the Polo Grounds

Back to the forum index
Linear

draft/recruiting trivia question

by Jay ⌂, San Diego, Monday, March 28, 2011, 12:13
edited by Jay, Monday, March 28, 2011, 12:35

Other people around the web have done studies on this before, but due to HR's awesome downloading ability I was able to quickly link up the Rivals recruiting data with NFL draft results, going back to the 2007 NFL draft. Anyway, here's the question. Feel free to guess.

What percentage of 1st-to-3rd round draftees (drafts 2007-2010 inclusive) were Rivals rated 3-stars or lower coming out of high school?

Tags:
draft

locked
  629 views

as another exercise

by HumanRobot @, Cybertron, Monday, March 28, 2011, 16:11 @ Jay

Could you compare the hit rate for the Top100 to the first three rounds? They're a similar size so that'd be a good comparison to draw.

locked
  446 views

are you a fellow PERL Monk now?

by HumanRobot @, Cybertron, Monday, March 28, 2011, 15:26 @ Jay

[ No text ]

locked
  439 views

Let's see ...

by Savage, Around Ye Olde Colonial College, Monday, March 28, 2011, 12:58 @ Jay

I'll take my chances with 25%-40%

Here's where I came up with my number:

There are approximately 200 schools that play reasonably high level football, each signs an average of 20 per year = 4000 players per year.

Approximately 250-300 kids are R4+*s each year.

So R4+*s make up about 300/4000 = 7.5% of each class.

Let's assume that a R4+* is 3x as likely as a non-R4+* to make it -- that brings us to 19.5%.
If we assume that a R4+* is 5x as likely as a non-R4+* to make it -- that brings us up to 28.8%

But even though it isn't supported by the raw independent trials type analysis, let's assume that some of the population can be thrown out. If we fudge a little to say that 1000 of those 4000 simply have no shot of going pro, and can be immediately eliminated from the pool of possible eventual draftpicks:
Let's assume that a R4+* is 3x as likely as a non-R4+* to make it -- that brings us to 25%.
If we assume that a R4+* is 5x as likely as a non-R4+* to make it -- that brings us up to 35.7%

I think that's reasonable. Anything beyond that would start to invalidate those who claim "stars don't matter!" (as though this doesn't in and of itself when looking at percentage of draftees)

locked
  516 views

I come up with 60%

by Jay ⌂, San Diego, Monday, March 28, 2011, 13:14 @ Savage
edited by Jay, Monday, March 28, 2011, 13:27

I'll show my work a little bit later. I will say that I find this interesting not from a "is Rivals f*cked or not" predictive kind of inquiry (I think Rivals does as good as job as anyone can at identifying high school talent). I think it's doubly interesting from the NFL point of view; the idea that draftable talent can come from anywhere (even Division IAA), and it isn't always manifest when a kid is an 18-year old senior in high school. There's a hell of a lot of development that can take place over the next four years.

locked
  510 views

I forgot to make my main point, which is

by Jay ⌂, San Diego, Monday, March 28, 2011, 14:37 @ Jay

"scoff at 3-star recruits at your own peril"

locked
  486 views

Isn't that like saying, "Scoff at 8 seeds and below

by Mike, Monday, March 28, 2011, 14:47 @ Jay

at your peril" based on Butler and VCU? The problem is identifying ex ante which of the low seeds/three stars will turn out to be Butler/Marcell Dareus.

locked
  466 views

I think it's more reinforcing

by Jay ⌂, San Diego, Monday, March 28, 2011, 15:16 @ Mike
edited by Jay, Monday, March 28, 2011, 15:37

(and this may be a blindingly obvious thing to point out) that the while the ratings services identify a certain select number of probable successful players, they will not identify the majority of successful players who will eventually rise out of the pack.

locked
  485 views

I get your point, I was just challenging its utility

by Mike, Monday, March 28, 2011, 18:38 @ Jay

There's no option to "bet the field" in recruiting. Each recruit is an individual wager, and the cost of each wager is the same (one scholarship). So even knowing that a majority of early round picks will not be 4/5 stars, you still can only offer individual 3 stars. Then the comparison becomes spending a scholarship on a 5-star prospect with a 20% chance of becoming a first-round pick, a 4-star with a 5% chance of becoming a first-round pick or a 3-star with a 1% chance of becoming a first-round pick (or whatever the numbers turn out to be).

locked
  443 views

oh, I don't think there's any real utility

by Jay ⌂, San Diego, Monday, March 28, 2011, 21:58 @ Mike

other than providing message board trivia.

locked
  443 views

sounds like "precision" versus "recall"

by HumanRobot @, Cybertron, Monday, March 28, 2011, 15:25 @ Jay

Rivals does a fine job with "precision" (the percentage of relevant documents of retrieved documents), but lacks when it comes to "recall" (the percentage of relevant documents which are retrieved).

http://en.wikipedia.org/wiki/Precision_and_recall

locked
  476 views

Can you carve that up by position?

by KGB, Monday, March 28, 2011, 15:24 @ Jay

It might be fairly uniform for "4's + 5's" versus "3 or fewer", but I wonder if further filtering might lend any insight, particularly as it relates to skill guys versus linemen.

locked
  452 views

yeah, I got all that shit

by Jay ⌂, San Diego, Monday, March 28, 2011, 15:27 @ KGB

It'll cost ya

locked
  472 views

What's with all the salty language today?

by PMan, The Banks of the Spokane River, Monday, March 28, 2011, 17:30 @ Jay

Did you just get back from the Magumba Bar in Drambui?

locked
  445 views

Well, I just want one rib.

by KGB, Monday, March 28, 2011, 15:32 @ Jay

Ah, fuck the cup. Jus' pour it in my hands for a dime.

locked
  456 views

I sho am hungry

by Dylan, Santa Barbara, CA, Monday, March 28, 2011, 17:17 @ KGB

[ No text ]

locked
  425 views

Your last point is the key

by BPH, San Diego, Monday, March 28, 2011, 13:09 @ Savage

The issue, as I see it, isn't the raw numbers, but the percentage of 5-star players who are drafted in rounds 1-3 versus the corresponding percentages for 4 and 3 star players. I believe I've read that the numbers are just as you'd expect: the percentage of 5 stars is significantly higher than the percentage of 4 stars, which is significantly higher than the percentage of 3 stars.

locked
  494 views

The article I posted below agrees (with charts).

by Savage, Around Ye Olde Colonial College, Monday, March 28, 2011, 13:14 @ BPH

[ No text ]

locked
  459 views

That's probably the one I'm remembering then

by BPH, San Diego, Monday, March 28, 2011, 13:15 @ Savage

[ No text ]

locked
  449 views

78%

by River, Hell of the Upside Down Sinners, Monday, March 28, 2011, 12:55 @ Jay

[ No text ]

locked
  477 views

79%

by Chris (HCC) @, Paradise, Monday, March 28, 2011, 12:59 @ River

Are we playing Price Is Right rules?

locked
  501 views

I knew I should have gone with 100%

by River, Hell of the Upside Down Sinners, Monday, March 28, 2011, 13:30 @ Chris (HCC)

[ No text ]

locked
  453 views

I was going to go with 45.1% for that reason.

by PMan, The Banks of the Spokane River, Monday, March 28, 2011, 13:01 @ Chris (HCC)

I hope I get to play the Mountain Climber game with the yodeling.

[image]

locked
  506 views

Plinko

by BPH, San Diego, Monday, March 28, 2011, 13:09 @ PMan

[ No text ]

locked
  453 views

ONE DOLLAR!

by NDTerp, I am not Jay. I never have been Jay., Monday, March 28, 2011, 13:09 @ PMan

[ No text ]

locked
  464 views

420

by Mike, Monday, March 28, 2011, 13:15 @ NDTerp

locked
  478 views

999?

by Chris (HCC) @, Paradise, Monday, March 28, 2011, 13:55 @ Mike

For a GD arbor (what the hell is an arbor).

locked
  458 views

45%

by Mike, Monday, March 28, 2011, 12:44 @ Jay

[ No text ]

locked
  480 views

higher

by Jay ⌂, San Diego, Monday, March 28, 2011, 12:46 @ Mike

[ No text ]

locked
  483 views

By the way, here's an old Doc Saturday examination of this

by Savage, Around Ye Olde Colonial College, Monday, March 28, 2011, 13:02 @ Jay

it's probably time to refresh that study

by Jay ⌂, San Diego, Monday, March 28, 2011, 13:37 @ Savage

Reliable Rivals data only goes back to 2002 (and 2002 is a bit spotty), so at the time, Hinton was looking at 3 drafts' worth of data (2006-2008, with the earliest year being somewhat unreliable).

I looked at Drafts 2007-2010 and the results are a bit different.

locked
  477 views
374368 Postings in 32685 Threads, 204 registered users, 87 users online (0 registered, 87 guests)
Contact
powered by my little forum