3 star mafia meets 2009 All-Americans

Submitted by TrppWlbrnID on

perhaps someone has taken a look at this previously, but with all the talk of recruit stars and future potential, i looked at the first and second SI all american teams and then tried to find what their rivals star rating was. the results are below.

first team
2) 5 star
9) 4 star
7) 3 star
5) 2 star
3) 0 star

second team
2) 5 star
8) 4 star
10) 3 star
2) 2star
2) 0

this is 50 total players (cj spiller was listed twice)
8% 5 star
34% 4 star
34% 3 star
14% 2 star
10% 0 star

anyway, i think the moral of the story is that over half of the first and second all-american teams are made up of guys 3 star and lower.

again, there is no correlation between all-americans and team wins, but hopefully this can help to dispel some of the "oh my god our recruits are not all 5 star players!"

there are a few guys that did not have a rivals profile and i could not find, perhaps they are no star or i just can find them. i will gladly edit if anyone has any additional info. these players are dennis pitta, te byu, jake kirkpatrick c tcu, grant ressle k mizz, luke keulchy, lb bc, matt dodge k ecu.

Blue_Bull_Run

December 20th, 2009 at 12:57 PM ^

That was exactly my first thought, too.

Really, what you need to do is this:

(Number of x-star players on the All-American team) / (Total number of x-star players that rivals rated)

So, for example, there are 25 5*/year, or a total of 100 in college football right now (or is it 125?). Anyways, 4 of them made the SI team. So you can say a 5* recruit has a 4% chance of being an All-American. (You may want to further refine the sample to exclude freshmen and sophomores)

I suspect that 4% chance is significantly higher than what a 3 star recruit could expect.

Blue_Bull_Run

December 20th, 2009 at 1:05 PM ^

cbuswolverine lists the total number of recruits listed by Rivals, below:

Rivals lists 23 five stars, 387 four stars, and 1537 three stars for 2010. I don't worry about stars too much myself, just sayin'.

Let's assume that total is the same each year. Then the chances of a recruit making the SI All-American team are:

5-star: (4)/(4*23) = 4.3%
4-star: (17)/(4*387) = 1.09%
3-star: (17)/(4*1537) = 0.27%

So really, you've just made a great argument that 5 star players are superior.

EDIT: damn, I think someone beat me to the point. Whipped out my calculator for nothing, man!!

osdihg

December 20th, 2009 at 9:21 AM ^

But also keep in mind that good punters/kickers can get 2 or 3 stars, which may account for some of your 2 stars.

For example, Zoltan Mesko was a 3* on Rivals, which is inconceivable because he should have infinity stars. Pathetic humans, they know not what they do!

cbuswolverine

December 20th, 2009 at 9:24 AM ^

The moral of the story is really that there are about 65 times more three stars than there are five stars. Rivals lists 23 five stars, 387 four stars, and 1537 three stars for 2010. I don't worry about stars too much myself, just sayin'.

*edit* my pony is slow

bacon

December 20th, 2009 at 9:57 AM ^

Your numbers suggest that at face value, the OPs original argument is incorrect.

5 stars make up 1.2% of the total, yet have 8% of the representation. That means there is an advantage to being a 5 star (duh). 4 stars made up 19.9% of the total and 34% of the all americans, again a selective advantage to being a 4 star. Finally the 3 stars were 78.9% of the original pool and only 34% of the all americans. I think that we can at least say there's some advantage to having a higher star rating.

I'm not a stats person, but I believe that when you normalize the data, 5 stars are 6.7 fold over-represented in the all american pool than what would be expected if there was no bias, 4 stars are 1.7 fold over represented. 3 stars are in fact pretty under represented (there should be 39- 3 star all americans, there are only 17).

BUT, I think that you could make the argument that the advantage for having higher stars is not as great as you might imagine, depending on what you expected.

clarkiefromcanada

December 20th, 2009 at 12:01 PM ^

I am also not a stats guy but I retain some of them when we do quantitative research in my clinical work. It is good to source out expertise so maybe some stats guys can clear this difficulty up for me.

I am wondering if taking the pure values you use looking at 6.7x over-representation for 5 stars as "all american" assumes the pool of talent is "equal" at the point of entry to University. Obviously, this is not the case and 5 star players would be (by the 3 star mafia and fans more generally) expected to maintain that rating/talent advantage over the course of four years (thus simply transitioning from Under Armor "All American" to plain old NCAA "All American"). This, of course, doesn't happen either as players develop differently and 2/3 stars mature etc. but the *idea* of recruiting 5 star players is that they have a built in talent advantage that is potentially generalizable over 4 years. In practice (Terrelle Pryor) this is certainly not so much.

So the question I have would relate to how do you normalize all of the talent entering to enable a statistically reasonable comparison? A second question would relate to the stars themselves and given the limited number of rating contributors is there really any validity there or is it just to sell subscriptions and increase Sunday morning debate on blogs?

Thanks for the help.

bacon

December 20th, 2009 at 4:51 PM ^

Ah, a fellow scientist? Sorry for the length of this response.

Full disclosure, I should point out that my data set is actually quite flawed as the numbers I used for the total 5, 4, and 3 star breakdown were the ones provided by cbuswolverine (thank you) and the numbers for the number of all-americans came from the OP (thank you too). At least three potential problems that I didn't think about when I was making my post that should be cleared up:
1. the number of total 5, 4, and 3 players in the entry pool should be multi-year since I assume that more than the senior class is considered when making the all-american list.
2. the numbers should come from the actual classes that the current all americans came from, not from this year (I'm not sure where they came from to be honest).
3. There's no accounting for the total pool of players (only the 5, 4, and 3 stars, so we I'm not sure that we could really say that the 3 stars don't have an advantage over the 2 stars or etc.

That said, IMO the 6.7-fold increase does in fact imply that the talent pool is not equal because if it wasn't there would be no selective advantage for being a 5 star over a 3 star. The only reason I even put it this way is to have some quantitative measure of the idea that the pool isn't equal.

One other thing that should also be considered (and obviously there are exceptions to this) is that kids with more talent out of high school theoretically benefit from "being" on better teams, with "better" coaching and have more of a chance to play on teams that get them recognized as all-americans. IMO, that also means that the 5 star should have a better chance of being chosen as an all-american down the road than a 3 star, so the playing field is inherently biased towards the 5 star and any normalization would have to take that into account (maybe there's at least a way to normalize for the fact that 5 stars are more likely to play on high-profile teams/games). In retrospect, I think this may have been part of what you were getting at in your second paragraph.

I think that you raise some interesting points that should be considered by someone with more skill and better data than I. Ultimately though, I think that given the amount of variability in this process, it's kind of crazy to assume that the number of stars that a kid has coming out of high-school either ensures or limits the amount of success that he will have as a college player. It's obviously just at best going to be someone's best guess, and here your last question is most pertinent because at the end of the day it's somebody's best guess at how good someone is.

clarkiefromcanada

December 20th, 2009 at 9:26 PM ^

Thanks for clarifying some of that; I am also a research scientist outside my day to day clinical work. My sense was that normalizing the entry data for any sort of meaningful comparison would require an incredible accounting of a broad range of factors that high end statisticians would have to take on. I have sat in many meetings regarding this sort of thing when we look at varied sample recruitment (multi demographic factors, illness conditions, durations etc.) and the stats guys try to clean that up...this is way more complex. I am thinking that if it was possible then maybe recruiting institutions/Scout/Rivals/Takkle/ESPN etc. would have wanted to avail themselves of that sort of data.

The implications would be pretty interesting if you could deduce the statistical advantage of, say, taking 3 star recruits from a specific region if they gave a measurable over-representation for becoming "All American" or getting drafted to the NFL etc. One wonders if the significant focus on the "State of Miami" (referencing Howard Schnellenberger) by current Michigan coaches relates to data analysis or simply the measurable volume of high end recruits in that area.

The more I think about this the more challenging it seems. Further help from pro stats guys?

mgoblahhh

December 20th, 2009 at 9:38 AM ^

Don't worry about the star rating, a lot of these kids come from small schools and do not get a lot of exposure. Take Ray Vanopil he is a STUD! He is only rated a 2 star, I saw him play twice this year and other than being a little small size he has the speed, a nose for the ball, and he is a winner ! I guess I am trying to say; trust the coaches and their ability to identify talent.

Tater

December 20th, 2009 at 9:41 AM ^

No matter what minor exceptions or mitigating factors may exist, the trend is obvious: star ratings are a decent predictor, but they don't mean as much as heart and good coaching.

Coaches know what kind of people they want on their teams, and build teams according to self-made "templates" that only they know. Because of the record the last two years, some are having a hard time with it, but there is an underlying issue here: RR didn't suddenly lose thirty points from his IQ when he moved to Ann Arbor.

Keep in mind, too, that coaches don't always get every player they want. I sometimes detect a tone in RR critics that seems to assume that RR is ignoring four and five stars at positions of need for random three stars at positions that are already full. Does anyone really think that a coach who has been successful everywhere he has been could succeed that way?

What it comes down to for me is that RR sees something in every recruit he offers; if RR didn't think a recruit could perform at UM, he wouldn't extend an offer. That is good enough for me, and I welcome every recruit who signs with UM, no matter what his star rating.

Simi Maquoketa

December 20th, 2009 at 10:33 AM ^

We gonna find out how much "heart and coaching" Rodriguez and his boys have this season, eh?

He's got guys like you really enamored. I don't know if it's TDoR's ability to rationalize or just hope like hell we couldn't have made a mistake--but I've never seen so much leeway granted to a guy (HE knows what he sees in recruits--that Ol' Black Magic Rodriguez has) who has shown zilch since he got here.

We'd be laughing at MSU if they had most of the guys in this class.

clarkiefromcanada

December 20th, 2009 at 12:08 PM ^

That sort of vitriol is pretty funny. I realize you don't much like the pro RichRod crowd...it's evident in your posts but a quick perusal of the Little Brother recruiting class does not inspire a great deal of awe...

http://recruiting.scout.com/a.z?s=73&p=9&c=8&toinid=684&yr=2010

You are correct in noting RichRod hasn't shown anything since his arrival but you give a coach his Seniors to show competency. If there is nothing starting next season it's going to be a long half decade in the 10's.

As an aside, the Little Brother would absolutely take most/all of the guys in the M class this year (and every year for that matter).

Carcajous

December 20th, 2009 at 9:48 AM ^

Assume 25 5-stars, 350 4-stars. and 1500 3 stars (just for argument's sake... I'm trusting cbuswolverine here).

If the AA teams are Jr. and Seniors (they are, mostly), if you are a 5-star, you had a 4/50 chance of being on the 1st or second team this year as a Jr./Sr. That's 8%.

If you were a 4-star, you had a 17/700 chance. That's 2.43%

If you were a 3-star (yikes), you had a 17/3000 chance. That's a .57% chance of being on the first or second team.

Of course none of this takes into account that you have another shot at it as a senior if you don't get it as a junior, but that would just magnify the differences. Also, this is depended on "in-the-head" math over bad coffee.... so whatever.

In other words, the moral of the story is that is you are a 3-star, the odds are vanishingly small that you will be an AA.

Moe Greene

December 20th, 2009 at 9:44 AM ^

I've never understood the rationale for why a player gets 5 or 4 or 3 stars. Then again, I doubt these folks are open about their methodology.

I'd also imagine it's highly endogenous - did DG's rankings go from 4 to 5 because of his recruiting, or was it just his performance?

This, of course, is yet another reason why I refuse to apply for front office positions for the Lions. I *suspect* I could do better....but what the hell do I know?

befuggled

December 20th, 2009 at 11:36 AM ^

But that's setting the bar pretty damn low, eh?

I'm not pretending to be a great evaluator of talent. I am however saying that I would have made the offensive and defensive lines my emphasis, and that I would not have drafted a wide receiver in teh first round four years in a row.

mejunglechop

December 20th, 2009 at 9:55 AM ^

A million thankz to you, this is totally the most earth shattering study ever. 2 stars have more all-Americans than 5 stars!!! It makes you wonder why people bother with 5 stars at all. Furhter, I think it really shows how much of a genius RRod is that our entier class except one is made up of 3 and 4 stars. Your insight is truly invaluable.

http://www.sundaymorningqb.com/2008/3/17/71811/4082

pistonsfan133

December 20th, 2009 at 10:12 AM ^

You also have to consider the competetition might be why a guy is an All-American, if he can put up amazing numbers on lower competition then he could be an all-american, but the recruiting sites see him at all of the camps and stuff against the top competetition, which could be why a guy isnt a 4 or 5 star

TrppWlbrnID

December 20th, 2009 at 10:27 AM ^

rrrrer (cat noise)

the point was not to say 5 stars are worthless, our own brandon graham was a 5 star and proved it all year long. the point was to try to settle down the folks who flip out about guys in this class not having as high a star level as some guys in other school's classes. i thought that point was pretty obvious, but that backing that point up with a small piece of factual evidence that showed that recruiting ranks did not automatically equal all american success. there are some good points above about the tendency of 5 stars guys to be AAs, but i think that this does back up the overall point.

mejunglechop

December 20th, 2009 at 12:52 PM ^

Nobody was arguing that star ratings guarantee a level of success. But if being voted an AA is your metric they are very very strongly correlated. I encourage you to read the link I posted above, it concludes that when you exclude kickers and punters, 5* are ~3x as likely to receive an AA vote as 4*s, 4* are ~2x a better bet than 3* and 3* are ~2x better than 2* or lower.

ThWard

December 20th, 2009 at 12:34 PM ^

(or induce someone into posting the beaten dead horse pic - which has, oddly enough, been beaten like a dead horse - my head hurts), but I suggest checking out the old Doc Sat's breakdown of this issue a few years ago. The relevant breakdown was actually posted on this site a week or so ago.

Again - yes, 3 stars can become AAs. There are so many more 3 stars than 4/5 stars that it's damn near inevitable that more AAs will have been 3 stars.

Let's just trust our coaches talent evaluation and development, etc., etc.

/Pat White was a 3 star'd

Blue Balls

December 20th, 2009 at 4:16 PM ^

players from 2004-present. Then and only then, argue that you need all these 4-5 star players to succeed-please. ND is going to be a force, like it or not. If Michigan Fans keep the knots out of their shorts and give Coach Rod time to get the players he "needs", Michigan will be once again feared on the field. Wasn't it the "football guru's" that allowed Tom Brady to be taken in the 5th round of the draft-certainly rivals has their shit together more than NFL scouts? The "three star Mafia", now that's funny. Go Blue!