While taking a break from some yardwork over the weekend, I kicked back with the latest issue of ESPN The Magazine. It contained a short article on this paper which purports to find market inefficiency in the NFL draft. I haven't yet read the paper (it runs some 60 pages) but the magazine version suggests that two pieces of evidence are an overvaluation of the top pick and an undervaluation of later picks. (Pick 43 has had a good run over the last 10 years--9 of the 10 picks are still in the league and they include Julius Jones, Corey Dillon, and Muhsin Muhammad.)

The first part strikes me as plausible--I doubt Eli Manning is going to be so much better than Philip Rivers that the Giants should have forked over an extra first, third, and fifth round pick to get Manning. (Of course, one should measure marginal revenue product rather than the mere effect on wins--Eli might have a bigger effect on ticket sales than Rivers.)

The second part--at least as reported in the mag--is less convincing to me. Suppose that draft choices are binomial events--success or failure--and the the probability of success is close to 1 for high picks and declines as the draft progresses. It would be entirely possible that some spot in the draft (maybe pick number 43!) would seem to be a sweet spot because over some relatively short (maybe 10 years) period of time there had been a run of successful picks. Stated differently, the finding that 9 of the last 10 picks at spot 43 are still in the league is no reason to conclude that the probability of success does not decline as the draft progresses. (Note also that still in the league depends on flukey things like injuries, not just the efficiency of the draft.)

Again, I have not yet read the full paper so I might not (or more accurately, ESPN The Mag might not) be doing justice to the paper. After reading it, I might add more.