A prediction on the 2009 season using a solid and proven mathematical model*

Submitted by OAMF90 on

A prediction on the 2009 season using a solid and proven mathematical model*

I believe that the two things in college football that will most dictate how many games you win are your quarterback and your schedule.  There has been much written about the quarterback situation and I can’t add anything to that.  I have seen little if anything, however, that focuses on the schedule. 

Considering that Michigan plays 8 of 12 games, including the first four, at home, the schedule is as bad as it can be.  If you break the games into three categories, likely win, tossup, and likely loss - 3 of the four tossup games are on the road where a team with a freshman quarterback is not as likely to do as well. 

I grabbed as many polls as I could find (the ones that rank all 120 teams)**, recorded the rankings for Michigan and their opponents, dropped the high and low ranks, and averaged them.  

There are several truisms concerning forecasts.  One is that they are always wrong and the other is that when you aggregate forecasts the result is more accurate than the individual ones.   Ask 100 people to guess how many jelly beans are in a jar and the average of all the guesses will be better than 90% of the individual guesses.  Consolidate a bunch of moronic preseason college football polls and just maybe the result is something approximating intelligence.

Only the Sagarin poll ranked Delaware and he had them at 116 so I just used that as it seemed about right.  I figured they wouldn’t be the worst team in division 1 but probably close to it. 

I then created a simple model, starting every game as a 55-45 proposition in favor of the home team.  I added/subtracted a point based on the difference in ranking,  so a home team ranked #20 would be given a 75% chance of beating a team ranked #40 (55 + (40-20)).  Given upsets in recent years, I topped it out at a 95% chance of winning or losing.

In the polls i used Michigan was ranked as follows; 35, 38, 41, 46, 47, 53, 55, 63, 77.  Dropping the high and low the average is 49.  The opponents average rankings are***:

WMU 73, ND 28, EMU 113, IU 91, MSU 29, Iowa 23

Del 116, PSU 11, Illinois 39, Purdue 76, Wisconsin 42, OSU 8

 

Based on those numbers here is a breakdown of what my simple model produces as an estimate of Michigan’s chances to win each of the games on their schedule. 

WMU                     .79            likely win

ND                        .34            tossup

EMU                     .95            likely win

IU                         .95            likely win

MSU                     .25            tossup

Iowa                     .19            likely loss

Del                       .95            likely win

PSU                     .17            likely loss

Illiinois                  .34            tossup

Purduw                 .82            likely win

Wisc                     .38            tossup

OSU                     .14            likely loss

Total expected wins = 6.28            (sum of all percentages)

 For this exercise I consider a likely win one of 75% or greater, tossup 25-75%, and likely loss less than 25%.

There are five likely wins (all at home), three likely losses (two at home), and four tossups (one at home).  I would feel much better if the Big Ten Schedule was reversed, giving Michigan their road games at home and their home games on the road.  Michigan would still likely win at PU and IU.  Still likely lose at PSU and OSU, but have a better chance of splitting ILL/Iowa/Wis/MSU and ending with 7 or 8 wins.  Michigan has to win one of the tossup games to get to the 6 wins my model predicts, and they will likely be the underdog in all of them.

The first two games should tell us everything we need to know.  If M starts out 2-0 then an 8 or 9 win season becomes possible as the Illinois and Wisconsin games would look much more favorable.   A 1-1 start points to 5 or 6 wins as it means we have lost our first tossup game and our ranking around 50 is probably correct.  0-2 means a season of 3 or 4 wins is coming as we have dropped one of our 5 likely wins and all the tossups start looking like losses.   

The silver lining?  Next year our quarterbacks will have a year of experience, and the Big Ten schedule is reversed, giving Michigan 2 winnable road games at Purdue and IU and 7 very winnable home games, leading to hopefully a 9+ win season.  If Michigan can go from 3 to 9 wins over RRs first 3 seasons I would be pretty happy.  Of course having lived through a two year quarterback battle between Demetrius Brown**** and Michael Taylor my happiness level may differ from your own.

Fast forward to 2012 when our quarterbacks are seniors and every player on the team is a RR recruit and…well…let’s not go there just yet.  Last time I did that was in 2004.  Looking ahead to 2007 when Henne and Hart would be seniors and Michigan would be in the BCS Championship Game….

 

 

*Model used was neither solid nor proven.

**the usual suspects, SI, Athlon, Sagarin, Rivals, Lindy’s, Sporting news, etc.

***If you are interested OSU (ranking range #5-#10) and PSU (range of #7-#15) had the closest consensus across the polls.  The widest range of rankings fell to Illinois (ranked as high as #17 and as low as #63) and WMU (#34-#89).  

****Demetrius Brown is awesome

Comments

nmwolverine

August 27th, 2009 at 11:43 PM ^

In July I predicted 5 wins based on an improved offense but an inadequate defense. I now predict 10 wins because the Big Ten sucks. I think we will play well, although not to Michigan standards. But it will be enough to pull out 10 wins. Then we will lose the bowl game. Because the Big Ten sucks. Then I will complain about another bowl loss.

Joined: 01/22/2009

MGoPoints: 1112

Enjoy Life

August 28th, 2009 at 12:00 AM ^

Perhaps you need to consult Nate Silver and his PECOTA system. (Which, yes, I know related to baseball -- Player Empirical Comparison and Optimization Test Algorithm). I think you just concluded that the higher ranked team is likely to win. Also, the rankings change significantly during the year, so the predictive validity is suspect (similar to Silver trying to predict an election based on a single point about 4 months before the actual election). My Weegee board says 7-5.

Joined: 07/02/2008

MGoPoints: 2658

Seth9

August 29th, 2009 at 10:16 AM ^

What on Earth does PECOTA have to do with this? First of all, PECOTA is designed for individuals, not teams. Furthermore, PECOTA was developed for a system with a lot more recorded data than college football. This has allowed statisticians to observe a lot of trends that enabled them to develop PECOTA. With respect to Silver, your comment makes no sense, not least to say that Silver predicted that Obama would win four months before the actual election. So, out of curiosity, what exactly is your point here?

Joined: 04/01/2009

MGoPoints: 2493

Irish

August 28th, 2009 at 12:08 AM ^

but you still have the problem of it all being based off of preseason rankings which mean nothing. Now if you could some how go into the future after each team had played 12 or so games and then ranked them and then brought that back into the present and used those numbers you might have something. I do applaud the effort, but all of those rankings really come down to guessing and even 100 guesses at a jar of jelly beans won't guarantee the correct answer, not that you would know the guess was correct till the jar was busted open in the end anyway.

Joined: 06/05/2009

MGoPoints: 3696

Mgoscottie

August 28th, 2009 at 11:54 AM ^

I say Iowa should be the toughest to win of all the games, maybe ohio state will be, but the timing of the season I don't have a lot of hope for the Iowa game. I don't see a lot of difference between purdue and indiana either and think both should be around 85ish. I think Wisconsin will be the easiest road game of the four and we'll be a favorite in that game. How quickly people forget how crappy they used to be and probably are heading back towards. There is very little talent on Wisconsin and not a lot of experience or great coaching. I do agree that Illinois could be one of our toughest games this year.

Joined: 08/25/2009

MGoPoints: 16363

psychomatt

August 29th, 2009 at 6:59 AM ^

College football is all about matchups. You can be a great team but just run into a team that you do not match up well with (e.g. Oklahoma v Boise State in Fiesta Bowl). (1) I agree with you that Wisconsin is not the same quality team they were three years ago. They have been slipping steadily on defense and have major questions this year on offense, including at QB. Nonetheless, they have a powerful offensive line, John Clay is a monster (if he turns out to be the starter) and we play them at Camp Randall, so I just do not have a good feel for this game. If this game was in A2, I probably would have given Michigan the slight nod. But being away, I left it at 50%. (2) Purdue and Indiana are no way the same. The Hoosiers are a mess. At least Purdue has some athletes. These two teams will almost certainly be the worst in the conference this year, but IU is significantly less likely to beat Michigan than Purdue. (3) As for Iowa, they should have a very good defense again this year (possibly better than last year), but their ability to put up points is questionable. Greene is gone and they are struggling to find a healthy RB to step into the position. Ricky Stanzi is a good, experienced QB, but at their heart they are a running team and without a strong tailback they are going to struggle. Michigan, on the other hand, should be much more effective on offense this year. They showed sparks of a high octane offense in the Purdue game last year and have had another 10 months for everyone to learn RR's system and work with Barwis. They also have quarterbacks that actually fit the system this year. I will be shocked if Michigan is not able to put up 5-10 more pts per game this year than last year. This is, of course, dependent on their finding someone who can field a punt, but this is not an issue limited to the Iowa game. The question is whether Michigan will have the defense to stop (slow down?) Iowa's offense and keep this game within reach. Since I do not see Iowa's offense being a powerhouse, I think this will be a very close game. The one caveat: it is a night game in Iowa City (not good). Thanks for your response though. We will know a lot more in a week!

Joined: 08/25/2009

MGoPoints: 3425

chally

August 28th, 2009 at 12:12 AM ^

It's the sort of rough math statistical nonsense that I like to do myself when bored. Actually, the only problem I have with it is the assumption that 1 place difference in ranking equals a 1% increase in chance to win. If it were me, I'd look to the average margin of victory per ranking difference among top 25 matchups over the last 5 years, then add 7 points for homefield advantage.

Joined: 11/25/2008

MGoPoints: 510

chally

August 28th, 2009 at 12:55 AM ^

So I was bored and a I did it. It turns out that on average a team wins by one more point for each place ranked above their opponent. However, the range of outcomes was far too random to meaningfully predictive for any game. If you're interested, by this analysis, since all of our away opponents are ranked higher than us, we're supposed to lose each of those games. Even with home-field we also lose to ND, PSU, and OSU. That's 5 wins.

Joined: 11/25/2008

MGoPoints: 510

Michael

August 28th, 2009 at 12:24 AM ^

In many ways, the prediction of a football team's season is like social science research. There is no potential for an experiment, because 1) football players are human and 2) each season is unique. (there are MANY more reasons I could add) While past models are relevant, their contribution to a true statistical analysis is limited. Economists and political scientists do not have the luxury of experiments - only theory which may or may not be based on quantitative analysis. What I am arguing is that attempts to predict a football season based on an quickly-assembled model are futile. Any model which purports to be predictive in this realm is at best unreliable. In this case, there is no indication that the preseason predictions used as data are reliable. The causal relationship is simply not adequate. Any model which purports to be predictive is at most somewhat reliable. I appreciate the efforts of our fanbase to make predictions based on "evidence," but it is simply not possible. Each season of college football is unique, as is each team at a given school for that season. There are simply too many independent variables to accurately predict an outcome (the dependent variable) based on data which is dubious. Now, what I see happening in many of these models is a qualitative theory seeking data which supports it, but does a terrible job at finding said data. This is more of a function of the dynamicism of college football as opposed to the ineptitude of the author. Data is so unreliable in the context of college football, it's almost ridiculous. I'm sure I'm going to get neg-banged because I'm pretentious (or whatever), but if someone claims to have a model then it is my duty as an amateur social scientist to point out its failures with the hope of advancing the discussion to a higher level. I'd like to get into more detail, but I will wait for other responses to escalate the conversation.

Joined: 06/30/2008

MGoPoints: 1623

MCalibur

August 28th, 2009 at 1:14 AM ^

Mike I hear you, but you yourself are making vast assumptions that seem to makes sense but aren't necessarily true. "Each season of college football is unique, as is each team at a given school for that season. There are simply too many independent variables to accurately predict an outcome (the dependent variable) based on data which is dubious." According to who? You? Who the heck are you? Florida returns 18 of 22 starters as well as a bunch of the 2-deep. You're really going to tell me that people can't reasonably say that they will be very, very good again this year? Again, I'm inclined to agree with you, but an opinion eloquently stated as fact is still an opinion. There are folks who are very, very good at doing this stuff who can explain most of the variation in the changes in a given team's success from year to year-- the folks at Football Outsiders. They project Meechigan to be third in the Big Ten at 8-4 (5-3) with the following disclaimer:
The correlation of mean projected wins to actual wins is 0.69 for all games, 0.61 for conference games.
So, a significant amount of the outcome can be predicted if you know what you're doing and you do it right. That said, I do not think that OAMF90 did it right...I'll reserve my judgement on wether or not he knows what he's doing. I'm looking for 7 wins with a solid chance at 8, not including the bowl game. I love that people are sleeping on Michigan this year. Bitch ass numbskulls...

Joined: 07/16/2009

MGoPoints: 2361

victors2000

August 29th, 2009 at 9:39 AM ^

I think his assessment of variables is true but he doesn't take into consideration the weight of the variables as well the length of time they are valid, as your point about the players returning for Florida relates. Still I like the attempt; just needs a little tweaking.

Joined: 03/21/2009

MGoPoints: 12906

Ernis

August 28th, 2009 at 1:22 AM ^

The value of a model is dependent on the philosophical and objective assumptions underlying the research. I have issue with several in the OP: I believe that the two things in college football that will most dictate how many games you win are your quarterback and your schedule. Philosophically, the premise that there is a meaningful distinction between two things amongst many which dictates a team's win count just can't be backed up. And in this case, no attempt has been made. Objectively, these two particular choices are dubious to begin with --although intuitively appealing to many, I'm sure-- and are also not supported by evidence. Objectively, again, the use of post-season rankings would be preferable to preseason rankings, since the former are reflective of a team's accomplishments that season. etc. But, as the author so eloquently stated, "There are several truisms concerning forecasts. One is that they are always wrong..." So it's all in good fun, and I think it was fun to read. The assumptions are laid out clearly and there should be no illusion of predictive accuracy. PS- one assumption with which I agree: the prediction of a football team's season is like social science research ... because football players are human human = social /win

Joined: 09/23/2008

MGoPoints: 11453

raleighwood

August 28th, 2009 at 12:22 AM ^

This is one of the better posts that I've read. Seriously. No bias, no BS.....just one person's interpretation of the situation. Very well done. HOWEVA.....I'm getting a little more excited about Michigan's QB play. Michigan had five games last year that could have gone either way (Utah, Wisconsin, Toledo, Purdue and Northwestern). They were 1-4 in those games. You can draw two conclusions from this. Michigan's sucked as a team. RR sucked as a coach. The truth probably lies somewhere in the middle. I get the feeling that this team (and coach) is a bit different. The offense HAS to be better. It looks like Sheridan (who should be improved over last year) is third in line. That means that QB play will be better this year. The defense, although more experienced last year, was the worst in the history of the program. It HAS to be better. In my mind, Michigan goes at least 3-2 in close games this year. Maybe 4-1. Throw in the gimmes against EMU, IU and Delaware State, along with the "probable" against WMU....and Michigan should be at least 7-5 with an upside of 8-4. Remember, the QB's that we see in September will be much more experienced in mid-October. It's quite possible that Michigan will be the third or fourth best team in the Big Ten by the end of the season. It's all of matter of how they handle those first tests of the season (WMU, ND and MSU).

Joined: 11/26/2008

MGoPoints: 4710

formerlyanonymous

August 28th, 2009 at 12:51 AM ^

This pretty much describes exactly my outlook on the season. My personal thought was the WMU game is a bit closer than the .79 and PSU/Illinois I thought might be a little too much in our favor. Up until just a few days ago I thought we were a 5-7 team with a 50-50 shot of 6-6. The more I look at WMU, the better I'm feeling about it. This statistical outlook makes me feel at least partially justified.

Joined: 06/30/2008

MGoPoints: 38964

mjv

August 28th, 2009 at 12:51 AM ^

I can't remember anything after I read "****Demetrius Brown is awesome". Please tell me that my sarcasm meter getting pegged after reading that is appropriate? just kidding. nice post.

Joined: 06/30/2008

MGoPoints: 3656

Muttley

August 28th, 2009 at 1:32 AM ^

He threw game winning, late fourth quarter TD passes to beat tOSU on the road and Alabama in the Hall of Fame. Also QB'd Mich to the 22-14 win over USC in the Rose. (OK, that Hoard guy had something to do with it.) And yes, he threw for several hundred interceptions against MSU

Joined: 07/07/2009

MGoPoints: -73245

Beegs

August 28th, 2009 at 9:16 AM ^

In those years mentioned (Alabama and OSU games) Mr. Brown also had a gentleman by the name of Jamie Morris at his side running behind a great O-line. I have a mental block getting past the 7 interception game in East Lansing. After that, Brown was dead to me.

Joined: 08/17/2009

MGoPoints: 103

harmon40

August 28th, 2009 at 6:10 PM ^

After all, how many Michigan QB's in the last 40 years can say they started a Rose Bowl and won it? 4: Griese, Grbac, Brown, Wangler. How many Michigan QB's in that span can say they won 2 bowl games? Henson, Collins, Grback, Brown. The game-winning TD against OSU in the Big House in '88 (mentioned in another post) capped one of the most dramatic Michigan combacks of all time. OSU scored with a minute left, Kolesar ran the kick to mid-field, then Brown found him on a spectacular long pass for the win. Also, his first bowl win, against Alabama, he threw the game-winner to -guess who- Kolesar in the back of the end zone. Of course, I can't think of any other Michigan QB's who can claim 16 picks in a season...

Joined: 05/02/2009

MGoPoints: 5743

steve sharik

August 28th, 2009 at 11:41 PM ^

...D Brown still had a year of eligibility remaining but blew it by becoming academically ineligible. He likely would've started the '89 season (especially b/c Taylor was hurt) and we could have beaten ND despite the Rocket show. I still believe if D Brown hadn't messed that up that we were looking at very possible double '89 NCAA titles in hoops and then football.

Joined: 08/08/2009

MGoPoints: 13860

mjv

August 28th, 2009 at 10:42 AM ^

D. Brown was a high variance guy. I remember the OSU and Alabama games pretty well. But the lasting memory I have of his career is the State game. The guy did have a strong arm, but Taylor was efficient and took care of the ball. Taylor is probably one of the (if not the) most underrated UofM QBs of the last 30 years. If not for his injuries, he would probably be better remembered.

Joined: 06/30/2008

MGoPoints: 3656

restive neb

August 28th, 2009 at 1:32 AM ^

but Michigan plays Delaware State, not Delaware, and Delaware State, as a 1-AA team (FCS), would not have a rank in 1-A (FBS). It doesn't change the fact that it's a "likely win."

Joined: 04/06/2009

MGoPoints: 1123

Muttley

August 28th, 2009 at 2:02 AM ^

Total = 6.68 (8-4 with the maize-colored goggles on)
Foe Your% My Points My %
WMU .79 -12 .80
ND .34 +3 .40
EMU .95-25.95
IU .95 -15.88
@MSU .25+6 .35
@Iowa .19 +7.30
DSU .95 -25.95
PSU .17 +8.25
@Ill .34+6.35
PU .82-12.80
@Wisc.38+3.40
tOSU.14+8.25

Joined: 07/07/2009

MGoPoints: -73245

The King of Belch

August 28th, 2009 at 4:56 AM ^

How one guy can be so frigging bored that he wastes half his day crunching all these numbers and two things happen: One: He doesn't tell us if he used past seasons to see how accurate it is. For instance: did it come close to predicting 3-9? Also: How these types of threads deteriorate into an even more boring pocket protector measuring contest between nerds who try to prove how mind-numbingly smart they are.

Joined: 07/21/2009

MGoPoints: 8

uniqenam

August 28th, 2009 at 8:17 AM ^

OAMF90 put some pretty hard work into this diary, and I (like many MGoBloggers) like to see a statistical breakdown of everything. If you want to knock someone's diary, do it because it was lazy or short; not because they put a ton of work into it.

Joined: 07/20/2009

MGoPoints: 4542

OAMF90

August 28th, 2009 at 9:14 PM ^

but about 10 minutes in Excel. the model was arbitrary, the inputs were opinion, the calculations were simple, and the tone was jovial. which was all kind of the point. In reality the entire post was an elaborate smokescreen contrived to get Demetrius Brown the mgoblog love that he so richly deserves. I believe mission accomplished on that score

Joined: 08/27/2009

MGoPoints: 2

David F

August 29th, 2009 at 5:16 PM ^

Way to contribute to the dialogue, ass. And why would predicting one season (last year) correctly make a difference? A sample size of one does not evidence make. So just stop.

Joined: 07/01/2008

MGoPoints: 414

Moe Greene

August 28th, 2009 at 6:47 AM ^

Are there many variables? Yep. BUT we can be have a bit more confidence about our results through the use of nonparametric matching techniques - which try to get cases that are identical to the one that we want to study. This is why many folks compare this season to RR year twos at other programs - they're essentially trying to hold one factor constant.

Joined: 08/01/2008

MGoPoints: 2654

steve sharik

August 28th, 2009 at 7:55 AM ^

...for teams that we know will be really, really good (e.g., Florida) b/c they were good the previous season and have almost everyone returning, or for teams that we know will suck b/c they were terrible the previous season and have nobody returning. And how many times has Michigan been ranked top 5 preseason and disappointed? (I don't feel like looking this up.) Almost every time, with the exception of maybe 1986. Historically and generally, when the public has had high expectations for Michigan, we underachieve. When the public has had low expectations for Michigan, we overachieve. I really believe this team is headed for overachievement, to the tune of 8 or more wins. I have no data to back this up, and I don't care.

Joined: 08/08/2009

MGoPoints: 13860

ZooWolverine

August 28th, 2009 at 8:51 AM ^

Teams that are ranked highly are almost always going to underachieve: if you're ranked in the top 5 you have to have a nearly perfect season to stay up there and that doesn't happen all that frequently, no matter how good you are going into the season. Take 2004, for example, Michigan has a 9-3 season, losing only to ND, OSU, and USC. We were even in the talk for the national championship game, although that talk wasn't particularly realistic. And yet, since we were ranked preseason numbers 8 and 7 (AP and Coaches, respectively), we underachieved since we ended up numbers 14 and 12.

Joined: 02/27/2009

MGoPoints: 3719

OSUMC Wolverine

August 28th, 2009 at 8:36 AM ^

I have complete faith in RR to use the lightning fast waterbugs he now has combined with a qb that appears to be accurate on the short routes. Hitting a waterbug in stride is what will make this offense click and when the lbs start playing the short stuff the running game will really take off. Staying healthy we could have a great season. And even better, when we open the season 4-0, we will be top 12 or so and reclaim our generally overrated early season ranking! :)

Joined: 09/06/2008

MGoPoints: 4403

oakapple

August 28th, 2009 at 4:42 PM ^

At the end of the day, you arrive at the probabilities that most people would have landed on, though you are too generous with the "toss-up" label, using it for any win percentage between 25% and 75%. In every one of the so-called toss-ups, it appears that Michigan will be a clear underdog. The most probable of these, according to your model, is Wisconsin, where the Wolverines haven't won in something like a decade. The five games in the "likely win" category will tell the tale. If Michigan beats these teams easily, it will probably have enough talent in reserve to win a couple more along the way. If Michgan loses to, or only just barely beats, any of the "likely win" teams, it probably signals a tough year when stiffer opposition comes along.

Joined: 02/06/2009

MGoPoints: 2130

Rasmus

August 28th, 2009 at 11:04 AM ^

But I'd be more comfortable with a bit more nuance in the middle. For me, "tossup" means something more like between 40% and 60%. Let's go with "winnable" (meaning a win is possible, but not likely) for between 25% and 40% and "losable" (meaning a loss is possible, but not likely) for between 60% and 75%. What I like most about your approach is that is underlines the fact that Michigan is not a preseason favorite in any of the middle games. There are no losable games, although WMU is close, nor any tossups (by my definition). Only winnable games. This is a tough schedule in terms of preseason rankings. I'm not sure what this means. It might mean that the pundits are not giving U-M enough credit for returning basically 10 of 11 starters on offense along with some of their best players on defense, and those "winnable" games are actually more like tossups. Or it might mean that the Big Ten has a lot of good teams this year, Michigan plays all of them, and it's going to be a long season in terms of wins and losses. [EDIT: Maybe "Likely close loss" and "Likely close win" would be clearer, making the range of categories as follows: Likely loss, Likely close loss, Tossup, Likely close win, Likely win. But winnable and losable is more fun, and somehow better expresses the uncertainty inherent in this.]

Joined: 07/02/2009

MGoPoints: 2412

MMB 82

August 28th, 2009 at 10:15 AM ^

when Harbaugh broke his arm in the MSU game and the season became a relative disaster at 6-6, including a loss to BYU in the Meh Bowl. The next fall SI predicted UM wouldn't do much better, so of course the 10-1-1 1985 team came the closest to a MNC than any other Bo team. Maybe I am a redundant optimistic homer redundancy, but I can't help but think IF (IFF!!!) we get good QB play this year we will surprise quite a few people- even OSU who historically have problems (suck) vs spread teams. But then I come to my senses and just hope we make a bowl this year (Insight- then I can just drive down to Tempe for the game).

Joined: 07/06/2008

MGoPoints: 18691

Durham Blue

August 28th, 2009 at 11:03 AM ^

Michigan is going to win all the "likely win" games. Michigan is going to upset two out of these three teams: ND, MSU, Iowa (I don't think any of these teams is nearly as good as advertised by the media). Michigan is going to lose to Illinois, Wisconsin and OSU (two really tough road games, revenge for Wisconsin and ugh....OSU). PSU will be really tough. Michigan will lose but it will be a really close game. Final record: 7-5.

Joined: 10/09/2008

MGoPoints: 30004

jericho

August 28th, 2009 at 9:53 PM ^

"I believe that the two things in college football that will most dictate how many games you win are your quarterback and your schedule. " Shows what you know. The biggest factor determining whether Michigan wins or not is my superstitions. If I'm wearing my lucky home jersey, our win rate is 93%. My away jersey is not nearly as lucky at a 87% win rate. If I serve green chili stew before the game, we win 99% of the time. You all should be begging for green chili stew next Saturday. Note: All win percentages exclude the 2008 season which never really happened in my world.

Joined: 07/04/2009

MGoPoints: 2632

summit595

August 29th, 2009 at 11:21 AM ^

This is a really cool analysis. Honestly one of the best thought out predictions I've personally ever seen. I'd be really interested to see what this formula predicts a few games into the season when we find out 1) how good our qb's really are gonna be and 2) how overrated ND really is. One point I slightly disagree with though which has nothing to do with your calculations - I don't think going from 3-9 to 9-3 under RRod is exactly what the fan base is looking for. Let's be completely honest losing to OSU 5 times in a row is what we're hurting about. Now being good enough to beat OSU means you get more wins, yes, but I really think he has to beat OSU sometime within the next couple years for the fan base to have any real satisfaction.

Joined: 05/08/2009

MGoPoints: 14