What Our Model Says After College Football’s Week Nine

Our model is eight days from receiving a jolt of information in the form of the season’s first College Football Playoff Rankings. For now, it has to base its impressions on the precedent past committees have set. Here are those impressions, following a week that saw further winnowing near the top:

Oklahoma Down

The biggest story of the week was Kansas State’s upset of Oklahoma in Manhattan. The Wildcats surged to a big lead in the second half and held off a furious Sooners charge, with the help of a fraction of a yard on the climactic onside kick.

Oklahoma, entering the day, had been in the thick of the race for a playoff spot, 46.1% likely to crack the field. They exit the week just 10.0% likely to make it.

Buoying OU’s chances is the fact they’re still a 63.6% favorite to win the Big 12, with Baylor the only remaining serious contender at 25.3%. Oklahoma gets the Bears on the road in a few weeks in what could be a preview of the Big 12 Championship. Baylor, undefeated as of yet and favored by the aggregate ratings our model uses in each of their remaining non-Oklahoma games, is not a playoff contender right now, measuring in at only 2.5% likely to make the top four. The Bears, though, project to be the highest-ranked team on Oklahoma’s schedule, which means the Sooners want them to win all their other games to make the hypothetical 12-1 résumé look a bit better. This is possible, and of course, Kansas State picking up a lot of victories wouldn’t hurt, but a 12-1 Oklahoma with a loss to the Wildcats would need help elsewhere—either from teams losing or from a change in precedent.

Adjusted Point Differential (APD), our homegrown but not-uncommon-conceptually metric that measures how well a team performs relative to other teams on their opponents’ schedules, is what we use to approximate the impact of “the eye test” and objective rating systems on the committee’s ratings. This is not an unimportant part of our model. Neither is strength of schedule. But what our model has seen as the tiebreaker in past seasons between, say, a 12-1 Power Five champion and an 11-1 Power Five team that missed its league’s championship, is a combination of each program’s two best wins and worst loss. Oklahoma’s worst loss might be decent, but it won’t be good. Their best wins, likely both Baylor in their best-case scenario, don’t figure to be all that great either. Without an enormous boost from the variables APD approximates, which they’re unlikely to get, and without a particularly grueling schedule, Oklahoma will not be in the driver’s seat even if they win out. And the teams that will be in that driver’s seat, according to our model, aren’t the ones you might expect.

Penn State Rises

It’s unclear if Penn State was the biggest beneficiary from Oklahoma’s loss. Other factors helped the Nittany Lions. They handled Michigan State on the road with ease. Wisconsin not only lost to Ohio State, but lost by a lot. But regardless of where their increased hopes came from, Penn State saw the largest overall rise in playoff probability across the weekend, jumping from 34.9% to 57.7%.

Our model currently only has Penn State as 19% likely to beat Ohio State in Columbus the weekend before Thanksgiving. It only has the Nittany Lions as 15.1% likely to win the Big Ten. Which means it clearly believes Penn State not only has a real chance to make the field at 11-1 but thinks such a possibility is reasonably likely.

This is where our model’s formula might be put to the test. Penn State has played a respectable schedule so far and has solid wins on the road against Iowa and Michigan State in addition to their home victory over Michigan. They beat Maryland by 59 points, which impresses APD, and the fact their victory over Michigan State came by 21 helps them in that field too. If they lose to Ohio State on the road, our model will treat such a “worst loss” as akin to going undefeated, considering Ohio State grades out as the best team in the country and road losses get a bonus in that measurement. The media, though, has yet to latch onto Penn State as a contender, and it has yet to jump on the narrative that Ohio State’s the best team in the country. If the committee agrees, or is even moderately swayed by this status quo, our model will have over-estimated Penn State’s playoff chances. We’ll find out a lot more a week from tomorrow night.

LSU Rises

Our model likes LSU too. It has the Tigers 51.1% likely to make the playoff. It thinks their 70% likely loss to Alabama would be a very good one, and its wins over Florida and Auburn at home will measure up well. But it sees the victory over Texas as less impressive in hindsight than it was treated at the time. It’s concerned two more possible Auburn losses (Auburn has yet to host Georgia and Alabama) will both come to fruition and damage that second-best victory. It’s similarly concerned that Florida will lose to Georgia, or even Missouri, and make LSU’s top-heavy schedule look more middle-heavy.

Stepping outside of the model for a moment and viewing it in the context of the media landscape, I’d guess that it’s underrating LSU’s chances and overrating those of Penn State. The committee is comprised of humans, and our research has certainly shown subjectivity to have played a role, which is part of why we have a feature in our model that allows us to adjust how wins and losses are being treated once we have rankings to demonstrate this. With that being said, I’d also guess the media is overrating LSU’s chances and underrating those of Penn State. The SEC has a stronger second tier than the Big Ten, with Florida, Auburn, and Georgia better on the aggregate than Wisconsin, Michigan, and Iowa. But at the third tier, Minnesota, Michigan State, and Indiana are better than Texas A&M, Missouri, and South Carolina. LSU doesn’t play Mizzou or the Gamecocks. Penn State plays the Gophers and the Hoosiers and already took care of the Spartans. It’s hard to see LSU having more top 25 wins than Penn State in the scenario in which they’re being compared as 11-1 teams. It’s possible the difference will be as large as two or three. That difference could make the difference.

The Pac-12’s Still a Longshot

Notably absent from the conversation so far is the Pac-12. While Oklahoma’s loss, according to immediate reactions, was a boon for the conference, their playoff probability as a conference actually dropped slightly after the week, from 8.5% to 8.3%.

What’s wrong?

It has to do with the conference favorite.

Oregon has a solid playoff case if they go 12-1. Their only loss will have been to Auburn at a neutral location, and they’ll have at least one top 25 win, assuming the Pac-12 South champion can stay in the committee’s good graces. The Ducks are still the conference favorite, 53.6% likely to win the league, and they’ve all but wrapped up the North division, two games ahead of their closest competition with just four games to play. But that probability is lower than it was before the week. Before Saturday’s escape from Washington State, the Ducks were 58.9% favorites. The difference isn’t huge, but it’s significant that it changed that way after a win, and the force that’s driving it is even more significant.

Prior to last week, Utah was viewed by the ratings we use as a point and a half worse than Oregon on a neutral field. After Saturday, they’re viewed as one point better.

That’s an important swing, because it means that Oregon’s favorite-hood is dependent on the slim possibility that USC will win the South and keep Utah out of the Pac-12 Championship. If that doesn’t come to fruition, Utah will be the Pac-12 favorite, and Utah’s playoff case is not as strong as that of Oregon, because the Utes lost to USC.

Even if Oregon does win out, their inclusion is far from guaranteed. They’d need a lot of luck to have more than one top 25 win, and results like Saturday’s don’t impress APD—nor do they figure to impress rating systems or the eye test. The committee does value conference championships. Ask 2017 Ohio State and Alabama how much.

The Top Tier Stays on Top

While the second tier in the playoff race—teams who control their fate but aren’t favored in all their remaining games—has lost half its members the last two weeks, the top tier remains intact. Ohio State, Clemson, and Alabama are all highly likely to make the playoff field. Ohio State is a near-lock, at 97.2% likely. Stepping outside the model again, I’m nervous that it’s giving that probability, since it’s dependent upon the assumption that even with a loss to Penn State, the Buckeyes would probably be in, which is a risky assumption for reasons covered earlier in this post. On the other hand, Ohio State’s played better than anyone else in the country this year. Every rating system we use has them as the best in the nation. They’re favored by 13 right now in that eventual matchup with Penn State at home. They’d be favored by 14 in a rematch with Wisconsin in the Big Ten Championship, and they’d be favored by even more against Iowa or Minnesota. The Buckeyes are very, very good.

Group of Five News

Air Force impressed this week, whooping Utah State to move into second place in Mountain West title likelihood. Ohio beat Ball State to reclaim the title of MAC favorite. FAU’s throttling of Old Dominion helped them move closer to Louisiana Tech in the Conference USA picture. SMU kept winning, with a big date with Memphis coming up Saturday. Appalachian State kept winning, making that November 9th date with South Carolina stand out even taller as the lone serious obstacle between the Mountaineers and a perfect regular season.

The Best Teams

As we do every week, we’ll close with who those ratings we use rate as the best teams in the country at this point in time. The number to the right of each team is how many points they’d be expected to lose by when facing Ohio State on a neutral field.

1. Ohio State (0.0)
2. Alabama (2.6)
3. Clemson (5.4)
4. LSU (7.8)
5. Penn State (9.9)
6. Oklahoma (10.1)
7. Georgia (11.4)
8. Auburn (12.5)
9. Utah (14.2)
10. Wisconsin (14.3)
11. Florida (14.8)
12. Oregon (15.1)
13. Michigan (15.5)
14. Washington (17.8)
15. UCF (18.1)
16. Iowa (19.4)
17. Notre Dame (19.5)
18. Minnesota (19.7)
19. Iowa State (20.0)
20. Baylor (20.0)
21. Texas A&M (20.4)
22. USC (22.1)
23. Texas (23.0)
24. Washington State (23.0)
25. Cincinnati (23.1)
26. Oklahoma State (23.3)
27. Memphis (23.6)
28. Michigan State (23.8)
29. TCU (24.1)
30. Appalachian State (25.0)
31. Missouri (25.0)
32. Boise State (25.4)
33. Indiana (25.7)
34. Kansas State (26.1)
35. North Dakota State (26.5)

The Barking Crow's resident numbers man. Was asked to do NIT Bracketology in 2018 and never looked back. Fields inquiries on Twitter: @joestunardi.
Posts created 3299

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts

Begin typing your search term above and press enter to search. Press ESC to cancel.