Cheeseburgers were moderately helpful for Notre Dame this week, as five of the six remaining game win probabilities in SP+ and FPI inched upwards (the sixth stayed level, if you’re wondering). The overall picture hasn’t appreciably changed from last week; both models have it as roughly a coin flip for the Irish to win out, with FPI a bit more optimistic than SP+ based on their respective views of the North Carolina game.
Recapping the methodology for these articles: These aren’t actual SP+ win probabilities, as SP+ creator Bill Connelly doesn’t publish those regularly. He does though publish his SP+ ratings every week for all FBS teams. We used a slightly tweaked version of the formula that Reddit poster rcfbuser (account deleted, pour one out) reverse-engineered using these ratings to get close to SP+’s published probabilities. That formula does some probability magic on the differential in SP+ rating for the two teams (with a bump for home field, and yes, the bump is smaller this year given the smaller crowds), and voilà . The FPI win probabilities are updated weekly by ESPN, so those are the real deal.
SP+ measures offensive and defensive explosiveness and efficiency on a per-play basis. FPI is ESPN’s proprietary metric and is much more of a black box, but we do know that it rests heavily on expected points added, which is a pretty solid advanced stat and also a per-play number.
Post-Week 11 Update – SP+ Matrix
As you move down each column, you find the probability of Notre Dame owning that many wins at that point of the schedule. In the fourth row, for example, you can see that the probability of owning 0 wins through four games is 0% and four wins is 100%, since we already won them. In the last row of the table, we’re tracking how much the probability of each win total has changed from the previous week. This is a function of our own outcomes, the quality of play that led to those outcomes, and SP+’s changing perceptions of our past and future opponents.
The 46.4% figure at the end of the Wake Forest row reflects our probability of winning out per SP+. That’s only up 1.22 points from last week, which isn’t surprising given that it was pretty high already and the Irish were idle.
Post-Week 11 Update – FPI
FPI continues its trend of liking Notre Dame slightly more than SP+ does this year, with the win-out probability at a lofty 56.0%. Notably, it still has a much brighter view of Irish fortunes against North Carolina, with that game probability nearly 20 points higher than SP+’s calculation. Let’s hope Bristol got this one right.
Week by Week Game Trends
| SP+ |
|---|
 |
| FPI |
 |
Not much to cover here given relatively small changes in both models. If Notre Dame does indeed win this weekend, there will probably be very little change next week as well given the remaining opponents.
Closing Thoughts
After the Clemson win, when Notre Dame’s SP+ win probabilities in all their other remaining games actually dropped, we noted this:
This feels like a good place to bring up one of SP+ creator Bill Connelly’s most common defenses of his rankings, on behalf of both the SP+ and FPI models: These models are meant to be predictive, meaningful helpful in estimating how two teams would fare against each other. They are definitely not meant to provide a ranking based on a team’s resume. Another way to look at is that SP+ and FPI are forward looking, while resume evaluation is backward looking.
And that still holds, but man, there’s some weird stuff going on with SP+ this year. Clemson’s SP+ rating has increased significantly since they played us, from 24.2 pregame to 27.1 right now. That well outpaces our own improvement from 21.0 pregame to 22.0 now, despite the game result and them not playing any games in the interim (bye followed by FSU postponement) while we beat a common opponent by a wider margin than they did. It’s more nuanced than that, of course, and there are opponent and conference adjustments involved that are much harder in a year with no intersectional FBS play. Nonetheless…
Those relative rating shifts mean, that after Notre Dame beating Clemson and beating up worse than they did on a common opponent, SP+ now thinks significantly less of our chances in a rematch. Had the first game been played at a neutral site, SP+ would’ve set the line at Clemson -3 and the ND win probability at 42.9%. As of right now it would set the line for a neutral site rematch at Clemson -5.1 and the ND win probability at 38.7%. The longtime readers here know that I’m a staunch defender of SP+ and a big believer in the work Bill Connelly puts into his model. I’m just having a hard time figuring out what the model likes this year.
Going back and comparing week by week, it’s pretty clear what happened in SP+. This week we and Clemson both moved up by 1 point for being idle, and last week we gained 1.8 points after beating BC while Clemson dropped 0.8 points for being idle.
The big shift was 2 weeks ago after we beat Clemson – they gained 2.7 points and we lost 1.8. I know his numbers showed a slight win for Clemson, something like 58%-42%, but that shift strikes me as really large for what seems like a fairly small advantage in a game where they were already favored. I guess when you’re that high in the ratings even a small advantage is exceptionally meaningful for assessing a team’s strength.
There’s a lot else going on in the model besides just head-to-head results – for example, Pitt demolished Virginia Tech and Louisville killed Syracuse, which theoretically would give just us a nice boost since Clemson didn’t play either of the winners. Virginia clobbered Abilene Christian, which probably gave some small bump to Clemson, and Syracuse getting killed by Louisville should’ve been an offset to that. Those are the only opponents who were in action for either of us, so opponent adjustments don’t explain this week’s shift.
One of the biggest parts of balancing the model in regular year is strength of conference adjustments. I’m not sure how Connelly is adjusting for conferences not playing each other this year, but there has to be some kind of arbitrary thumb he’s putting on the scale for it and I think that’s where most of the wackiness is coming in. It’s an unenviable task.
Plus prior season weighting is all over the place, due to the wildly different number of games many teams have played.
Also a good point. In past seasons, preseason projections count for a diminishing proportion of a team’s rating up until game 7, and then it’s based entirely on that season’s performance. Connelly said he’s keeping that game 7 milestone in place for all teams regardless of when their season started, so for example, Wisconsin’s rating still factors in about half weight on preseason projections while they’re long gone from ND’s rating.
I think that we also have to keep in mind that in this season of covid, team performance might be a little bit more variable week to week due to starters randomly missing games because of covid test results / contact tracing as well as some teams having practice weeks occasionally disrupted due to small outbreaks in the team. Variability can throw a little noise into the models