Rankings Roundup after Week 13: CFP, SP+, FEI

Submitted by Blue@LSU on November 29th, 2023 at 7:52 AM

Let’s do this.

COMPARING CFP, SP+, AND FEI (RANKINGS)

The penultimate CFP rankings are out and Michigan is in at #2 after beating the toughest bunch of tough bastards…Oh never mind, I’m not doing that shit anymore. Ryan Day* showed us who he is in the 1+ month leading up to The Game. And he turned out to be exactly who we thought he was. 

*[This toughness ranking is purely for Ryan Day. I don’t hold any real animosity toward the OSU players. Of course I hope they lose every game forever, but as far as I know, they haven’t said shit and there was minimal trash talk before the game. But correct me if I’m wrong.]

Anyway, the advanced stats are really liking Michigan. Want to see something beautiful? Here’s a look at the combined SP+ and FEI scores (not rankings) for the CFP top-25 teams.

FEI has two tiers at the top: Michigan/Oregon with UGA/OSU/PSU below them (or to the left). SP+ has Michigan/UGA in the first tier, with OSU/Oregon/PSU one tier below.

Here are the SP+ offensive and defensive rankings for the CFP top 25: 

And here are the FEI offensive and defensive rankings for the CFP top 25:

Look at them Bayou Bengals! Bless their hearts. It looks like they’re trying to pull off the AWOI. 

 

B1G AND MICHIGAN OPPONENT SP+ AND FEI RANKINGS

Meanwhile, in the B1G…Nah, I’m not doing that shit anymore. Fuck ‘em.

SP+ & FEI OFFENSE AND DEFENSE (RAW SP+ SCORES)

Standard interpretation applies:

  • Top-right: Good at offense and defense
  • Bottom-right: Good defense, below-average offense
  • Top-left: Good offense, below-average defense
  • Bottom-left: Not so good at football

Recall that these are standardized coefficients, not raw SP+ scores. Each team’s offensive and defensive score represents the number of standard above/below the mean that category. For example, Michigan is 1.25 standard deviations above the the average SP+ offense, and 2.27 standard deviations below (this is good) the average SP+ defense.   

And the same graph for FEI

FEI is a bit more forgiving than SP+ of the B1G offenses. Apparently it doesn’t watch B1G games (except for Michigan. I like Michigan’s offensive FEI score). 

SP+ & FEI RANKINGS BY CONFERENCE (RAW SP+ SCORES)

Who’s ready for the old Box and Whiskers? (I’m talking about the graph, perverts). The box is the interquartile range with the median value marked. The whiskers are the lower and upper extremes. Anything outside the whiskers is an outlier observation.

Spoiler alert: Michigan is an outlier.

Sorry, no snarky remarks from me this time. I’m burnt out and can only think of so many ways to say the B1G sucks ass. 

Go Blue!

Blue@LSU

November 29th, 2023 at 8:02 AM ^

Of all the ranking systems, FEI has apparently been the most accurate at predicting game outcomes according to these people. Thanks to the poster that pointed me to the FEI on one of the earlier roundups. 

https://twitter.com/CFPResumeRanks/status/1729484767727366584

 

maizenblue92

November 29th, 2023 at 8:12 AM ^

Excellent post as always. I have enjoyed following this all season. Small request, can you post the SP+ ratings of our opponents this season? I want to see who Iowa's offense most compares to. 

oriental andrew

November 29th, 2023 at 9:37 AM ^

Don't look now (or please do look), but UNLV is playing in the MWC championship game against Boise State on Saturday. Would love for UNLV to pull off the victory!

Fun fact - there was a 3 way tie in the MWC and because all three teams didn't play each other, they went to the next tie-breaker - computer ratings. 9-3 UNLV was first by a clear margin, followed by 7-5 BSU and 7-5 SJSU. 

MGlobules

November 29th, 2023 at 8:13 AM ^

Thanks for these! Here's a question: Which of these rankings, in all of y'all's view, takes the most considered view of the kind of offensive production we exemplify?

I'm thinking of how--when I watch the games a second time--I'm realizing that those running plays we depend on so heavily, which have me grinding my teeth in real time, actually keep us moving the ball, gaining first downs, using huge chunks of clock. . . OSU had more yardage, but when you re-watch, you really see (as Borges said) that we were in control and OSU chasing all the time. (In the best of all possible worlds one might even be able to calculate what the most efficient SCORE for the game might be!) Which stats best denote this kind of activity, that keep a team with (almost) two whole elevens on defense fresh for the second half bludgeonings we've been seeing?

EDIT: PSU and OSU really had fine teams this year, as these graphs, repeatedly, show. We just had an objectively better one. An oblate sphere takes funny bounces, but I feel pretty good going into the playoffs that our talented staff can maximize our many strengths. Will be crestfallen if we aren't in the final with Georgia, and in that game right up to the end. And on this same note, I really believe that we can titrate our offensive exposure through Iowa and (say) an FSU so that we have a great deal left in the arsenal for a Georgia. Oregon. . . we could surprise and blow them out or have to go full ham.

 

MGlobules

November 29th, 2023 at 8:28 AM ^

Yes, but amid all of these different attempts to characterize team strength how is efficiency being framed. . . is my (probably naive) question. Does FEI do that? Blue@LSU--above--says FEI may well be doing the best job, so am going to go see how that's put together.

EDIT: Just starting to dig, but I liked this:

https://18stripes.com/an-interview-with-brian-fremeau-of-the-fremeau-efficiency-index/

MGoCarolinaBlue

November 29th, 2023 at 9:48 AM ^

IIRC, both SP+ and FEI seek to measure efficiency, but FEI considers efficiency per drive/possession while SP+ looks at efficiency on a per-play basis.

Both systems however consider efficiency in terms of scoring.

It would be interesting to find some way to measure instead the per-possession change in win% probability as computed from the game state at the beginning and end of each drive.

What that would do, which these current measures aren't doing, is incorporate the value of a drive that burns clock while leading, even if it doesn't result in points. It would also properly weight the difference between a drive that gets 7 in 2 plays vs. a drive that gets 7 while burning half of the 3rd quarter.

BlueLikeJazz

November 29th, 2023 at 8:25 AM ^

Thanks as always for doing this. Gotta be honest—I’m really nervous about Oregon beating Washington and matching up with Michigan in the playoff. They look really dangerous.

UgLi Eric

November 29th, 2023 at 9:03 AM ^

Oh boy yes. I watched the Oregon / Washington game, and Oregon lost that one. But my feeling was more like, "Dan Lanning tried too hard and cost his team a few points" more than "Oregon didn't look good." 

Live, watching that game, my thought was holy shit Oregon are fast. They scared me then, before they were universally ranked about in party to us. Now they scare me even more...

But if you want to be a champion, you have to be fearless and beat the best.  So our warpath is Ducks, Dawgs, Cows and/or whatever Iowa is...

Amazinblu

November 29th, 2023 at 9:00 AM ^

It seems like five teams are in that “upper right” quadrant.  Michigan is one of those five - and has defeated two of the others over the last three weeks.

Like others, I think Oregon and Georgia are solid teams.  And - it’s one play, series, and game at a time.  The next play will be at 8:17 pm on Saturday evening.   

Thanks for doing this, great information.

Go Blue!

PopeLando

November 29th, 2023 at 9:27 AM ^

Love this feature 

The advanced stats are kinda backing up the rankings: there’s about 3 teams that are CLEARLY top tier, and 3 more that are only half a step back.

I remember the travesty of a ranked Colorado earlier in the season. Glad that those mostly have been excised from the rankings.

I’m really interested in doing a post-season “Iowa and Inverse Iowa Awards”

UMQuadz05

November 29th, 2023 at 9:32 AM ^

About your asterisk- this was weirdly a non-hateable group of OSU players.  I feared MHJ, JTT and the like, but in direct contrast to their coach they all kept their heads down and let their on field results do the talking.  

Amazinblu

November 29th, 2023 at 9:32 AM ^

I don’t know if a relatively static metric - 247’s Team Talent Composite - could be combined with this other data.  My feeling is - the result would indicate that Michigan has been outperforming it’s “team talent” for a while.  That would lead to the questions of why and how.

My answer would include: development, S&C, coaching, scheme, effort, and in-game execution.  All critical characteristics of the current culture which seems exceptional.

Go Blue!