Even the stars misalign. A lot.

www.postandcourier.com/article...n-many-failures

LOT of truth in that article.

The day anyone can predict whether injuries or off the field issues are going to occur by looking at a 16-year old is the day that these guys become millionaires. Head coaches fail just as much as these guys do.

That being said, if you line up the prospects from 1 to 3000 each year and set equal divisions, there is a huge correlation between success rate and those rankings. There is no group that approaches 100% success, but that doesn't mean the data isn't valuable or that the science is inaccurate. Folks just have unfair expectations of what they should say. Unlike basketball, football is a violent sport which requires college kids to stay eligible, healthy and out of trouble for four years. That is a HUGE source of error for these analyses. If this part of the data could be removed (all the kids who get hurt), I would bet you would not only get a good correlation between predicted (normalized) and actual success, but your predicted success would come much closer to 100% for the highest rated.

I look at the recent 4*'s (and higher) that UNC got and only a handful had full careers that were not cut short by injury, grades, or some off the field issue that shortened their eligibility, that didn't have a good amount of success: Mike Paulus, Cam Sexton, Christian Wilson, DPM, Herman Davidson, Jhay Boyd--that's really about it. Even still, there are many more accurate predictions (particularly in the division that is known as the 2* ranking and comprises the lower 65% of the entire population). There obviously are some guys in that group who have success (and sometimes them being passed over is very explainable when they are 17: maybe new to football aka a project) but seems like that is where the focus is on--not the fact that the rankings services predict that they will NOT be successful and 95% are not. I think all of it should be analyzed instead of pointing out your favorite example of a 5* who didn't pan out or a 2* who exceeds everyone's expectations.

I love you, too, Cornbread.

---------------------------------------------
--- bucinor wrote:

I love you, too, Cornbread.

---------------------------------------------

Did you want to discuss the article? Finding yet another article that says the stars aren't perfect is nice and all, but there are reasons that the stars mis-align. The writer hints at it, but doesn't do an adequate job (IMO) of explaining, for example, why football recruiting rankings have complicating factors such as a higher risk to injury and higher risk to off the field issues due to the requirement to stay in school for 3 or 4 years. It's very relevant if you want to have a true analysis of the value or accuracy of the rankings. Otherwsie, it's just saying what we all know--that the rankings show correlation but their predictive power is not close to 100% for any of their groupings.

from the article ... kind of incredible:


An analysis of Rivals.com’s top 100 players from 2006 to 2009 reveals that 42 percent of the nation’s top prospects became busts, meaning they failed to either play in 40 games, start 20 games or have one above-average season in their college careers.
.
.
.

In contrast, only 14 percent of those 100 prospects have been, or are projected to be, first- or second-round picks in the NFL draft.

---------------------------------------------
--- BethelRegiment wrote:

from the article ... kind of incredible:


An analysis of Rivals.com’s top 100 players from 2006 to 2009 reveals that 42 percent of the nation’s top prospects became busts, meaning they failed to either play in 40 games, start 20 games or have one above-average season in their college careers.
.
.
.

In contrast, only 14 percent of those 100 prospects have been, or are projected to be, first- or second-round picks in the NFL draft.

---------------------------------------------

What is incredible? That "only" 14 have this definition of success or that they can predict 9 of the 64 (out of the pool of the hundreds of thousands of HS kids who play football) when they are 16-17 years old...six years from the draft? Top two rounds of the draft (64 players) seems like an odd choice for defining success when you are starting with 100 players by the way. There seems like there would be a cap of success at 64%. The first three rounds is usually the choice of most due to the approximate equality of the two sample sets. However, since there is a significant portion of the top 100 that will get injured, go play MLB or something, or never complete his eligibiltiy due to off the field issues, I would think it would be possible to get 100% predictive success if you remove those that shouldn't count from the analysis (unless you feel predicting those kinds of things like injuries, grades, or legal issues should be part of the analysis). There might be 64 left from that 100 who stay healthy, eligible and out of trouble.

But I agree with Luginbill. The vast majority of the high ranking kids that don't pan out are completely unrelated to on-field performance. You see this with UNC's recruits and while I don't follow any other school nearly as closely, many of the NCHSAA kids who have not panned out have failed to do so due to off-field issues. Perhaps some of those reasons should be captured in their rating if you feel there should be an element of personality or psychological rating, but obviously some of them cannot be predicted like injuries.

These discussions also seem to always dwell on the top echelon instead of the entire group of data and specifically the lower echelon. My favorite argument is that sometimes folks point out that there are more 2* kids on an all-conference team than 5*'s as evidence as to how the star system is backwards... I don't know why the recruiting services don't use a more linear rating scale to eliminate this faulty logic, but if they did, it would take care of this immediately. To put it in perspective, it they used an evenly distributed 10 star system, the group of 10 star recruits (i.e. the top 10%) would include all of the 5*'s and pretty close to all of the 4*'s. The "9* group" would include the remaining 4*'s and the highest rated 3*'s whereas the 8 and 7 star groups would encompass the remaining 3*'s and the highest 2*'s. That would be pretty much 95% of the "successful" players with a large chunk of the remaining successes being from the group of players who were projects or walk-ons or some other subset that eluded the ranking services.

Rivals, scout, and ESPN grade these prospects according to stars, and we can see the hierarchy in which they're rated, but I wonder how Coach Fedora rates them. Which "3" star does he consider a "4" star, and so on. Makes you wonder which recruits Coach is really excited about. Honestly, I trust his evaluation over ESPN's.

---------------------------------------------
--- redhotchiliheel wrote:

Honestly, I trust his evaluation over ESPN's.

---------------------------------------------

Yeah, I do too, but those evaluations aren't mutually exclusive, right, and I would argue that the services rely HEAVILY on the input of coaches like Fedora? The evaluations of the coaches at the two dozen schools within 500 miles of Chapel Hill are the ones who make me first think that a kid is a quality recruit. They know a lot more than any of us and more than any of the recruiting analysts. I mean, I'm pretty sure that Mark Richt knows what a good tailback looks like and there is a reason that everyone wanted Todd Gurley and Keith Marshall but very few wanted some of the other tailbacks in NC in that class. Coincidentally, there is a reason that those two kids had good star ratings. They could each blow out a knee or get kicked out of school or flunk out tomorrow and never play football again...to me, that wouldn't mean the services were wrong.

I love the diamonds in the rough. Those are the ones that are truly hungry and to prove people wrong.