Under/Upper-Classmen and Win %
Apologies to the Mathlete if this constitutes stepping on his turf.
Last year's Michigan roster had a very high percentage of first- and second-year players. I looked back at the past 16 seasons and compared under vs. upper classmen on the roster and mapped it with the team's winning percentage.
It's not airtight but I think there are some trends worth observing:
We've only had one really good year with an underclassmen-heavy roster, and that was the year with fifth-year senior Tom Brady under center.
Other other two "really good years" had an upperclassmen majority and followed a year where the opposite was true.
Noticeable year-to-year shifts in underclassmen:upperclassmen tend to go along with similar shifts in winning percentage. When the percentage of upperclassmen dives, so does win %, and generally vice versa.
It suggests that this year will be better than last year...the huge sophomore class jumps to upperclass status, we've got a small freshman class and only 1-2 of them will be counted on for production (Peppers, possibly Canteen and Cole), and a fifth-year senior QB.
Thanks for the work. I'm sure mathlete won't mind. Your chart says 1st or 2nd year, but I assume underclassmen would include RSSO, yes? Also, by "roster" do you mean scholarship or total roster?
I chose to track it by year in school...so 1st or 2nd year versus 3rd/4th/5th year. Scholarship players only.
So, RS SO is an "upper-classman" yes?
Yes.
The good news is there is an obvious correlation between experience and wins. The bad news is that we haven't won 9 games or more when we are over 50% 1st/2nd year players since 2000.
I'm not sure this bodes well for our season, other than to say it's likely we'll get to 8 wins this year.
That said, great work, thanks.
in get us an extra win or two but a eight win season wouldn't surprise me this year. I'll be happy if it looks likee we are taking steps in the right direction as I expect we will.
Is this by eligibility or simply by year in school? Is a RS FR a 2nd year player or 1st year player, is a RS So an upperclassman?
Nice chart/info by the way, just interested in the clarification.
The real exceptions there were the Rich Rod years with trying to quickly ship offensive schemes.
Thanks for the work.
Good work! I don't know if you calculated it, but I would be curious to know the R-squared values for both the correlation between upper- and underclassmen percentage and overall winning percentage. I expect that you have some interesting evidence to support the "youth" aspect of last year's performance (surely a dynamic issue) which some (a seemingly vocal but small minority) seem to dismiss very easily.
that I went to business school and did poorly in my business stats class. I was very good at math through advanced algebra but I don't grok stats enough to do what you are suggesting.
Wow, this looks like a shockingly strong correlation. Pretty crazy.
Very nicely done.
This is tangential, but I still can't believe we went 3-9 a few years ago. That chart kind of highlights the awfulness of that record.
Someone brought up Justin Feagin a few weeks ago and I was surprised to learn that he had actually seen the field at UM. I was so checked by that point in the season that I didn't know.
Great job on actually creating something relevant to our ongoing offseason discussion on youth vs wins.
but to see it on a graph gives it life. My first reaction was why the arbitrary 16 years? But the very next microsecond my next thought was, the amount of time and effort to tweeze out the data to make the graph, let's be happy with a trend that accounts for three different coaches and multiple team turn overs.
I have a nerdy spreadsheet that had roster data going back to 2002, so that was my starting point. Forrest Gump like, I decided since I had gone that far, I could go a bit further. I figured nothing was to be gained by including 1997, obviously (snif) an outlier year. Plus most importantly, mgoblue.com has the yearly rosters starting with 1998.
That '97 was a bit of an outlier and so made a good end point.
If ever you want to go earlier than '98 you could always use Michigan's Bently Historical Library, it has Michigan football information, to include roster information, back to 1879.
What I really found interesting is the major jump in Win % from a year where the under classman get a lot of experiance playing and the next year when the team closes the gap of under and over. 2005-06 and 2010-2011. I really hope that is the case this year.
While the transition of the underclassmen to upperclassmen probably played a big role there were a few other factors going on as well.
Jake Long missed most of the 2005 season due to injury. He was arguably UM's best oline at the time. In addition, English was in his first year as DC in 2006.
2010 to 2011 was a complete overall of the coaching staff.
I don't like when data doesn't back my opinion/ agenda. According to your data, we should be expecting between 7-9 wins. I am expecting 11-12 (regular season) based on my very scientific "that's what I want the record to be Damnit!" /s
Great work though. I love this sight for many reasons, but the statistical models are always great, even when Mathlete's go above my head. Like you said, hopefully a 5th year senior QB can increase the percentage.
I hope the experience from last year causes us to exceed the predicted win% from this chart.
Hopefully by 3-4 wins.
I'll toss Mone's name in the hat due to what appears a much quicker-than-average first step for a man his size, thus almost always creating an avenue for him to penetrate the backfield. Mattison will use whatever he can to force a double team from the base defense and he's the type of DC that will always use his best set of tools, no matter the length of guarantee. He figures he won't have to worry about it for four years anyway and so it really doesn't matter if they're ready for prime time if they can offer some afternoon relaxation. Have no idea if it were just me, but he and Henry seemed to be in the offensive backfield on a routine basis during the spring game. Some are probably want to explain this away due to a breakdown on the part of the OL, but if that were true, it would allow for others across the front to do the same thing and this didn't appear to be the case. Pencil him in.
Retention is critical, because without it you have underclass-heavy rosters continuously. The improved retention of players under the Hoke regime should begin to pay off this year and really pay off next year.
Just looking at the depth chart of MSU and Wisconsin you can see how they are offsetting a step down in recruiting from OSU / UM by redshirting almost everyone and benefing from the value of a bevy of RS SR and RS JRs. 80%+ of MSU's 2 deeps are guys who are 4th/5th year. It is a smart strategy if you are not getting a bunch of top 150 guys because 4 years down the line, with good coaching, development, scheme, and S&C the difference between player #328 in the country and player #189 is going to be nearly nil. Especially if player #328 is a 5th year senior and playerr #189 is a RS SO.
I remember going into that game last year looking at their DL vs our OL and the interior was scary - they had Tyler Hoover a 6th! year senior (he had some sort of injury waiver for a year) going against true freshman Kyle Bosch. Hoover was 24 years old in the Rose Bowl. This stuff matters no matter how talented you area.
Again it doesnt mean just having old(er) players is an automatic scheme for winning because BYU has some bad years but if you are taking player #21 out of Ohio rather than player #7 out of Utah or Wyoming you still have a lot more to work with in terms of raw material. The more I look at how MSU and Wisconsin are doing it the more I am for advocating as many guys as possible - I dont like our strategy of wasting redshirts on players who are just special team players. If for example a Ferns is out there wasting a year of eligibility with a stacked LB core ahead of him and we lose him as a 5th year senior so he can run down the field 30 times on kick off returns - it simply makes no sense. That is where you can use a Thomas Rawls or Josh Furman type (last year).