S&P Ratings without Preseason Ratings built-in
Figured it was interesting to see how these rankings look without the weighting of the pre-season. MSU is interesting and purdue as well.
#2 OSU
#8 Wisconsin
#9 PSU
#15 Michigan
#22 MSU
#31 Purdue
#41 Minnesota
#42 Maryland
#47 Iowa
#48 Nebraska
https://www.sbnation.com/college-football/2017/10/1/16392556/ncaa-footb…
Five weeks into the 2017 college football season, we know at least a few things:
- Bama is Bama.
- Clemson looks great when it has to (then throws things back into cruise control).
- Ohio State responds well to early-season losses.
-
Outside of Bama, there’s very little separation among elites. That didn’t make for a very fun Week 5, but it could make for one tense November.
"I am again including where each team would rank if there were no preseason projections involved. Predictive success requires inclusion of such projections — for teams that have played five games, projections carry only 35 percent weight at this point — but I like including the non-projection rankings here. It both addresses a lot of questions people might have (“How is [Losing Team A] ranked so high???”) and gives you a better feel for which teams are trending up or down."
October 2nd, 2017 at 8:27 PM ^
October 2nd, 2017 at 9:27 PM ^
Who?
October 2nd, 2017 at 8:31 PM ^
OSU gets crushed at home by Oklahoma and proceeds to beat up 3 terrible teams and suddenly they're #2 in the country? ok
October 2nd, 2017 at 9:14 PM ^
October 2nd, 2017 at 9:45 PM ^
October 2nd, 2017 at 9:56 PM ^
Ohio State only had a 4% win expectation against Oklahoma. They were rather soundly beaten by Oklahoma --- that was no fluke --- and in Columbus no less.
Oklahoma also hasn't had any bad performances in their other games. Their game at Baylor was a bit close on the scoreboard, but still OU had a 99% win expectation there. Also a win expectation of 99-100% vs. UTEP and Tulane. Weak opponents, sure, but they did what a good team is supposed to do. Beat them soundly.
Given each of the above --- especially the nature of the OU win over OSU --- I really can't understand having OSU ahead of OU.
I like Connelly's work overall, but this is simply a miss.
October 3rd, 2017 at 12:31 AM ^
October 3rd, 2017 at 1:47 AM ^
I wouldn't be pointing to the Iowa game as support for that opinion though. Road conference games against average-ish foes are often tricky even for legitimate for conference championship contenders. Just in the last 3 years:
2016: PSU @ Indiana, OSU @ Michigan State, Michigan @ Iowa
2015: MSU @ Nebraska, OSU @ Indiana, Michigan @ Minnesota, Michigan @ Indiana, Iowa @ Indiana (poor Indiana in 2015).
2014: OSU @ PSU
The evidence for your opinion is more in PSU's game vs. IU rather than the Iowa game.
The former was not close on the scoreboard, but was also a home game where PSU had only a 63% win expectancy.
The latter was close on the scoreboard, but was also a road game where PSU had a fairly strong 85% win expectancy.
October 3rd, 2017 at 8:04 AM ^
Isn't the win expectation before the game? I don't think it's an in-game thing and you're saying it never dropped below 99% right?
I think that's the problem. It was supposed to be a near certainty that they would win but the game itself was actually competitive.
Also, yes they've played some bad teams, but Ohio State has absolutely hammered them, and as much as people think a solid win is a solid win, it's really not. How badly you beat a team, even a bad one, matters. That's why Kenpom is so accurate as well. Maybe Oklahoma is better, but Ohio State is still a really good team unfortunately. Like the poster above said, it's not a resume. If they played again, OSU may very well win the next time. We're talking about very small sample sizes still.
Having said that, I don't think Connelly's numbers can in any way account for specific matchups, and I feel like Oklahoma was the perfect matchup problem for OSU because of OSU's secondary and Oklahoma's dangerous passing attack.
October 3rd, 2017 at 10:38 AM ^
In-game and post-game.
ESPN (and others) have their in-game win expectations.
Connelly's numbers are post-game numbers. Look at each team's end-game statistics (yards, turnovers, yards/play et cetera) and based on both (1) the difference of such and (2) historical results, how often does a team w/ those sort of stats win. Any "garbage-time statistics" are thrown out.
Not a perfect metric, but I do like it quite a bit.
(for the statistically inclined, Connelly's almost undoubtedly running a logisitical regression)
October 2nd, 2017 at 8:33 PM ^
October 2nd, 2017 at 8:44 PM ^
October 3rd, 2017 at 8:06 AM ^
You're looking at the rankings that still have preseason weights in there.
October 3rd, 2017 at 9:23 AM ^
They dropped both ways after beating a much higher rated team on the road. How do you explain that and why be dismissive of their dropping when the preseason ranking is included?
October 2nd, 2017 at 8:44 PM ^
October 2nd, 2017 at 9:07 PM ^
October 2nd, 2017 at 9:12 PM ^
October 2nd, 2017 at 9:15 PM ^
October 2nd, 2017 at 9:26 PM ^
Penn State has demolished everyone so far, which isn't necessarily a guarantee they'll be a CFP contender... But it helps. UM is getting weighed down by the Cincy game. Also, as Bill C himself will tell you, there is a reason he keeps preseason predictions as a component till Week 7 (I think, maybe it's 6 games.). UCF is #4, for example, because the computer doesn't know about Maryland's QBs.
October 3rd, 2017 at 4:00 AM ^
No, they have demolished every team they have played so far that isn't Pitt and Iowa, the two teams on their schedule that aren't worse than mediocre. Neither were dominating performances. I've yet to see a reason to think they haven't just been prancing through a cakewalk portion of their schedule.
They can still convince me. They haven't yet.
October 3rd, 2017 at 7:30 AM ^
Doh! I forgot about Iowa.
Pitt, they beat handily.
October 3rd, 2017 at 8:09 AM ^
You can knock the numbers, but his numbers outperform Vegas, which I'm sure means they'd outperform you.
Also, I think we can definitely win in Happy Valley, especially if O'Korn plays like he did against Purdue, but you're obviously being a homer if you think at this point in the season Michigan has looked better than Penn State.
October 2nd, 2017 at 9:08 PM ^
I'm surprised MSU is at 22 after getting trounced by ND at home. Also, S&P not a big fan of Iowa so they shouldn't have a big boost from them.
October 2nd, 2017 at 9:18 PM ^
October 2nd, 2017 at 9:56 PM ^
I think ND is good. But that's still a game that was 38-10 in East Lansing before a late score. I know MSU outgained ND pretty significantly yardage wise but I thought S&P removed "garbage time" from their calculations.
October 3rd, 2017 at 12:15 AM ^
That was also a 3 turnover game for MSU which affected the result rather significantly.
October 3rd, 2017 at 8:10 AM ^
Yes, I don't know exactly how Connelly treats turnovers, but I know for the most part he treats them like Brian does. Mostly luck.
October 2nd, 2017 at 9:16 PM ^
October 2nd, 2017 at 9:32 PM ^
You know, if you weighted rankings on overall defense against mustaches, Michigan would probably still be #1 on several metrics. I suppose I hadn't thought of it like that before, but maybe it is something that will be taken seriously in future seasons. Good call.
October 2nd, 2017 at 9:21 PM ^
MSU at #22 is astounding to me. MSU vs. USC would be a bloodbath for the Trojans it seems, and yet they are lower-ranked.
I've listened to Connelly's podcast, and the take away I get from this is that because we don't have a lot of data still for some teams, the rankings without preseason weight can be misleading. Michigan is probably a top-10 team, Penn State is too. I still don't know about Wisconsin; it's a team that can turn it on in spurts but I watch them play offense and it just looks janky.
October 3rd, 2017 at 1:42 AM ^
I'm not so sure about that. Caveats about the transitive property and all, but MSU beat Western much more easily than USC did.
Trojans are incredibly inconsistent.
October 2nd, 2017 at 9:22 PM ^
Our ranking is probably misleading due to the qb change. If Okorn plays at the same level he played against Purdue or better than we are easily a top 5 team, if not than who knows I guess.
October 2nd, 2017 at 9:26 PM ^
October 2nd, 2017 at 9:32 PM ^
October 2nd, 2017 at 9:45 PM ^
October 2nd, 2017 at 10:13 PM ^
Our season is about to take a big turn up or down, largely depending on John O'Korn.
No one is really sure how that will go, so predicting the outcome for Michigan right now reflects a lot of guesswork and indecision.
From what I've seen of O'Korn with the pressure of being thrown into a game, and what Coach Hamilton has done for his confidence, I like our chances. The "improvement week" is a great opportunity to get everything in sync.
Go Blue!
October 2nd, 2017 at 10:30 PM ^
October 2nd, 2017 at 11:08 PM ^
October 3rd, 2017 at 4:36 AM ^
...soundly beaten 2 of them (Louisville, VTech) on the road.
Meanwhile, Bama's win over previously top 3 FSU continues to look worse...
With Mississippi State and LSU getting embarrassed, Bama won't be able to inflate its resume with wins against over-ranked teams...
Look at Bama's remaining schedule:
TAMU (4-1), Arkansas (2-2), Tennessee (3-2), LSU (3-2), Mississippi State (3-2), Mercer (2-3), and Auburn (4-1).
Only Auburn deserves to be ranked.
I can see a scenario where a top-10 Auburn team upsets Alabama in Jordan Hare on the final weekend of the regular season and plays Florida (or Georgia) in the SEC Championship Game.
In that scenario, an 11-1 Bama may get left out of the playoffs IF:
- the SEC Champion (Auburn/Florida/Georgia) has only one loss
- 12-1/13-0 Clemson wins the ACC
- 12-1/13-0 Oklahoma wins the Big 12
- 12-1/13-0 Michigan/PSU/OSU wins the Big Ten.
October 3rd, 2017 at 7:25 AM ^
October 3rd, 2017 at 8:58 AM ^
I'm not sure that Alabama's win over FSU is going to look bad. I think most people would recognize that FSU is a very different team without Francois.
October 2nd, 2017 at 11:27 PM ^
October 3rd, 2017 at 9:04 AM ^
In the offense UFR thread I said 3 more losses and im going into #fireborges mode towards drevno. If MSU wins this weekend I am activating #firedrevno. I dont see that happening but IF it does drevno is the fall guy.
October 3rd, 2017 at 11:20 AM ^
The sample size is still too small for meaningful results. Ask again later.