Statisticians and economists weigh in on predicting NFL QB success from college

I was wondering what economists and statisticians had to say about the best predictors of NFL success from college players.  My depressing conclusion after the jump..

I had vaguely remembered that the only real indicator of NFL success was this Parcells theory of college success:  did a guy start for 3 years, did he have 27 wins, and something else.  But I was wondering what evidence lay behind this.  Sadly, the answer is none.  As a guy with a Ph.D. I was happy to dive back into some academic texts.  But let me just provide the top few and their key conclusions, and some references for the nerds out there:

1) Malcolm Gladwell New Yorker Article

This is more of an easy reading intro to the topic.  It raises the general idea (note that it is long and has a bunch of stuff about teachers in it).  I remember reading it when it came out, but it was all the more poignant to re-read relative to the Redskins because it talked about a) how there where 5 highly rated QBs in McNabb's class, but he was the only one with success and b) it features a scout talking about how awesome Chase Daniels was going to be in the NFL (sorta).

2) Berri-Simmons Article

This is the current reigning champ of the statistical analysis of how to predict QB success from college and combine performance.  It had four main conclusions from my standpoint: 1) if you grade QBs on cumulative things like games started or total yards, high draft picks rack up more of these (duh) than low ones.  However, 2) If you look at things like value added per play or yards per play or more performance based metrics, there is an extremely weak correlation to draft position.  Additionally, 3) NFL draft position is disproportionately influenced by combine performance, specifically height, BMI, Wonderlich, and 40 yard dash, but 4) combine performance has virtually zero correlation with actual performance.

So my plain English recap is: 1) GMs fall in love with taller, faster, smarter prospects and draft them higher 2) these prospects have no better chances than others of succeeding 3) After drafting a QB that they picked, they play him a lot, but 4) he has no better chance of succeeding than some random guy picked behind him.  

3) A yet more academic treatment of the topic, but with excellent bibliography to follow further if you are interested

This one does conclude that on average 1st round picks perform better than 7th round ones (duh!) but that there is no correlation between any indicator of college performance, or combine performance, and success.  The implicit suggestion is that NFL scouts/teams are able to use other info (interviews, etc.) to figure out that one group should probably go early and another group (of QBs) should go later.  However, the colloråry to this 'success of scouts theory'  that is that the relationship is not that strong - e.g. this is true when comparing 1st round vs. 7th round success, but that if 6 QBs were picked in the 1st round, a dartboard would be as good a predictor as actual draft position in the 1st round of success (not quite, but you get my point).

Anyway I found it interesting, but sobering.  One word summary:  crapshoot.  Or maybe better, lottery ticket.  Let's hope we get lucky (no pun intended), but for god's sake draft whoever is left, and don't trade up for your guy, unless he is a once in a generation guy like Ryan Leaf Andrew Luck.