Football Outsiders projects Michigan to finish 7-5 in 2013
Football Outsiders did a preliminary projection and simulation of the 2013 Big Ten Standings today. They projected Michigan to most likely finish 7-5 and 4th in the Legends Division with a potential win range between 5-7 and 9-3. I was pretty surprised by this as usually Football Outsiders models are some of the better ones in football. Article is ESPN Insider ($):
4. Michigan Wolverines
Projected finish: 7-5
Win total range: 5-7 to 9-3
Chance to win Big Ten: 1 percentThis projection model isn't too kind to the Wolverines, a program with an elite historical tradition but one that is still seeking its identity in the Brady Hoke era. The offense was prolific in spots under quarterback Devin Gardner after he took the reins midseason, but there is youth along the offensive line that will probably keep Michigan from being a contender. If the Wolverines can help themselves on defense and special teams, it can ease the burden. But Michigan started twice as many drives from inside its own 20-yard line (24 percent) as its opponents (12 percent) and lost the field position battle eight times, including in four of its five losses.
December 3rd, 2013 at 12:22 PM ^
Well we went 2-4 against those 6 "Challenging games" plus the loss to PSU that I didn't think was "challenging". This team fell far short of expectations.
I'll go ahead and disagree with this analysis.
Our OL will be better with Miller at center and Kalis in at guard, our WR group will be better with Darboh, Dilleo and Gallon, with some Cheson mixed in, the Funch is a year older, Devin is a better QB than Denard, and Green will be better odds are than anything we had running last year.
On defense, yeah, we lost Ryan, but we are getting Countess back. And Ryan may come back in October.
I can easily see us winning 9 or 10 games this year.
I will say though, that 2014 will be the year we run the table.
Here's someone (their Twitter handle is @HuskerMath) that did this slightly differently but arrived at some similar results to the ESPN article (HERE). In this model, they also had Michigan at 7-5, but this one also displays some of the probability tables used and explains the method in at least some detail. Like the ESPN one, it is based on past performance.
Others have said it, but I also agree that 7-5 with our schedule seems like the worst case, one where perhaps there are numerous injuries. Hoke and company have built a well-rounded, deep team (at several positions where depth was needed to boot) and I think this year, they will be able to do some of the things they've wanted to do stylewise since they arrived and move closer to what the long-term plan was anyway.
Like the alternate model I linked here, it seems like anything which is based on past results that pulls from scoring offense and scoring defense might spawn interesting predictions for Michigan depending on how many years of data they used. Even if the defensive production stays at a similar level to last year between not having Ryan right away and other changeouts, I still see 7-5 as not terribly likely (many unfortunate things would need to happen, I think).
Both the Football Study Hall and Football Outsider's models use past samples of 5 years. Both have 3 of those 5 years being the Rich Rod years. I think UM may be one of those 'exception' teams where the sample data doesn't do nearly as good of a job predicting performance as other team's.
we ended up going 11-2 and winning a BCS game........
disagree with this. Not trying to be a homer here either, i think our team is deep and solid.
I would like to hear what Brian has to say about this prediction.
For those wondering, in addition to the usual returning starters, what positions, etc that most of these models use, Football Outsiders also uses a 5 year program 'rating' using all the stats from the past 5 years to predict upcoming success. So bascially, right now that includes the seasons 2008-2012, aka the lowest of the low for the UM program. If you look at their tables, we currently have a program efficiency rating of 29, which is one ahead of Iowa and 5 behind MSU. The lowest point we had was after the 2010 season, when we were 50 in the nation. What does this mean? Not entirely sure, but eyeball tests suggest we are out performing our program efficency rating.
LOL. If that's truly the case, then their predictions for Michigan are worthless.
I am all the sudden feeling good about this prediction of 7-5
Feel free to be bearish about Michigan's program efficiency. The returns on the 2010 coaching change investment appear to be respectable thus far for the entire program, holistically. Michigan does not yet possess enough continuity with the current coaches for a homerish 12-0 but I see 10 wins as a reasonable expectation for the upcoming season if CG steps up in place of JR and DG flies high. Say the homefield seal breaks and we lose to ND and Ohio. We also play at a decimated Michigan State; and face Nebraska at home. I saved my ticket from the 2011 game: Section 21, Row 11, Seat 21.
In short: I believe Michigan is most likely to lose to ND and/or Ohio next year but I see all games on the schedule as essentially winnable based on the incomplete information available to fans.
Well then that's a fairly easy prediction to make... add 3+5+7+11+8 as the number of wins we've had the last five years =34, divide them into 5 seasons and we have!.... 6.8 if we round that up it'll be 7 and I'm being optimistic, who wants to hire me to be an analist?
Si if MICH goes 7-5, that means Les Miles, right?
but I think there will be a lot of tracking flights in December!!
how good DG can be. I think he is going to go nutz this year.
No Way we lose 5 games this year. I predict an undefeated season. Our defense is going to be so much better and our offense will be at least as good. I think having the brawlers up front in the middle of the line are going to be at least as good as last year, even with their youth. They are definitely going to be able to run the ball more and passing is improved exponentially with Devin and some big receivers. 7-5? I really don't think so.
CMU W
ND W (could have easily won this last year without an inexplicable 6 turnovers)
Akron W
@UConn W
Minn W
@PSU W
Ind W
@MSU W (defense destroys MSU's weak O and exposes their overrated 2013 D)
Neb W (last year was a fluke because of denard's injury. We are definitely winning this)
@NW W
@Iowa W
Ohio W (Ohio overrated and cannot sustain their playing above themselves last year out of anger and spite and also getting really lucky several times)
This year we are the best team in the B1G. the Hoke Factor comes to full fruition in year 3.
It's easy to see why this model predicts a bad season for Michigan:
- It has a five-year look-back period to determine program strength, so it includes all three years of the RR era, which makes us look like a mediocre program that just had a lucky season in 2011 rather than a historically good program that had a 3-year deviation.
- One of its primary variables is whether we have a returning starter at QB. We don't, but Gardner is a top recruit who showed that he's a very good QB in his limited time last year.
- Another variable is returning starters. What the model doesn't realize is that nearly all the guys we lost (Omameh, Barnum, Campbell, Vincent Smith, Roundtree, Floyd, Campbell, Roh) were average at best - Denard and Kovacs were the only two stars we lost. And almost all those guys are being replaced by highly-regarded recruits, some of whom (Gardner) fit the system better than they guy they're replacing.
So, in other words, don't look at a single ridiculous prediction (I don't think any college football analyst is predicting us to be as bad as 7-5) and say the model sucks. This model actually has a pretty good track record at predicting how strong most teams will be. But if you look at why the model says we're not supposed to do well, you'll see reasons why our particular situation doesn't fit the model well.
if we went 7-5 this year. While the interior OL is rebuilt, what we had last year wasn't so great and Kalis and Braden come with high praise. Mattison has shown what he can do with spare parts on defense, and he has a lot more to work with this year. I think Devin is going to be terrific. One of the RBs will be good enough. Anything less than 9-3 would surprise me.
Will go 10 bucks to keep it friendly to anyone who wants to take it!
there are 8 easy wins there. 8-4 is the worst case scenario. Football Outsiders can suck it.
I'm semi-interested in who you consider an easy win, Northwestern or Michigan State? Two teams that we oh so narrowly beat at home last yar and will be facing on the road this season.
especially with the issues concerning the punting game. . .
...assuming that there hasn't been a head coaching change in the last five years that's completely turned the program around from recruiting to assistant coaches to the teaching of fundamentals.
Last year our 4 losses were away games, and even with our OLine struggles and bad turnovers we could've, or even should've, won 3 of them. This year with those opponents at home and what I see as improvements in our OLine and passing game we should be able to edge those opponents out.
That's going to be a really big mojito!
How many games was the 97 team supposed to win? Just sayin'.
I don't think this prediction reflects how good Devin Gardner is going to be. The secondary is deep, as is the offensive line. The offensive line has young depth, but that young depth is big, talented and highly rated. Also, never, never, never discount the effect an impact RB can have on a football team. Maurice Clarrett's impact on the 03 Ohio team can't be underrated. Derrick Green will make an impact, and will make this team far more consistent.
The one thing really against us for the division crown is schedule. MSU's cross divison games are Purdue, Illinois, and Indiana. Nebraska's are Illinois, Purdue and PSU. We get PSU, OSU and Ind. So we play probably 3 of the 4 toughest legends division teams, one of which is by far the best. MSU plays one, the worst of the top 4 (Ind), and Neb plays 3 and 4 (PSU and Ind). Basically, our divisional games are CRAZY important this year. Need to beat both MSU and Neb to keep a cushion. At least NW has to play OSU and UW, so that gives a little leeway in the race against them.
I will go with my default pick of 9-3 with the bowl being a toss-up. I see no reason to think otherwise in either direction. Michigan is a baseline 9-3 school, and has been since Bo retired. Sometimes, they do better, sometimes, they do worse. But 9-3 seems to be the starting point more often than not.
As that they (a) often produce crazy results for individual data points despite being pretty good overall, (b) often lack any sort of "sanity check," and (c) suffer from the fact that you can only include so many variables before you risk overfitting the curve.
In other words, models have blind spots, and trying to make sure every data point is where you THINK it should be is a recipe for a terrible result. I happen to think Michigan will outperform this result comfortably, but you can't really be mad.
I forget which basketball model had Michigan at like #42 entering this year, and in the discussion portion they were all, "yeah, Michigan is almost certainly too low, but they're really young so that happens."
This isn't really true. When models get calibrated and validated any professional modelers will look at the outliers to try to determine what might be missing.
That doesn't mean models are perfect, but if by sanity check you mean they need a subjective adjustment factor - that's exactly the kind of thing Brian has been railing against in the computer formulas. No place for that.
For anecdotal evidence - people were arguing Kenpoms model needed to do the same thing for Wisconsin and they ended up a pretty good team. Models are worthless if they are just putting numbers to our opinions. Unexpected outcomes to question our thinking is precisely the point.
The offense has to be dragging down the numbers. After Denard got hurt, Michigan produced approximately zero yards on the ground last season, the whole interior line graduated, most of us believe our top two running backs in November haven't registered for classes yet, and Gallon is our only proven receiver. There's a lot not to like there.
Models can't realistically account for the less tangible reasons why most pundits and fans see Michigan as one of the two best teams in the conference this season. Modeling is really hard to do in a sport like college football due to high roster turnover and small sample size (among other things), so I tend to be skeptical of what modeling is done.
A 7-5 prediction doesn't even bother me as much as suggesting 9-3 as a best case scenario. Best case. Even a reasonable ohio fan (all three of them) would admit our best case is better than that. Our best case the last two years was better than that, and this team is better than they were.
Even if I hated Michigan there is no way they go 7-5. If a meteor hit and most of the team is vaporized, then maybe.
Because I will take that money to the bank
If we establish any kind of decent running game I don't see how this team doesn't win 10, 11 and maybe all 12 games during the regular season.
If the O line struggles this entire season ( like last year). Hoke will need to take a long look at his offensive line staff
4 years is enough time for any coach to make a mark, either positive or negative. I'd rather have a o line coach who can coach up players than someone who recruits 4-5 stars only to have them fail miserably. The jury is still out on Funk but, after watching our O line the last few years his coaching ability is suspect at best