clock menu more-arrow no yes

Filed under:

Predicting Rutgers Football’s Future: A Deep Dive into SP+ and FPI Analytics

New, 17 comments

After a gut wrenching loss at Michigan, a look at how the Scarlet Knights football season projections have shifted.

Rutgers v Michigan Photo by Gaelen Morse/Getty Images

Before we get to the analytics, a minor confession: I was one of the three contributors who predicted a Rutgers upset last Saturday at Michigan, and I was nervous as all heck about making that pick. It’s tough putting yourself out there with a pick that seems so uncomfortable. I felt that way until about 5:30pm on Saturday, when I started to think all of us insane people were on to something.

Now that a couple of days have passed, more than the final score I remember how stunned the Michigan faithful were for that entire second half on Saturday; it was like they were collectively thinking “RUTGERS is doing this to our esteemed football program?!” Even Harbaugh looked frightened. Cool stuff. I think Michigan’s fan base has bought into the Rutgers 2021 hype.

***

One more retrospective thing before I get forward-facing. That decision to go for it with 0:22 remaining in the second half, Michigan up 17-3, with fourth and ten from the Michigan 44? There are analytics that go into a decision like this, and here... Rutgers pretty much had to punt the football there. Even if you don’t trust your punter, Rutgers had to punt there.

Jon Bois of SBNation created the Surrender Index several years back to analyze the relative strength or weakness of punting decisions. It’s built off of NFL data (so it’s not an exact comparison, though, again, Adam Korsak is really good at punting the football). If you plug in the exact parameters of Rutgers’ late second quarter situation into the link above, you get a surrender index of 1.55. Without context, that’s meaningless — generally, the higher the number, the more of a surrender the punt is, and 1.55 is a really low number. My understanding is the average NFL punt has a surrender index between 3 and 4, so again, punting would have been more than reasonable there.

(You’re probably reading this right now and saying “Fred, I don’t need any analytics to tell me Rutgers should have punted there — look at what happened in the game!” I pretty much agree with that take. I’ll say one more thing and move on, that had Rutgers pinned Michigan there, they would have taken knees to end the half at 17-3, and had Rutgers been able to convert that fourth quarter field goal try, we are talking about a 17-16 Michigan lead with 9:38 left in the football game. So mid-game decisions have consequences.)

***

OK, on to the analytics. As I mentioned in my most recent post, the two predictive models in college football are both currently owned by ESPN, the Football Power Index (free link) and Bill Connelly’s SP+ (paywall link).

In a vacuum, for reasons that are super nerdy and not worth getting into, I prefer Connelly’s SP+ rankings (they’re tempo-adjusted better than FPI). But the problems with SP+ rankings right now, which I’ll feel better about in 1-2 weeks, are that in the early going of a given season they’re more heavily weighted toward the previous season, which breaks down in terms of accurately predicting what’s going to happen with a team like 2021 Rutgers (weird, pandemic-shortened 2020 season + new coaching staff + rebuilds are tough to time). So, I’m going to talk about both.

With respect to SP+,

  • Rutgers is now ranked 53rd in overall efficiency (two weeks ago, they were ranked 72nd).
  • Defensively, Rutgers is top-25 in efficiency, currently ranked 24th (two weeks ago, they were ranked 44th).
  • Offensive efficiency is ranked 75th (they were ranked 90th two weeks ago).

Obvious improvement here, I won’t get into week-by-week trends but the improvement was really from the Michigan game. The SP+ model continues to love Michigan, ranking them 8th in overall efficiency, so what it’s basically saying right now is “Michigan remains a good team, and it turns out Rutgers is pretty good, as well.”

Looking at the Football Power Index numbers,

  • Rutgers is ranked 35th in FBS, with an 83% chance of making it to six wins, and a predicted season record of 6.8-5.2 (you can round that up to 7-5).
  • That 35th overall ranking is good enough to be ranked eighth in the Big Ten, ahead of Maryland, Purdue, Indiana, Minnesota, Northwestern, and Illinois.
  • FPI also provides efficiency data, and nationally, Rutgers is currently ranked 8th in defensive efficiency and 6th in special teams efficiency. (Take these with a grain of salt; again, SP+ is tempo adjusted and ranks Rutgers more modestly in both areas of production.)

***

Looking at both the SP+ and the FPI numbers above, they are super encouraging to the glass-half-full Rutgers football fan. I think to myself, what a good time this is to be a Rutgers football fan and also (to very weakly paraphrase a very profane quote from the movie “Pulp Fiction”) let’s not get ahead of ourselves just yet. Every national college football bowl projector person with a pulse is going to project Rutgers into some kind of bowl game this week, and it all means nothing if Rutgers can’t scratch out three more wins this season.

Taking a step back, I’m aware of and respect the idea I sometimes hear in the comments that projections are fun and all, but the game is played on the field. Couldn’t agree more. When I write these articles, I’m hoping this is supporting info that makes the game watching experience more enjoyable. Hopefully Rutgers can continue the strong play supported by these analytics, and put up another positive effort against Ohio State on Saturday.