings difficult is not knowing how large the gaps are between teams in the committee’s eyes. Our model, though, sees significant gaps between Ohio State and LSU, between Clemson and Georgia, between Michigan and Notre Dame, and between Cincinnati and Memphis, for what it’s worth. We’ll learn more this weekend, and we’ll have more on what we might learn come Friday, so be sure to come back and see us then.
