I'd like to know whether you are calling them a W or not. I'm assuming we lose to Nebraska and Ohio given the records above.. I'm just wondering where the other losses are coming from. I think I'd rather lose to SDSU than ND.
Mid-Week Metrics: Projecting the Big Ten
College football is 7 days away. Michigan football is 9 days away. It is time for a little Big Ten preview. Last year my numbers pegged Michigan at 7-8 wins. This year you’ll have to read on to see my predictions for Michigan and the rest of the Big Ten.
The Nuts and Bolts
If you just want to see the picks and the nice standings you skip on ahead. If this section confuses you or brings about more questions than answers, you might want to head here.
My methodology is along the same lines of user Undefeated Dream Season of 1992’s great post from last week.
Begin with the PAN from the team’s previous season. Regress that season half-way to a team-specific mean, which for me is the five preceding years, then adjust for returning starters. Every team ends up with a rating which is then plugged into the full season schedule and simulated a whole bunch to produce average results for every team in the FBS.
I weight returning starters based on what I can find validation from in past seasons. I am continually tweaking this because it is very difficult to separate out, but my best method currently accounts only for returning QBs on offense. A returning signal caller is worth 1 extra point per game vs average and a loss of QB is a 1 point reduction, leaving a 2-point spread. Once accounting for a regression to the mean and the QB effect I can’t find any other correlation across returning offensive starters. On defense the break-even point is seven returners. Each player returning above or below seven is worth 0.8 points per game. Return all 11 and it’s 3.2 points per game. Return 3 from the previous season and it’s –3.2.
For prediction purposes I exclude special teams because their success or failure isn’t typically consistent from one year to the next like offense and defense are. Almost all teams are predicted to have 2+ losses because even though you know several teams are going to run the table or have just one loss, which teams is a challenge and my numbers are based on averages across multiple “plays” of a season.
The Power Poll
Michigan checks in right in the middle of the Big Ten at +4 predicted. I didn’t know what to do with Purdue at QB since they have just been a mess the last two years, but hedging to the negative is probably the right call. Those of you familiar with my numbers know that Northwestern has some sort of crazy luck/skill at exceeding their numbers year in and year out. They are the one team in my ratings out of 120 that just never work and it’s always to the Wildcats' favor.
It should also be noted that I had Wisconsin as a non-returner at QB even though they kind of do have a returner. If Russell Wilson is counted as a returner, Wisconsin jumps to the top of the table.
The power poll tells you how good I think a team is but to get a read on how they will be predicted to do you have to factor in opponents and game locations.
Woody Division (R. Wilson as returning starter)
|Team||W||L||Conf W||Conf L||Conf SOS||NC SOS||SOS|
If you drop Wisconsin down based on Wilson, Ohio State sneaks into the top spot.
SOS indicates the average PAN rating for all opponents on the season.
|Team||W||L||Conf W||Conf L||Conf SOS||NC SOS||SOS|
Michigan, at 8-4 (5-3 Big Ten) comes in second in the division, but Michigan, State and Iowa are all virtually indistinguishable in spots 2-4.
The Big Ten is highly bunched this season. Whether it’s Wisconsin, Ohio or Nebraska that makes it through the championship game depending on the scenario, I am projecting the Big Ten winner to have the most conference losses of any conference winner in 2011. [Ed-M: I'm predicting SEC fans will give us shit for that.]
Overall, the Big Ten is slotted third in my preseason conference ratings behind the SEC and what’s left of the Big XII. In conference strength of schedule, the Big Ten ranks fourth behind the Big XII, Pac 12 and SEC. The SEC is the only conference with a weaker non-conference lineup than the Big Ten.
Michigan’s strength of schedule is ranked 12th in the country, Notre Dame and Ohio [Ed-M: He means OSU; next thing you know Mathlete's gonna be pointing at things too.] are #1 and 2. The SEC has the seven toughest conference schedules among its ranks but its cupcake-loaded preseason leaves them lower overall.
Predicted winners from other conferences:
|Team||Conf||W||L||Conf W||Conf L||Conf SOS||NC SOS||SOS|
|W Virginia||Big East||10.2||1.8||5.8||1.2||0.7||-3.3||-1.0|
|Boise St||Mtn West||11.8||0.2||6.9||0.1||-2.1||3.1||0.1|
Four of my top five match the AP top five (Boise, Oklahoma, Oregon and Alabama) but beyond that I have a handful of teams I think are over/underrated:
but as far as expectations I would think that a loss to ND would be much more acceptable than a loss to SDSU. ND should be a much more talented team and although I don't want to lose to either team, I would think that a loss to ND would hurt our team much less than SDSU because of the talent level on the ND team.
Very interesting idea. 1 quick thing, in your original post explaining PAN, the 3.95 should be 4.95.
Also, I may just not be understanding correctly, but how is the defense taken into consideration exactly? I get that it is taken into consideration when adjusting for the final PAN, but you have this whole scoring system for offense and nothing for defense. The way I undertsand it now, you're not really considering how good a defense is, but an offense's ability to perform against the average for the defense it is facing. So is there a PAN type stat for defense?
I'd take 8-4 this year with a win over the Buckeyes and losses to ND, Nebraska, Iowa, and an upset from Illinois or NW.
EDIT TO ADD: Forgot to mention that these kinds of things are much appreciated and what separates this blog from any other team blog on the web. How you crazy math guys come up with this stuff, I'll never know. It's witchcraft, I tell you.
Thanks for all your hard work on your posts, they are great reads and have plentu of valuable information.
They're forecasted for an average of 0.2 losses; this means your program thinks they're *heavy* favorites against Georgia, correct? That is a very surprising outcome...interesting.
Great work, Mathlete. Cool stuff.
returns a bunch of starters from last season. Boise State is a favorite to beat UGA.
Hey, have you done this for past seasons? How good generally are the predictions pre-season?
Interesting that Northwestern is predicted to be 4-8. I would be surprised if that actually happened, unless Persa's knee really hinders him that much. As you said, it seems more likely that the model seems to skew the numbers for Northwestern for some reason.
Achilles, not knee. Right? Or did Persa's knee go out too?
Persa might not be as mobile as he was pre-injury, but an Achilles repair should not re-rupture this year. I'm with you, by the way. I see Northwestern doing better than 4-8. And I see 7-5 as a real possibility this year for Michigan.
7-5 and a Gator Bowl loss in 2010 got Rich Rodriguez fired. Would 7-5 and a bowl loss in 2011 get Brady Hoke fired? We'll see.
No, it wouldn't. 7-5 and a bowl loss in 2013 might though.
Also, that was a stupid question.
It was a stupid question.
Brady Hoke can go 5-7 and not get fired. Brady Hoke could probably go 3-9 and not get fired, if the reason was that Denard, Martin, Demens and Woolfolk all suffered injuries. Which could happen. Or not.
So, yeah. It was a stupid, rhetorical, provocative question.
went 3-9 in 2008 and didnt get fired.....
i would be beyond happy with 8-4. thanks as always. a few questions:
was the regression term (halfway) from undefeated's post? i didn't see it in a quick skim or remember. have you had an opportunity to run anything along these lines with PAN to figure out in-season correlation? if you were just doing some best guessing, what would change if you switched up the regression term?
Provided one of the wins is over that suckhole Dantonio.
I really don't like that man.
I don't like him is because of their fuckhead fan base.
Oh, did I mention my two sisters went there??? lmao...
...but I think you're dreaming. 8-4 is perfectly acceptable, anything better you should allow yourself to be pleasantly surprised.
I just can't see us finishing any worse than 13-0. I mean, I started with your 8-4 prediction and then started thinking who that would mean we would have to lose to. You are predicting 1 non-conference loss, most likely ND, but I can't imagine we would drop that one under the lights with all of the excitement. So that puts us undefeated coming into the big ten schedule, then the only two teams I could see us maybe losing to are tOSU and Nebrawska. I think the whole team is smelling blood with tOSU wounded, so mark that down as a win at home. Nebrawska is the wild card, but us coming in at 10-0 and having the Heisman front-runner at that point in the season, I think momentum caries us through that one. Obviously we will win the bowl game, WE'RE MICHIGAN. So there you go, 13-0.
Michigan goes 12-0 but does not play in the conference championship?
Toledo gets one of those predicted 9 wins in their 2nd game.
Nice job, Mathlete - love this analysis.
Just curious-- do you have an advanced degree in math or a math-related field? You seem to have your stuff together from all of your posts.
So is it just interest, or do you work with numbers for a living?
Indistinguishable between Michigan, Iowa, and Michigan State. I'll take that for a coaching transition year. That seems to fit with the concensus pick of 8 wins this year.
I think we all have some fear as we are transitioning and will accept an 8-4 season. I hope we get to look back on this season as one where we found our footing and started competing for the BigTen Championship every year.
for using "meteoric", or some facsimile thereof, twice in the same post.
Nice...now we dont even have to play the season...can we sim the season to see the results?
quick reset is required. ("No, no, no, I am NOT getting upset in Week 4...")
Did I miss something? Notre Dame is now in the Big 10?
I am assuming that he included them because they play against three B1G teams each season.
With their schedule, are they not also considered under rated?
given that no matter who the starter is now, he has no quality experience at the position. Remembering what it was like to be a scared rabbit doesn't seem like it should be an advantage.
Nice breakdown. Think 8 wins makes sense, and I agree that the bunching at the top of the conference should make the final champion harder to distinguish early on than in years past. I actually think OSU is going to suffer more than your numbers maybe presume with the loss of Pryor and the rest of the players for the first 5 games. They caught a break with Miami having eligiblity issues, but that should be a close game against a talented team that might be playing with even more FU potential than OSU. MSU should also benefit from a still-reeling OSU in their matchup. Plus, OSU's offense was so reliant on Pryor the past couple of years that whomever they trot out there will be a downgrade and may stall the offense even more. I actually see them struggling to get to 9 wins, maybe closer to 8.
Still, great breakdown. I am not sold on WVU winning the Big East just because they seem very schizophrenic when given high preseason expectations, but most of the other conference winners sound about right.
I think we end up better than Penn State. I also think Northwestern and Illinois should get a little more love than what they are getting but I understand why they aren't yet. Other than that this all looks reasonable to me.
But now i am beginning to understand why it had relevance in my life. thank you mathlete. Although I see the glass half empty, with losses at Evanston, E. Lansing, Iowa City, at home to ND, NB and OSU. People forget that Tressel went 7-5 his first year with the same kids (except some freshman name Mo) that won the NC a year later. Rasta.
Interesting that you see an effect for defensive starters but not offensive starters (except QB). Or that a returning QB matters about as much as any returning defender.
Just curious: when you say you regress halfway to the mean, is that because you've modeled and tested that halfway leads to the best prediction, or because it just makes sense?
In any case, I'm always impressed with the thoroughness and thoughtfulness of the analysis!
ND, MSU/Corn, SuperCorn, and OSU?
that we will beat one in that group of four. And that we'll lose to one in the group of @Northwestern, @Iowa or @Illinois.
What kind of a countdown clock to you put up in 2012 if you go 0-for-clocks in 2011?
And if our SoS is #12 in a year where everyone is calling it "easy", maybe we shouldn't base our success estimations on schedule, because it's always going to be hard.
will not face 3 11-1 teams like last year. Big 10 is expected to be down big time. Michigan face creampuff teams like WMU, EMU. SDSU and ND are solid teams but beatable. Minnesota and Purdue are very winnable games. That's 6 wins right there. Northwestern is a wildcard due to them exceeding expectations for the last 3 years. Iowa is down and is a winnable game with loss of several key players like Stanzi, DKJ, Claybourn, Sash, etc. Illinois lost key players like LeShoure, Wilson, Liguet so it's a toss up at best but winnable. MSU is a very solid team and probably is the favorite to win the matchup. Neb, OSU(even without Pryor) and Wisconsin are all 3 games that I would consider a loss.
8 games is a minimum expectation for me and 9 wins is realistic. With tons of returning starters plus Woolfolk, Michigan should improve.
...Air Force is underrated...
God Bless my Falcons.
Let's go AF!
These predictions ..Lets just see how the season plays out..I think Michigan,much like the Lions are gonna surprise some people this year....Ah yes Tigers,Lions,Wolverines,Red Wings...Its a great time to be a Michigan fan........GO BLUE.......