Visual Comparisons of Michigan's Class w/ Its Rivals

Submitted by LSA Superstar on

I did this post last year, and it got a lot of positive feedback.  It's difficult to measure the overall quality of a class, because quality is defined by how many players you get and how great those individual players are.  So I feel a graph is a little bit of a better demonstration of how various classes compare than a straight numerical rank.

The first chart shows Michigan's 2017 class as compared to MSU's and OSU's.  Each team's best player is ranked first, with each subsequent player listed in descending order along the x-axis.  The y-axis is measured based on the 247 composite rank - a value of 1 is a consensus number one player, .98 tends to be the line before five stars and four stars, with .89 typically being the line between four stars and three stars.  Click to embiggen.

Analysis?  I don't think I need to elaborate.  Michigan's class is stellar, but it's definitively eclipsed by OSU's which is vastly superior pretty much all the way through.  Both dwarf MSU's class, which is in a very distant third place.

What I think is also relevant is comparing these classes with each teams' from 2016:

Here, you can see that Michigan's 2017 class is better than the 2016 class.  The Michigan 2017 class is also equally as good as OSU's 2016 class.  Finally, you can easily see that MSU's recruiting fell off dramatically from last year.

MGoBoard
Football

BursleyBaitsBus

February 1st, 2017 at 7:11 PM ^

If Michigan can somehow manage another 10 win season in 2017 with Ws over the rivals... 2018 will be insane recruiting wise and on the field. 

Joined: 01/20/2016

MGoPoints: 12434

Logan88

February 1st, 2017 at 7:35 PM ^

I think that OSU's insane class this year was fueled by their crazy NFL draft from last year (5 first round picks, 12 overall). It seems like recruits respond to NFL draft success even more than National Championships.

Michigan is expected to have a pretty deep NFL draft class this year (although not as many first round picks as OSU had last season). If UM has 10-12 players drafted in the upcoming draft, I think UM could see an even better class in 2018 than we just saw for 2017.

Joined: 01/05/2009

MGoPoints: 8078

YoOoBoMoLloRoHo

February 1st, 2017 at 8:15 PM ^

UM should gain solid draft boost this year along with a smaller class in 2018 - expect a very high caliber average ranking next year. Also, this staff finally has established relationships with HS coaching staffs so they are not playing catch-up. Meyer did not have to establish these relationships with less than a year hiatus and his staff had a network as college coaches. UM actually has even more upside. Meyer will continue to recruit at the top. Dantonio is probably well past his best recruiting days.

Joined: 11/01/2012

MGoPoints: 18054

Blue In NC

February 1st, 2017 at 7:18 PM ^

Thanks.  These graphs are very helpful.  To me it looks like UM's 2017 class is clearly (but not significantly) better than OSU's 2016 class (equal at the top but with better and more depth).

Joined: 06/30/2008

MGoPoints: 8575

ppToilet

February 1st, 2017 at 7:39 PM ^

and appreciate your efforts, OP.

However, the visual representation is perhaps misleading by the spacing on your Y-axis. That is, the "visual" separation may not represent as "meaningful" of a separation as it appears. For example, Michigan and OSU's best recruit is about 0.99 while MSU's best recruit is about 0.91. However, the scaling has that recruit about 50% of the way down on the Y-axis. This is "visually misleading" as that recruit is not likely to be "half as good" as the OSU and Michigan recruit.

The suggestion of putting other B1G teams on here is useful because the bottom dwellers would force compression of the Y-axis. Sorry to nit-pick: one of my gigs is reviewing research.

Joined: 04/18/2011

MGoPoints: 5261

The Maizer

February 1st, 2017 at 8:48 PM ^

As a researcher, I strongly disagree with this. You would have a point if recruit ratings actually varied from 0 to 1, but that is not the case. The lowest ranked recruit for Rutgers, for example, is a rating of 0.77. I would argue that DPJ is more than 28% better than Rutgers' worst recruit, agreed? 

Joined: 08/07/2015

MGoPoints: 10228

ppToilet

February 1st, 2017 at 9:16 PM ^

Not sure we disagree that strongly. My point was that the Y-axis was too skewed visually. In your example, since 0.77 is below the bottom of the Y-axis, this graphical representation would suggest that DPJ is infinitely better than Rutgers' worst recruit. While I would certainly like and hope this to be the case, the odds are that it is not true.

Again, in my example, I was comparing a .99 recruit versus a .91 recruit. Perhaps a better visualization of the Y-axis would be logarithmic to your point?

Joined: 04/18/2011

MGoPoints: 5261

The Maizer

February 1st, 2017 at 9:49 PM ^

Okay, you caught me; I don't feel strongly about the visual representation about this set of data. I think the important point is that quantifying how much better one football player is over another (particularly in a position invariant way) is not particularly feasible.

Joined: 08/07/2015

MGoPoints: 10228

ak47

February 1st, 2017 at 9:07 PM ^

While it would look nicer for Michigan I disagree with this. For example being rated a .99 could make a player a top 10 player in the country while being a .97 puts you at 50th and a 4 star. That person is a really good recruit but there is a noticeable drop off in prediction success rate from 5 to 4 star and like top 25 and not. Being a a 5 star top 25 ish player should look noticeably different on a graph evaluating recruits than a 50th ranked 4 star.

Joined: 05/05/2011

MGoPoints: 33614

ppToilet

February 1st, 2017 at 9:30 PM ^

Not to rehash that old topic, but the last I looked at it the success rate of a 5-star making it into the NFL was about 25%. This dropped off for 4 and 3 stars, as I recall, but as there are numerically more 4 and 3 stars then more of those make it to the NFL.

Predicting success is very difficult. A simplistic graph makes us feel better sometimes because we're #winning. Then, when the game is played, it sometimes does not seem that our athletes are light years ahead of others (excluding Rutgers). Maybe it's because our expectations are incorrectly set and because these rankings do not take into account coaching, development (physical and otherwise) as well as positions of need. That's why I agree with the "bigger is better" ranking class sizes. I also think there are different positions that should be weighted greater than others.

I think these sorts of graphs can be useful for a program to rate itself (especially if the coaching staff is stable), but there is so much variability that it's a bunch of guesswork. As I mentioned above, maybe a logarithmic scale would "look better".

Joined: 04/18/2011

MGoPoints: 5261

maquih

February 2nd, 2017 at 8:21 AM ^

I see these kind of comments all the time on all sorts of graphs. But I just don't understand -- the y-axis is clearly labeled, so how is it misleading? We make those visual adjustments all the time and use zoom all the time on cameras, while watching tv, etc. Is a slow motion replay that zooms strongly into the play to show a controversial call because it makes the ball look the size of the entire screen, is that misleading? Is a high school graduation headshot photo misleading because it leaves out 90% of someones body? No, the contexts are clear like it is with this y-axis. We are zoomed into the 80-100 ratings zone for a closer look, what's misleading about that? A chart that went from zero to 100 and made everything look bunched together would be both less useful and misleading because the gap between a 3-9 State team and Michigan's recruiting is actually quite large.

Joined: 11/04/2010

MGoPoints: 1780

Stuck in Utah

February 1st, 2017 at 7:41 PM ^

It would be interesting to make additional graphs for previous years, then to reevaluate when they graduate and re-graph to see 1) how development went, or 2) how inaccurate the recruit rankings can be. 

Joined: 07/07/2013

MGoPoints: 416

AAB

February 1st, 2017 at 7:42 PM ^

I love this.  Also, I know a lot of it is a class size thing, but the fact that OSU doesn't even have the #1 overall class is just insane given what they put together.  

Joined: 01/14/2009

MGoPoints: 8284

M-Dog

February 2nd, 2017 at 10:00 AM ^

They should rank kickers seperately.  We dropped from the 4th ranked class to 5th because of our 11th hour offer to a punter with a slot we had open.  

That did not really make our class "worse", it made it better.

These rankings discriminate against kickers and they shouldn't.  If you are the #1 ranked kicker you should be at least a 4-star, the same way a #1 ranked TE would be.

A quality kicker / punter is a difference maker.

 

Joined: 07/06/2008

MGoPoints: 257517

Lil boy blue

February 1st, 2017 at 8:23 PM ^

Most are niche type players or project pieces who could have very productive careers in the system. Wisconsin always competes by recruiting within the Barry Alvarez system. Harbaugh will sit in the homes of the highest rated OL and RB recruits in the near future and talk about a lowly 3 star, NR player or maybe even a PWO playing in the NFL and ask them to imagine what he can do with them. Just a matter of time before this whole show takes off Edit: I am aware Glasgow was a walk on ... there will Ben more examples to add

Joined: 12/30/2016

MGoPoints: 1985

stephenrjking

February 1st, 2017 at 9:46 PM ^

Sure we will. Even great recruiting teams whiff all of the time, and Michigan is now on a level where they're in play for a lot of great players. So we miss on a Wilson or a Slaton, but we hit on a Gary and a DPJ and a Solomon.

I hope we hit on a few more, but Michigan is already hitting on a lot of these guys you're talking about, the ones that are clearly starting with more talent than the 3-stars Harbaugh developed into NFL players at Stanford. 

Joined: 03/15/2012

MGoPoints: 20143922