Michigan #1 in SP+...with a huge caveat
Tweet speaks for itself. Bill Connelly never shares this for obv reasons, but it's still fun!
September 19th, 2021 at 11:49 AM ^
Honestly, ill take it.
September 19th, 2021 at 11:51 AM ^
Beat Rutger
September 19th, 2021 at 11:51 AM ^
Memorialize this very moment as a Michigan fan. Let’s live in this glorious moment for as long as possible.
September 19th, 2021 at 11:51 AM ^
I replied to an earlier comment, but it showed up here. I am taking my own command to memorialize this way too seriously.
September 19th, 2021 at 11:52 AM ^
Better than the worst Rutger
September 19th, 2021 at 12:58 PM ^
Or to my ear, Ruckers.
September 19th, 2021 at 2:17 PM ^
My 12 year old won't stop pronouncing it Rug-ters
September 19th, 2021 at 8:15 PM ^
My wife calls them "rudders, or whatever," which I like best
September 19th, 2021 at 1:12 PM ^
Re-Rutger Rutger
September 19th, 2021 at 2:24 PM ^
Indeed. Can't look past our arch rival from New Jersey.
September 19th, 2021 at 5:39 PM ^
“Questions.”
September 19th, 2021 at 3:33 PM ^
EDIT: dreaded double Rutger!
September 19th, 2021 at 4:39 PM ^
Can we award Corum his September Heisman while we're at it?
September 19th, 2021 at 11:48 PM ^
Who are you and why do you come here?
September 19th, 2021 at 11:52 AM ^
Why would you still factor in preseason projections? We're 25% into the regular season.
September 19th, 2021 at 11:54 AM ^
It's a fudge factor to make prognosticators look good. A sort of self fulfilling prophecy to game the system.
September 19th, 2021 at 12:19 PM ^
It's not a 'fudge factor'. Previous seasons are predictive. Your take is just uninformed.
September 19th, 2021 at 12:23 PM ^
2020 says "Hold my beer".
September 19th, 2021 at 12:44 PM ^
It doesn't only look at last year. It's the last 5 years I'm pretty sure. Also I'd guess he's weighing last year less than usual given it wasn't a full season.
September 19th, 2021 at 1:45 PM ^
For many teams, how they performed 5 years ago doesn’t seem particularly informative. It tells us nothing about the players on the field and in many cases little about the coaches.
September 19th, 2021 at 7:27 PM ^
My bias last year predicts my bias this year.
I got it.
September 19th, 2021 at 12:01 PM ^
Polling and ranking has let preseasons projections corrupt their conclusions for decades. This isnt honestly any different.
September 19th, 2021 at 12:19 PM ^
SP+ is 54% against the spread picking every single game but sure it's the same right!
September 19th, 2021 at 1:52 PM ^
Doesn't S&P+ throw out garbage time data? And if so, would that explain why it does not perform well as a way to pick against the spread?
September 19th, 2021 at 2:34 PM ^
Uh, 54% is very good against the spread. It was even more successful until Vegas figured out to use the SP+ spread to inform the lines. I don't think you can find a better comprehensive system for picking ATS for every team/game in FBS.
September 19th, 2021 at 2:40 PM ^
O.k. I never bet so 54% did not sound great to me. Anyway, if 54% success is considered good, then I guess I will just look at whatever pick S&P+ predicts, make my bets, and retire early.
September 19th, 2021 at 2:55 PM ^
54% is what you need to break even against the vig, so you won't get rich at that rate.
September 19th, 2021 at 3:45 PM ^
Isn’t the spread literally designed to have half the people pick every game correctly? If the masses can get 50%, a model getting 54% is not super impressive. We’ve moved all the way from a coin flip, to a little better than a coin flip.
September 19th, 2021 at 7:42 PM ^
All the systems we're familiar with--Sagarin, Massey, FEI, SP+--have outperformed the Vegas spread over the last several bowl seasons. SP+ has actually been the worst of the four, though I think Connelly's last round of tweaks probably fixed the problem. It's been interesting to me that the metrics that incorporate more data have, in general, been less predictive. I didn't expect that.
Michigan is now less than half a point behind Alabama in Massey, by the way. I don't think we've ever been that close since before the ND game in 2008.
September 19th, 2021 at 2:37 PM ^
54% against the spread is pretty good though?
September 19th, 2021 at 3:47 PM ^
No it does not. Football Outsiders has a more comprehensive analysis an FEI ranking, basically an S&P that does factor in garbage time.
Last week Michigan was 6th.
September 19th, 2021 at 12:21 PM ^
But the whole point is (or is supposed to be) that computer rankings are different than that, that they’re based on objective data. They shouldn’t be influenced by how some dude thought a team was going to perform before the season started.
September 19th, 2021 at 12:44 PM ^
sp+ isn't based on subjective data like that though. Maybe learn the basics about sp+ before taking shots at it?
September 19th, 2021 at 2:36 PM ^
You know they have data that is put in that calculates the preseason projections right? It’s not “how some dude thought a team would perform.” It’s how the data predicted they would perform.
September 19th, 2021 at 12:15 PM ^
... because his analysis shows that including them increases the accuracy of his model.
September 19th, 2021 at 12:24 PM ^
What does “accuracy” mean here? The data is what it is. If the data says we’re the best through three weeks, he should publish that. If it changes, so what?
September 19th, 2021 at 12:47 PM ^
"is what it is" is meaningless. Accuracy means accuracy of predictions about future games.
September 19th, 2021 at 12:57 PM ^
Accuracy in predicting the score of a game. He's been doing this for years and the data has shown repeatedly that including preseason projections, and then slowly phasing them out as the season progresses, increases the accuracy of his predictive model.
September 19th, 2021 at 12:18 PM ^
Because three games is a very small sample size in a sport with a wildly high amount of variance and many teams haven't really played many exceptional opponents after three weeks.
September 19th, 2021 at 12:20 PM ^
Because it makes the rankings far more predictive. 3 games is an absurdly small sample size. They gradually get phased out in sp+ as more data is collected.
September 19th, 2021 at 12:23 PM ^
3 games? My man, it takes like 20+ college basketball games for the Kenpom rankings to make sense. 3 game is nothing when it comes to sample size in predictive metrics.
September 19th, 2021 at 12:31 PM ^
SP+ is:
What is SP+? In a single sentence, it's a tempo- and opponent-adjusted measure of college football efficiency. I created the system at Football Outsiders in 2008, and as my experience with both college football and its stats has grown, I have made quite a few tweaks to the system.
It is, as always, important to note that SP+ is intended to be predictive and forward-facing. It is not a résumé ranking that gives credit for big wins or particularly brave scheduling -- no good predictive system is. It is simply a measure of the most sustainable and predictable aspects of football. If you're lucky or unimpressive in a win, your rating will probably fall. If you're strong and unlucky in a loss, it will probably rise.
Preseason data is de-emphasized every week as more data points are available, but inclusion of those projections makes the model better.
September 19th, 2021 at 2:30 PM ^
They gradually phase out the preseason projections as real game data is accumulated. This early, preseason projections are still part of it. Though a much smaller part than they were in previous weeks
September 19th, 2021 at 11:52 AM ^
This makes some sense. If you were born right before the season started and knew nothing other than what you've seen on the field the first three weeks, I think you could make a reasonable case for Michigan, Georgia, and Ole Miss looking like the best college football teams. Though I'd also throw in BYU and Iowa rather than Texas Tech and Florida.
September 19th, 2021 at 11:56 AM ^
Florida at 5 but no Bama?
September 19th, 2021 at 12:03 PM ^
Obviously the eye test says Bama is the best again. But it’s not like Florida got destroyed by Bama. It came right down to the end. I’m sure if they add preseason to the S&P, Bama is clearly number 1.
September 19th, 2021 at 12:50 PM ^
Florida outgained Bama by over 100 yards.
September 19th, 2021 at 12:03 PM ^
Bill Connelly on The Solid Verbal on Friday is quite enlightening.
About the 11:00 mark of Friday 9/17 podcast:
https://www.solidverbal.com/2021/09/17/bill-connelly-talks-numbers-latest-college-football-news/
Q. Do you ever have a moment where you are at odds with your SP+ just as a fan? Where you just look at the numbers and go "come on man, this can't be."
Connelly: You mean like when it has Michigan 6th? Michigan's the best example right now. They've absolutely looked great for two weeks. Don't know if they can pass, which will probably be a thing at some point, but we know that they've looked really good so far and Western Michigan is not terrible. Washington, eh, we'll see what they have to offer exactly. But they've looked undeniably good and they've moved up in SP+ and I just wish they hadn't. And the funny thing too is that Michigan fans wish they hadn't. No fan fan base yells at me as much for liking their team more than Michigan does, and I kind of get it because we know what could happen next.
Say what you will about his SP+ model, but the dude is spot on about the nature of Michigan fans.