I always enjoy when data backs up my optimism! Go Blue!!!
Mathlete’s Michigan Season Preview
Of all the Michigan season previews you read, this has a pretty good shot at being the nerdiest. I will be focusing the preview on the statistical contributions of the departed and the returning. The culmination will be a series of beautiful charts telling you just how many games Michigan is going to win this year (or at least how many they won in one of the thousands of simulations I ran). The key metric you need to know is PAN, Points Above Normal. Normal is technically the NCAA FBS average, but you could also think of it as BCS conference replacement level. A Normal team would generally place in the bottom 2-3 of any BCS conference in a given year or in the top 2-3 of a non-BCS conference in a given year. For a little more detail on how it works, there is a short primer at the end of the column.
On to the preview.
Tate Forcier took the majority of snaps last year and has the more robust data set from last year while Denard has shown vast improvement through spring and camp and is the rumored front-runner for the starting position. Here is what they did last year.
Forcier: Had a passing PAN of 37 points on 278 qualifying plays (QP). This was good for third in the Big 10 (behind Tolzien and Cousins) overall but was second behind Tolzien on a per play basis. Tate’s rushing PAN (which excludes sacks) was –5 on 90 carries, meaning in 2009 Tate was worth about 32 points above the average NCAA player, second only to Kirk Cousins of Michigan St in the Big 10 and ahead of the 27 points in 13 games posted by Terrelle Pryor.
Robinson: Very limited sample size, but with 27 QP Denard had a passing PAN of –1 point. His –.04 PAN per attempt would have ranked him second to last in the Big 10, ahead of only Ben Chappell at Indiana. There was little element of surprise when Robinson was on the field that the ball was staying on the ground. Despite that, Robinson posted a positive PAN (+1) on the season. At .01 PAN per play he was ninth in the conference of players with at least 50 QP and second among all quarterback on the ground, behind Terrelle Pryor’s .05.
Despite an unknown quarterback situation coming into the season, it reasons that a sophomore Forcier would exceed a freshman Forcier and if Robinson can beat him out, then the quarterback production from Michigan should be about the best in the Big 10 this season.
In the most wide open position battle on the team, Michigan is looking to replace departed seniors Brandon Minor and Carlos Brown. In injury limited time, Minor was still the 5th most valuable back in the Big 10 last year with over 6 PAN on only 95 QP. His .06 PAN per play was third in the Big 10 behind John Clay of Wisconsin and Jaycen Taylor of Purdue. The production of Brown will not be as difficult to replace. WIth 80 QP Brown had a –3 PAN for the season. Brown was more active in the passing game with 10 catches on the season.
Looking to fill the void will be two players who saw limited action last year.
Vincent Smith: Had 31 QP last year and a 0 PAN. Running backs tend to skew negative so a 0 PAN would have put him 10th in the Big 10 among qualifying backs.
Michael Shaw: 33 QP in 2009 and a –4 PAN. At –0.13 PAN per play, he would have been in the bottom quarter of Big 10 backs.
In limited time last year, Smith outperformed Shaw. With Smith missing the offseason due to injury and Shaw battling academic issues, it looks like several players will get their chance to prove themselves on the field.
Wide Receivers and Tight Ends
The good news is that almost everyone is back from last year. The bad news is that no one did much in 2009, Michigan did not have a single receiver in the top 15 for the season. Greg Mathews is lone departee and was second last year with 25 PAN (receiving PAN do not directly correspond to other PAN because there is no good way to offset the good plays, catches, with bad plays so all you end up with are the positive plays).
Roy Roundtree: Finished the year red hot out of the slot. His 27 PAN in the last four games of the season were more than any other Wolverine had for the full year. If he could have produced at that pace for all 11 FBS games last year, he would have been top 2 in the Big 10.
Martavious Odoms: Third on the team with a 21 PAN last season but looks to have lost out on the primary slot role due to Roundtree’s emergence.
Kevin Koger: Michigan’s main tight end was fourth in 2009 with a PAN of 16.
Junior Hemingway: Came out of the gates on fire with the team’s season high single game PAN of 11 against Western Michigan but only managed a 15 for the season.
Darryl Stonum: Provided more value as a kick returner than he did as receiver, his 2009 PAN was 9.
Based on his strong finish to last year, Roundtree is the only member of the receiving core who looks to have a strong role locked up. Whether it’s from a single individual or the group as a whole, Michigan’s wide receiver will need to do better than 2009’s 8th place ranking for receiver value.
So we lost one guy. He was OK. Brandon Graham exploits are well noted but just for the PAN refresher, defensive players are measured by the number of negative (for the offense) plays they make and their magnitude. Graham led all Big 10 players with 47 points taken away from the offense, a stat normally dominated by linebackers.
Ryan Van Bergen: Finished 12th in the Big 10 among D-lineman with 19 points taken on 25 plays made.
Mike Martin: Lacked the big play ability of some of his Big 10 counterparts with only 10 points taken away, but was second among all tackles with 30 plays made.
Despite losing Graham, the move to more three man lines and the strong performances by RVB and Mike Martin last year make this the most secure position group on the defense.
Michigan’s linebackers flat out did not make plays last year. Only Indiana and Illinois got less production from their linebackers. Michigan’s “best” linebacker was Jonas Mouton who came in 25th among Big 10 linebackers in points taken.
Jonas Mouton: led the group with 17 points taken on 34 plays.
Obi Ezeh: Right behind Mouton with 16 points taken on 35 plays.
Craig Roh: Would have been a top 25 defensive lineman in the conference last year and should see his production increase above last years 13 PT as he moves to more of a standup role.
This group has experience but a long history of mediocre performance. The hope is that the coaching consistency can help this group turn the corner but at this point Roh is only one who seems able be a positive difference maker.
Classified as a safety, Stevie Brown was the most productive defensive back in the Big 10 last year with 32 PT. Classified as a linebacker he was still top 10. Donovan Warren had only Stevie Brown and three other DB’s in front of him at 26 PT. Where the linebackers are a question because of limited production the secondary is a question because of limited bodies.
Jordan Kovacs: the lone returner that managed to crack the top 50 DB’s last year. The walk-on garnered 16 points taken on an impressive 30 plays made.
The only good thing about the secondary going into this season is that the expectations will be very low. Any success they can manage will likely be more than most fans are counting on.
One of the biggest challenges Michigan has in climbing the hill this year is their schedule. Based on my PAN projections, Michigan has the 22nd toughest schedule in the country and second toughest in the Big 10 behind Minnesota. Only Minnesota and Indiana have easier Big 10 schedules and that’s because they don’t get to play Indiana and Minnesota.
So where does this all net out? Here is Michigan’s team PAN since 2003 along with my 2010 projection.
As detailed in the comments of by wannabe Blogpoll, PAN projections are done using 2/3 historical performance (2003-2008) and 1/3 previous year performance. Based on previous years, this combination of program history and most recent performance yields the most accurate pre-season predictions.
So what does that mean for wins? Using all teams PAN predictions I ran a Monte Carlo simulation for the season and at the 6.6 PAN above, this is your win expectation chart for 2010.
Over a 50% chance of 7-8 wins, a 93% chance of bowl eligibility.
If this projection seems either too high or two low, I ran 3 additional simulations, one with Michigan’s PAN increased by 3, one decreased by 3 and one with performance at last year’s level.
At last year’s performance the chance of bowl eligibility drop to 56%. At the optimistic side, wins average about 9 with a 27% chance of 10+ wins.
*PAN is calculated by assigning every play a value based on how much the play helped or hurt the offense’s chances of scoring. Every down, distance and line of scrimmage combination is assigned an expected value, the average points scored across college football in that same situation. If a play increases the expected value, the respective teams and players are credited with the amount of increase.
All plays are then adjusted based on strength of opponent. Plays against weak opponents are penalized and downgraded while plays against strong opponents are bumped to reflect the degree of difficulty.
Only games against FBS (D1A) opponents, games against FCS (1AA) opponents are non-existent in any numbers used in this work.
Qualifying Plays (QP) are all plays in the first half and plays in the second half when the game is within two touchdowns. End of half run out the clock drives are also excluded.
a badass post from the mathmadman. If we can muster 8 wins this season, I will be freaking perma-smile-man. Go Blue! Only 10 days left until I'm sitting in the BIG HOUSE!!!
I thought it was "2 wrongs don't make a right, but 3 rights make a left".
If you run this for every NCAA team, how much money can we all make in Vegas on the wins over/unders?
Thanks for the hard work.
(Played by John Goodman!):
Great post. Excellent work. Thank you.
I love this site
So what your saying is we are going undefeated