It took that long to regrow 'em

For a moment yesterday everybody ("everybody" meaning the 10 of us who excessively pour over roster data) got really excited that we might have some spring weights, as a spring roster was apparently circulating among reporters down at IMG. Some people even noticed huge weight differentials between the data on MGoBlue.com's official rosters and the numbers on the spring sheet.

That was a false alarm: weights on the official website haven't changed since last spring, while except for Nolan Uluzio going up 4 pounds there's no difference between the 2016 spring roster weights and those listed in Michigan's media guide prior to this year's bowl game.

But it spurred me to finally update all those weight listings on the MGoRoster database. "Wgt" now means the last weight they were listed at on any Michigan roster.

The excel spreadsheet that is based on has all sorts of historical playing weights and start data too. And I realized that means we can actually say, quite accurately, how big a starter was at any given position.

Tall and thin or short and squat? Some positions seem to have preferences.

Methodology: The data are from my heavily edited version of Bentley's database (there are hundreds of errors in there but since they've only ever annoyed myself and Bruce Madej nobody's ever got around to fixing them), and filled in with subsequent roster data from media guides and whatnot. For guys who never enrolled I used their weight as a recruit on Rivals). Snapshots for a year are fall weights.

I also had to go through my starter data and refine it—originally I broke them up into wider position groups but this required some precise definitions. For the years Michigan went to a 3-4 it was at least obvious which DE was basically a 3-tech versus which correlated to SDE, likewise with the WDE/RLB distinctions, and which ILBs were MLBs or WLBs. The hardest part was parsing out hybrid spots versus when Michigan just started in a nickel, but it was quite obvious by which player they used. If Michigan was using the slot positions in a special way (e.g. slot receivers and spurs under RR), I counted them as such. That includes Funchess as a slot receiver and the base nickel with Countess or Leon Hall, but doesn't include Arrington, Manningham and Greg Mathews lining up anywhere (they're all wide receivers).

Ultimately I could say things like freshman Henne was X tall and started Y times, and once I had that I could properly weight that in a calculation of all QBs to determine an average QB starter.

And make charts!

Chart of freshman weight vs starter weight by position:

Starters are bigger than freshmen, duh. But certain positions required more growth than others. I thought the disparities at various positions were interesting. Both defensive end-liked positions demanded the biggest gains (even including some pretty tiny starters at times), with WDE (13%), SDE (12%) and DT (6.9%) the top three in starter/recruit disparity:

No, they rarely get taller. But I did find it interesting that the slot positions didn't see a lot of change from when those guys stepped on campus to when they graduated. But then they were rather young.

Chart of Average Starter Age:

If you didn't click for big you probably can't read on the left axis where it says a ONE equals a TRUE FRESHMAN—I figured that is less mathematical but more intuitive because "5th year senior" etc.

On average Michigan was starting a guy at least in his third year in the program. But nickel saw a lot of youth, while MLB (yikes!) and center (that seems obvious) rarely strayed from redshirt juniors and seniors unless they couldn't avoid it, or they had David Molk on hand. WLB was often used for MLBs in training.

I bet quarterback would have been much higher if the last decade's troubles didn't interfere. We can show that.

Wow, Michigan has had effectively older quarterback play EVERY SEASON since Tate and Denard were freshmen. Let's post this. Or one more thing:

Effective STAR (recruiting) Ratings of Starters

Remember all that work I did to create a consensus star rating between the sites, kind of like 247's composite but slightly less useful? Here's how much of it was on the field any given year:

Poor Rich Rod.