Is Michigan staff better at finding/devloping 3* players?

Submitted by pescadero on February 26th, 2019 at 1:56 PM

There is a strong "trust the coaches" vibe to Michigan picking up 3* players - and a relatively strong belief on the board that Harbaugh does an above average job choosing/developing 3* players. After next season we'll have a better idea how those 3* players from Harbaugh's first two classes end up in the draft.

Historically - just about 9% of all 3 stars (.8->.89x composite) get drafted in the NFL draft. There were 25 such players signed in the 2015 and 2016 classes.


Average performance would be 2-3 players drafted from that group.


Devin Bush is at this point the only certain draft pick I see from the 25 players. Higdon is possible. Khaleke Hudson is pretty likely.

That would put Harbaugh as pretty average over those two classes. A few other guys hitting (Kemp? Uche?) might put him over average.

...but to get to anything more than slightly above average (for those classes) will require 4-5 of the following to be drafted.

Shelton Johnson
Karan Higdon
Reuben Jones
Grant Perry
Keith Washington
Jon Runyan Jr.
Nolan Ulizio
Andrew David
Carlo Kemp
Chris Evans
Nick Eubanks
Eddie McDoom
Khaleke Hudson
Elysee Mbem-Bosse
Nate Johnson
Quinn Nordin
Josh Uche
Kingston Davis
Steven Spanellis
Josh Metellus
Sean McKeon
Michael Dwumfour
Devin Gil





February 26th, 2019 at 2:08 PM ^

This post was like True Detective.   Started out with a pretty strong idea but it never came to execution.   Not sure just picking guys drafted from 2 classes is enough of a sample size to get anything done.  If Higdon gets selected in the 7th rd or goes undrafted his career was a success at Michigan and out performed his 3 star status.   

Not sure how you quantify it but drafted in 7th rd is success,  not drafted but makes an NFL roster is failure is not the right grading system.    


February 26th, 2019 at 7:07 PM ^

Doesn’t 247Sports Composite consider any player rated above 0.8900 to be a 4* recruit?

I thought the composite had a very clearly defined range for their star ratings.

In Devin Bush’s case, he was 31 spots above the highest rated 3* recruit (Brodarious Hamm [0.8893] - great name).


February 26th, 2019 at 4:25 PM ^

" Not sure just picking guys drafted from 2 classes is enough of a sample size to get anything done. "

It really isn't - it is at best a slight hint pointing out direction.

I'd say you probably need a good 4 classes worth to make any sort of judgement (either way) that can't just be dismissed with "sample size".


February 28th, 2019 at 10:09 AM ^

A sliding point scale might work.  You certainly need to get more credit for a 3 star 1000 yard RB who made all Big Ten vs a 3 star that washed out with neither making the NFL. 

Also something to be considered would be early 3 star guys who became 4 star guys possibly or partly because of Harbaugh’s early interest.  


February 26th, 2019 at 2:16 PM ^

I don't think it's fair at all to include the 2015 class. Those aren't really Harbaugh's guys. He had no time to scout players or build proper relationships with them. Therefore, what we are really looking at is something like this:



Devin Bush

Khaleke Hudson

Josh Uche

Josh Metellus

Kwity Paye



Sean McKeon

Nick Eubanks

Michael Dwumfour

Andrew Stueber

Ben Mason

Donovan Jeter/Benjamin St-Juste/Joel Honigford (development pending)


February 26th, 2019 at 4:39 PM ^

IMO - there are 2 players (given performance to date) that are likely to be drafted from that list: Bush and Hudson.


Everyone else on that list is at this point a maybe... and Metellus, Dwumfour, Steuber, Mason, Jeter, St. Juste, Honigford? I'd consider it a good showing if more than one of those guys gets drafted.


February 27th, 2019 at 10:24 AM ^

Why is Metellus a maybe? Delano Hill got drafted in the 3rd round after a 2nd-team All-BIG senior year. Metellus was a 2nd-team All-BIG selection last year.

If you assign likelihoods to each tier (e.g., likely draft pick=90%, maybe draft pick=50%, long-shot=10%):

4 likely draft picks (Bush, Hudson, Metellus, Higdon) x 90% = 3.6 draft picks


4 maybe draft picks (Uche, Paye, McKeon, Eubanks) x 50% = 2 draft picks


6 long-shot draft picks (Dwumfour, Steuber, Mason, Jeter, St. Juste, Honigford) x 10% = 0.6 draft picks


You have 6.2 expected draft picks* out of 25. That's an expected draft rate of 25%, which is WAY above the national average. So it feels like you're being overly pessimistic here.

*Note this is also excluding contributors from last year like Gil, Kemp, and Evans. So I think this estimate is actually somewhat conservative.


February 27th, 2019 at 11:43 AM ^

" Why is Metellus a maybe? Delano Hill got drafted in the 3rd round after a 2nd-team All-BIG senior year. Metellus was a 2nd-team All-BIG selection last year. "

Hill was an inch taller, 15lbs heavier, and ran a 4.47 at the combine.

Metellus has a reported hand timed 4.5x - which if accurate is likely a 4.7 laser timed 40.

I absolutely wouldn't be at all surprised to see both Metellus and Higdon go undrafted.

I don't think McKeon has even a 5% chance of getting drafted at present - much less 50%.


...and your "long shot" percentage is higher than the AVERAGE percentage for folks with composite scores in their range.



February 27th, 2019 at 2:02 PM ^

Those are all good points, but I still think you're being overly pessimistic to make a point. 

Moving McKeon to the long-shot category (and changing the long-shot percentage to 5%) and Metellus and Higdon to the maybe category (and changing the maybe category to 40%) STILL gives you an expected percentage drafted of almost double the national average.

And that doesn't even consider guys we're not talking about or who develop better than expected. What would you have set the odds for guys like Rudock, Clark, Chesson, or Ryan Glasgow midway through their careers.


February 27th, 2019 at 3:29 PM ^

Rudock was mid 3* recruit (0.8559) who increased his stock a little at Michigan - but didn't play significantly differently from his Iowa career. It was no sure thing that he would have been drafted coming out of Iowa - but a 6th/7th round pick like he ended up would not have been that surprising.


Clark and Chesson were also mid 3* recruit - so I'd expect about a 10% chance they get drafted.


The Chesson class definitely beat out the odds though - you'd only expect 1 player drafted based on history (10 players between 0.8->0.8x) and they had 3 players drafted (Chesson, Willie Henry, Jeremy Clark).





February 26th, 2019 at 4:36 PM ^

I absolutely don't care about the NFL draft, and completely care about the CFP.


That being said - being drafted is the best proxy data I've been able to find.


If someone has the data correlating all-conference selections to composite rating that would be a much better proxy - but I was unable to find that.

SMart WolveFan

February 27th, 2019 at 11:11 AM ^

Actually, I think there is a spreadsheet out there comparing recruiting rating to college win percentage.

The data is not as clear cut plus shows a lot of confirmation bias (because ...well, you know Bama, right?), but it does show the fact that, in the aggregate, the teams that win more games have better recruiting rankings.

It also shows that good recruiting is no guarantee of success, look at USC and FSU and how they fell off quick even with plenty of recruiting talent; also, it shows sometimes it's harder to manage a lot of top players to that special season since OSU had more success before their gigantic recruiting classes and since have found ways to ruin seasons also even Bama had more success wining the championships before 2012.


February 26th, 2019 at 4:33 PM ^

Depends on whose metrics you are using.

The data I have regarding draft rates groups them into groups by composite - and considers everything 0.8 -> 0.89x to be 3 stars.


Given that dichotomy - lets do away with all "star" talk, and just use the composite categories.

9% of players with a composite score .8->.89x get drafted in the NFL draft. There were 25 such players signed in the 2015 and 2016 classes.



February 27th, 2019 at 10:36 AM ^

My issue with any of these ratings analyses is that they are almost always retroactive.  Chase Winovich was a 3* OLB when he signed with UM; he wound up being a 4* when it all worked out in the end.  So do you consider him a 3* to the coaches are "penalize" them for getting in early?  Similarly, judging the 2015 class is a bit unfair considering that was a half-year between Hoke and Harbaugh.  So a number of those guys were fliers the staff probably had to grab to fill out the class.

Also, the goal of college football is to win college football games, not necessarily to get guys to the NFL.  OSU won a TON of games with QBs who never even sniffed the NFL radar because they were great for college, and I doubt anyone would ding the coaches for "wasting" guys like Pryor, Miller, and Barrett.  Yes, player development is important and you want to see guys improve, but a guy like Josh Metellus who is an above-average college safety and got on a couple of all-conference teams but will likely struggle to get drafted doesn't make his growth at Michigan "bad". 

Also, the difference between a 3* and a 4* is sort of silly in some of these comparisons.  To wit, Nick Eubanks was a 3*, the #344th player in 2016.  Chris Evans was a 4*, #338th player in 2016.  If you can divine a marked difference in two players 6 spots apart, go for it.  


February 27th, 2019 at 11:47 AM ^

These are all good points BronxBlue, especially in regards to the retroactive grading (what was the recruit at the time of offer versus signing day).   And to some extent this is my issue with our current recruiting strategy - we seem to be offering and accepting very low ranked recruits (below 3 stars) extremely early in the cycle.

Now if these 2 or even no star recruits bloom into 4-5 stars by signing day then one could make the case that our recruiting is beyond excellent as we're in on the talent far in advance of even the recruiting sites.  But if they dont significantly improve I'd have to question what we are doing right now as we both know a team heavy with 2-3 star athletes is just NOT going to beat OSU and other elite teams consistently.  They just can't.

I would love to see that bit of analysis.  What was the star rating at time of offer versus star rating on signing day.  That would explain a lot or would validate those who (like me) are concerned right now.

Im thinkin bou…

February 27th, 2019 at 12:14 PM ^

What may be a more interesting data set is the correlation between some accolade (all-conference, drafted) and the star the player is ranked when the commitment was accepted.  Much more interesting to see how UM and ND loved kyle hamilton before he became ranked above dax hill.  @seth, I'd love to see you do something like this.

Ali G Bomaye

March 1st, 2019 at 9:17 AM ^

I would be surprised if Metellus, Dwumfour, Evans, Hudson, Higdon, and Runyan weren't drafted. It's unlikely that any of them will be high-round picks, but they've all shown enough to get a look from the NFL.