Jimmystats: So Many Stars. Stars Are Good

Submitted by Seth on February 3rd, 2017 at 4:36 PM

I have revamped my player database, and learned how to make gorgeous interactive charts. Wanna see where this class stacks up?


Mouse over the dots to see whom each belongs to. The orange ones are this year’s class (they limit free users to only a few colors and I was trying out a bunch of these).

[UPDATE: Didn’t see that they limit your views too. No more interactivity—if somebody knows a good site to build these let me know. In the meantime if you download this and open it in your browser I think it will work for you.

The spreadsheet still lives here and includes a ton of updated data thanks to some readers who helped me out. If you want to see the actual ratings and rankings that went into these numbers I’ve put that all on a separate tab. Behold:

I had some help. Reader David Moorhead pulled out all of his old recruiting issues of The Wolverine that had data going back to 1990. Much of the National Recruiting Advisor (ancestor of Rivals), Parade, Lemming, PrepStar, Street & Smith, USA Today, and SuperPrep (Scout predecessor) data came from his work. Also reader Jeff Alotta helped me play around with the math some as I tried to rebuild how I give out star ratings for regional and national position ranks.




The receivers and front seven look amazing when stacked against the players who’ve come through here in the modern era. It’s also a very balanced class. And it’s huge. Getting to Best Class Ever™ would be tough. The Class of ’94 formed the basis of a national championship team and while not everyone stuck around almost everyone made it to the NFL. The next class then produced two guys in the conversation for greatest football player who ever lived in Woodson and Brady. On the other hand this class matches either of those in average quality, and then doubles the size.


That appears to be the case, statistically, when I compare the star ratings of past players to how many games they started at Michigan. Note not just the trendline, but where the NFL players came from:


click biggerates.

The average star rating (on my 5-star sliding scale) of a future NFL player coming out of high school was just under 4.25. That’s roughly equivalent to a top-125 player who’s the #2 player in Ohio or the 9th best cornerback in a deep year.

That r-squared is saying “they’re related but star rating is no guarantee.” Note however that lots of starts don’t necessarily mean quality, e.g. Ezeh. You should also note that the number of little diamonds bunched at zero starts gets thinned out considerably as it gets into the 4-star range. This is consistent with every other study that compares on-field performance to recruiting ranking, which always show you can get great players from the 3-star ranks but the higher-rated players are progressively more likely to contribute.

Let’s blow up that bottom corner to see the 5-stars who had fewer than 10 starts at Michigan:


It’s hard to look at that and make a claim that the scouts got it wrong. Five of the seven left with eligibility remaining to play with another Power 5 school or the Yankees—Fargas, Simmons, and Mallett would all play well as upperclassmen elsewhere and stick in NFL. Baraka couldn’t stay sober, so that wasn’t even a scouting issue. Henson is a special case.

That leaves Green and Grady. On review of every other consensus 5-star running back in recruiting database history up to when Green committed that seems to just be horrible bad luck.  Grady and Green were overrated or undeveloped, which sucks since every other RB rated as highly was either awesome or lost his career to something not related to talent.


The last two classes, like the 2012-’13 hauls under Hoke, are making up for the two smaller classes between. The 2016 class had to play a lot of guys right away to fill the depth chart and this year will be the same. There’s no way around having an incredibly young team this year. By next year however these hauls will start to show. I think we’re done with 30-man classes for the time being.

National championship teams have to get close enough for luck to do the rest. The 2016 team was that with terrible luck despite a lot more misses than normal for the quality and size of Hoke’s early classes that built it. Harbaugh’s found Michigan another one or two shots at it again down the road. That’s all you can ask.


The Maizer

February 3rd, 2017 at 4:55 PM ^

Your interactive chart says:

Plot twist! Payment required... This embedded plot has reached the maximum allowable views given the owner's current subscription. Please visit the subscriptions page to learn more about upgrading.

This makes me sad because I enjoy a good chart.

Naked Bootlegger

February 3rd, 2017 at 5:01 PM ^

Good way to end a Friday.   Two years from now, we will reap the benefits of back-to-back stellar classes.   We should have a fantastic nucleus of talented players, led by either a senior Speight or seasoned Peters that has beaten out Speight head-to-head.   I like it.

The Maizer

February 3rd, 2017 at 5:04 PM ^

I don't know a good analysis solution here, but number of career starts isn't necessarily indicative of a player's value. A mediocre player at a thin position may get more starts than an excellent player behind an also excellent player on the depth chart.

Having said that, it would be interesting to look at this per position as well to see where the correlation is the strongest.


February 3rd, 2017 at 5:36 PM ^

The final paragraph is fantastic. A great summation of a complicated process, put plainly and yet losing no meaning.

Great opinion, and great writing.


February 3rd, 2017 at 5:53 PM ^

Seth, this came out great!

This line says it all:

"National championship teams have to get close enough (recruiting-wise) for luck to do the rest."

Harbaugh is there  The luck was not there this year, but luck does not root for any team in particular.  It will even out.


February 3rd, 2017 at 7:10 PM ^

I have argued this year, the gambler's fallacy has it's limits.

Sure, your rnext coin flip is 50/50 but if you are betting on a series of coin flips and the first 5 are tails, the odds that the next five will favor heads is just math.  

I don't know if bayesian inference really says that we shold see more heads in this scenerio, but I have always read it that way.  

I am sure I will get some pushback so for clarity sake I want you all to understand that their is no doubt the next coin flip is always 50/50; I am talking about a series of bets...





February 3rd, 2017 at 7:24 PM ^

No previous results, no matter how bizzarre or deviating from the norm, have any impact on the future results of a coin flip. If you keep betting for the next one to heads eventually you'd be right but not because you were "due" the result. 


Last seasons bad luck (and officiating) doesn't mean we're due for a lucky season. Even if it did, I think Hoke pretty much mortgaged all of Michigan's future luck on his first season. 


February 3rd, 2017 at 7:32 PM ^

I am talking about betting on a series of events.  In the case of the coin flip even though each flip is 50/50 on its own there are expected sequential outcomes and the math does hold, I think over the sequence...

Am I taking crazy pills?



February 3rd, 2017 at 9:03 PM ^

There is a difference in saying before you start the odds of the sequence HHHHH are one in 32, and in saying that if I have seen HHHH then the odds of the next flip being tails are greater, because it was 1/32 before I started watching.

The difference is due to the fact that the "improbability", if you will, has already been taken into account.  The odds that you would end up in the universe where the sequence was HHHH was 1/16, and the odds that you'll continue into the universe where the sequence is HHHHH are 1/2, and HHHHT is also 1/2.

So, no, luck doesn't have to even out, especially if you bet on it.



February 4th, 2017 at 1:32 PM ^

You obviously passed math, statistics & probability - and yet I can disprove your equation, and all the talk of the gambler's fallacy with one simple proposition:

You are forced at gun point to call either H or T -  "No Country For Old Men" style -  and you've just witnessed the sequence HHHH unfold. Now it's your turn, your life depends on getting it right - you taking H or T?

It doesn't seem so theoretical now, does it? Is there still a 50-50 chance that something that was wildly improbable before the sequence began will occur, or does probability start to play a part?

Something can be both right and wrong at the same time in math - weird, huh?!


February 4th, 2017 at 3:11 PM ^

It does not matter what I pick.  

If HHHH occured, and then you waited a year to flip again, would you still think it mattered that you hit HHHH a year ago?

The reason nobody ever flips 100,000 heads in a row is not because the coin tries to "make up" for the fact that you just filpped 4 heads in a row or 10 heads in a row or 20 heads in a row ... it's because the odds on the next flip are truly 50/50.  Tails is just as likely as heads on the next flip.



February 4th, 2017 at 3:47 PM ^

Anton Chigurgh is still demanding an answer friend-o...

And he's not waiting a year.

I'm willing to bet that with your life on the line, having witnessed HHHH, you would take your chances that the odds would even out, so to speak. You would call Tails.

That's what I mean by math being both right and wrong at the same time in some instances.

Yes, the coin still has only 2 sides, and only 2 possible outcomes, but the odds of one of them coming up 5 consecutive times are prohibitive.

Like if I'd told you last September you can either put 32-1 that on Feb 5, 2017 the Falcons would win the SB, or that a coin flipped 5 times would come up H all 5 times, I'll bet you again that you'd be rooting for the Falcons tomorrow, a potential handsome payday awaiting. That's how unlikely it is to come up H 5 straight times. 

And while I recognize that if it's already come up HHHH the most improbable part of the sequence has already occurred in this instance, I know I'd take Tails. 

Matt Damon solved this whole problem on the board in Good Will Hunting. You should totally rewatch that...


February 4th, 2017 at 4:03 PM ^

I'd argue with you more but we are running out of room!

We did this exact experiment in a probability class that I had in college, because the argument you are making is intuitive and it is what most people would think.

But a tail is no more or less likely than a head, even after 10 heads in a row.

If HHHH already came up, I would have no problem picking H on the next flip.

I would have a problem saying that I can flip 5 H's in a row.  It's not the same thing. 

Predicting the outcome of flip #5 (an independent 50/50 event) is not the same thing as picking the outcome of all 5 flips.


February 4th, 2017 at 4:21 PM ^

Well, while we're having this dialogue allow me to say, from my vantage point a mere 97,000 Mgopoints below you, that you are one of my favorite posters on here, and you often make me chuckle. So thanks for that.

I'm still taking tails.

And I still say you would have taken the Falcons at 32-1 to win the SB over a coin coming up the same 5 consecutive flips. And they were actually 150-1, so you'd have been cheating yourself!

We can now get back to saying who were the most under- and over-rated prospects coming out of HS in Michigan's past quarter century:

Eugene Germany & James McKinney stick out to me as most over-rated, aside from the most obvious, those 5* RB busts... I'd like to say Ryan Mundy, but he was good elsewhere once he got better coaching (or PEDs possibly?)

And a 2* gem in Michigan Mike of course has to head the over-performed team, QBd by that guy going for his 5th SB ring tomorrow!


February 4th, 2017 at 5:07 PM ^

Damn!  Now I can't say anything snarky.

The beauty of the argument is that if you take tails . . . you won't be wrong either!

My #1 bust would have to be RB Sam McDuffie, because he looked like a bust before he even left high school.

Everybody got all hyped-up about that video of him hurdling a small skinny defender and got way ahead of themselves.  He was ranked much too high.

Then one of his big HS games was nationally televised on ESPN.  He didn't do squat against better competition.  I wondered what all the fuss was.

He came to Michigan and the pattern continued.  He was not going to cut it against real competition and he ran in a way (constantly jumping off the ground and spining in the air) that would get him hurt. 

Also speaking of PED's, Justin Fargas did well at USC after he left a mediocre stay at Michigan, but he looked 'roided up at USC, IMO.  I hope I am wrong about that.

That's one I would not bet on.



February 4th, 2017 at 6:55 PM ^

I just recall that a very good '03 team had been taken to task by usc in the Rose, and our D looked very slow in comparison. McKinney & Germany were supposed to give us the type of front they had (like our '16 team actually was!) Looking back I guess Germany wasn't as highly ranked as I'd remembered, but it was a usc-M battle that we won, so I thought that meant he was a stud at the time.

Neither ever played a meaningful snap that I recall. In '06 we were fantastic on D til the last 2 games, but in '07 if those 2 and the others in their class had panned out maybe the Horror never happens! And '08!?! Fergodsakes!!!

Some of those '04-'06 D recruits could've really helped, but so many were just busts. Those 2 + Slocum, Mouton, Antonio Bass (unfortunate injury), Stevie Brown (til has last season), Terence Taylor, Adam Patterson (rated higher in his class than Taco, Wormley, Van Bergen, Mike Martin were in theirs, or than Hudson, Malone-Hatcher or Irving-Bey are in this!!!). All '04-'06 busts. Led to some bad times!

There were a lot of disappointments on that list as a whole - it really gives you pause when you see guys you know were doomed to be out & out busts rated higher than anyone in the current class except the 5* players...


February 4th, 2017 at 4:21 PM ^

Told to me by an engineer a long time ago when I proposed a possible strategy for betting on roulette, "You are assuming the roulette wheel is perfectly engineered so that each number has an equal probability of coming up on any spin.  Such a wheel does not exist."

Seems like a better bet going with heads after heads has come up 4 times in a row.  Either:

a) the odds are 50/50,

b) there's something not quite right with the coin such that it is more likely to come up heads, or

c) there something not quite right with the coin such that it is more likely to come up tails AND the previous 4 results of heads has been exceedingly improbable -- far beyond 1 in 32

Why not take a chance on a & b over c?


Ty Butterfield

February 3rd, 2017 at 8:32 PM ^

The luck will never even out for Michigan. One of my earliest memories is the trip on Desmond Howard and it just goes on from there. Even in 1997 Michigan was screwed by a kicked ball in the Nebraska vs Missouri game. I have no doubt Harbaugh will win a lot of regular season games at Michigan. Beyond that I am not sure. I have had season tickets with my Dad for over 20 years but this will probably be the last season. Not sure I can justify spending the money after yet another season ending loss to OSU.


February 3rd, 2017 at 6:10 PM ^

Hey, Seth, could you do a "Sum of Stars" chart that is per recruit?  The simplest way would be to divide the sum of stars by the number of recruits, while a more refined version would be to divide the sum of stars *for each position group* by the number of recruits per position group, then the sum of the average stars per position group by the number of position groups.  This last is complicated, as it doesn't work out quite right in years where there is no recruit in a position group... so presumably K/P's would need to be left off...

In any case, an "Average Stars per Recruit" would level out some class size issues... it wouldn't be fully useful on it's own, but as an addition to the raw sum, would give a bit of insight into who got high net stars due to lots of players, and who got, on average, higher star players.

Thanks for the data!  Great to sift through and consider!


February 4th, 2017 at 4:31 PM ^

With 25 recruiting classes, on average one would expect 4 of the top 100 recruits in each class. If the team is composed of players from the last 4 recruiting classes, there should be about 16 top 100 recruits; 20 if one looks at the past 5 recruiting classes.  Of course some players (e.g., Jabrill) leave before 4 years is up. 

Regardless, it looks like next year's team will have a few more top recruits than many previous classes.  But most of them will be true freshmen and sophomores.


(Side note:  It strikes me a bit odd that Ambry Thomas and Henry Poggi are only 2 spots away from each other on this list.  The excitement for Thomas seems to be much much higher than it was for Poggi -- I certainly recall there being some for Poggi, but perhaps that came along with an expectation that he'd put on 40+ lbs. of muscle and play D-line.  Of course many would argue that Thomas is better than his ratings.  We'll see.)

Goggles Paisano

February 3rd, 2017 at 8:42 PM ^

If there was a chart that showed lowest ranking vs. most success I would say it would go to BJ Askew.  I was surprised to see his low rating.  He had a pretty nice career college and pro.  


February 4th, 2017 at 1:56 PM ^

Askew was a fine player, but I feel certain that the player who most outperformed his HS rating - and this may honestly be for all of college football, not just Michigan - is Mike Hart.

I don't know how Seth & Co's revisionist system of awarding points works, but he gives him 3.97 while also posting him as "not ranked" by Rivals, and rated as a CB by Lemming. Superprep did call him the #12 RB - but I don' think that was for the whole country, just for the NE region. If they had him that highly ranked nationally they were indeed prescient and I hope their scouts are employed by one of the services still.

My personal recollection is Hart was a 2*, maybe low 3*, who was supposed to be no more than a change of pace back for Max Martin. His tape was posted with "watch this dude torch 0* future Budweiser truck drivers" caveats.

And just a few spots below Hart on the chart is Tom Brady. His career may have overshadowed Askew's and a few others as well...

I also feel pretty certain that we'd be among the NCAA leaders for most players underperforming their ratings... sadly! Late Lloyd, RichRod, Hoke were not coaching dudes up as well as we would've hoped!!!!!!!!

One good thing to come out of this commodization of recruiting is the competition has made them have to do a better job on projecting college success, for survival's sake. So when a Derrick Green happens now, it's a shocker. Look back through some of our top 50 recruits on that list and you'll be scratching your heads to figure out how some could've been so highly ranked - indeed who some of them even are?!



February 3rd, 2017 at 8:43 PM ^

Unreal how stacked this class is. What's even better are the somewhat unknowns like St Juste. All indicators are that he was seriously underrated and unknown before this class was already winding down to signing day.


February 4th, 2017 at 11:31 AM ^

I'd recommend using R and the ggplot2 for static graphics also. You could make an mgoblog theme in ggplot and, after some initial effort, you'd be able to produce nice stuff fairly quickly. It's probably also worth checking out D3 and 538's github page. This site prides itself on being data-driven, so I think it would be worth the effort to pick these things up.