The Weekly Maths has Gone Rose-Colored

Submitted by The Mathlete on October 11th, 2012 at 7:56 AM

Until sometime Saturday evening, Michigan’s season was somewhat of an enigma. Entering Big Ten play the anecdotal evidence for the season was a mixed bag. Good performances against bad teams, weak performances against good (Air Force offense and Notre Dame) and great (Alabama) teams. Purdue entered Big Ten play with a similar resume. Big wins against three bad teams and a close loss against Notre Dame. Vegas installed Michigan as a slight road favorite more on preseason expectations than what has happened thus far.

Michigan responded with an old school MANBALL beat down and the picture is starting to become clearer. This is not a juggernaut Michigan team that can contend with the top teams in the country, but it is also a team that hasn’t faced a major let down the likes Michigan St, Iowa, Nebraska and the rest of the Legends division have seen already this year. Michigan’s remaining schedule shows only one team rated higher than them on offense and defense (Ohio St) and three teams that eclipse them on offense (Nebraska) or defense (Michigan St and Iowa).

Projected Michigan Wins

image

[After THE JUMP, where does Denard rank among 2012 NCAA QBs?]

Michigan probably still has some losses on the board but losing more than 2 more games is becoming less likely and there is nearly a 40% chance of losing 1 or less to finish out Big Ten play. At 6.3 projected Big Ten wins, Michigan is comfortably in the driver’s seat of the Legends division. Nebraska sits almost a full game back and Michigan State is a full game and a half projected behind. A win at Michigan Stadium in a week and a half could be an elimination game for the Spartans.

Over five games we have seen good Denard and bad Denard, good Borges and bad Borges. At the end of an up and down stretch Denard sits at +6 EV with only six quarterbacks sitting between the totality of what he has done and the now immortal Geno Smith.

Rank Player Team Games EV+/Game WPA
1 Geno Smith W Virginia 4 9.8 200%
2 Johnny Manziel Texas A&M 4 8.5 107%
3 Seth Doege Texas Tech 4 7.7 81%
4 Collin Klein Kansas St 4 7.5 122%
5 Taylor Martinez Nebraska 5 7.4 50%
6 Braxton Miller Ohio St 6 7.0 202%
7 Tyler Wilson Arkansas 4 6.8 75%
8 Denard Robinson Michigan 5 6.2 102%
9 Tajh Boyd Clemson 5 6.0 163%
10 Cody Fajardo Nevada 5 6.0 178%

This is where I just lose people who talk about benching Denard. There are going to be a few times you want to pull your hair out cheering for him, but opposing DCs (except Tim Tibesar who Toussaint keeps up at night, apparently) lose significantly more hair.

The defense has been top 10 caliber and getting better after a rough showing in Dallas. And it’s not as if Mattison doesn’t have the track to believe in the trend. In this year’s Big Ten, Michigan is by far the best positioned team going forward. Until the last game of the season I have Michigan as an underdog only once more, in Lincoln on Oct 27. A win there and Michigan is looking at a 60+% chance of entering Columbus at 9-2 and already having the Legends division wrapped up.

And there is still a very good chance Purdue is the team Michigan sees in Indianapolis. No post-season eligible Leaders team is projected to win more than 4.5 games and Purdue (4.4) and Wisconsin (4.3) are the only eligible teams projected to win more than 1-2 games.

For all the math I have, looking forward this is the equation I see:

Last week’s domination + the utter awfulness of the 2012 B1G = Good Things Ahead for Michigan

Game Chart

image

Michigan’s dominant first quarter pushed the odds up to 75% and by the time Ramon Taylor was in the end zone, the number was 90%. From there on Michigan bled it out Thomas Gordon sealed it with his interception of Robert Marve.

Game Notes

Offense: +14 EV, 28% WPA

Rush: +4, 12%

Pass: +10, 16%

Denard Robinson: +15, 27%

Fitzgerald Toussaint: –1, –0%

Jeremy Gallon: +2, 3%

Defense: –1, 4%

Rush: +2, 4%

Pass: –3, +0%

Ramon Taylor pick 6 (return not included in above numbers): +5, +6%

Note: WPA percentages won’t add up. There were some at stake in the second half that I didn’t include and special teams etc have a slight contribution, as well.

Prediction

I must have a thing for Illinois. Last time I was in town for a game was RichRod’s final home win, the epic offensive showdown that featured 132 points. This year’s Michigan team is certainly much more complete and Illinois, well, maybe a bit more chew than in 2010. My numbers have this a 99% chance of Michigan winning. When you get to this point, the margin is a pure guess. Michigan could go conservative with an early win and you could end up with a margin like Purdue or it could just escalate quickly.

Michigan 44 Illinois 6

Comments

eschaton811ydau

October 11th, 2012 at 8:28 PM ^

the mathlete= the knowledge and all of the knowledges wild ass predictions are from some crazy next generation statistics that the mathlete/knowledge is hoarding to himself. of course he's using this secret knowledge to accumulate wealth and street cred by betting solely on wildly unlikely prop bets.

death by trident

October 11th, 2012 at 8:56 AM ^

Also, opponent adjusted S&P+ rankings show some interesting things, and they pretty much agree with your analysis on Denard.  Right now Michigan is 4th best in the country when it comes to making successful plays.  If we can limit turnovers, we should be just fine the rest of the year.

joeyb

October 11th, 2012 at 11:21 AM ^

Thanks for adding expected wins back in. That's my favorite part of my favorite diaries. Just curious, what are the percentages on each game? I'm assuming that Nebraska and OSU are just below 50%, which is what would drag our expected losses up to almost 2.

turd ferguson

October 11th, 2012 at 2:16 PM ^

I'd like to see these, too.  I'll confess that I'm skeptical of the Mathlete's models, because it never feels like there's nearly enough uncertainty in there.  Last week, I think he had us with a 96% chance of beating Northwestern and only a 20% chance of beating OSU, and now we're seeing Illinois at/above 99%.  If the Mathlete let us bet against his predictions, I'd almost always bet the underdog.

The Mathlete

October 11th, 2012 at 1:10 PM ^

Most of the perceived variance is based on the fact the due to the lead, the second half is not included in these numbers. First half saw 6/8 passing and 4 big conversions while rushing had a decent 27 carries for 118 yards and 4 conversions. Second half the rushing really dominated but was excluded due to the size of the lead.

RakeFight

October 11th, 2012 at 1:41 PM ^

Probably splitting hairs here because I think your point is simply that OSU is the toughest game left, but by what measure is OSU defense rated higher in defense other than points per game where OSU holds the edge by a whopping 0.5 points allowed per game?

Michigan team defense leades OSU team defense in: Total yards allowed per game, Total yards allowed per play, First downs allowed, etc.