Rany Jazayerli's MLB draft study

My last post was partly about the 1984 Bill James draft study, which found that college players proved to be better choices than high-school players, and that hitters provided a better return on investment than pitchers.

In the comments, a couple of readers pointed me to separate studies by Philly and Rany Jazayerli. I haven't gone through Philly's analysis yet, but I did read through Jazayerli's.

It's a great 12-part series, that appeared at Baseball Prospectus over the course of a year or so. (Here's a link, as provided by commenter VKW – look for part 1 to part 12 in the article listing.) It's so thorough that I'm surprised it hasn't had more publicity (although maybe it has; I don't keep up as well as I used to).

If you don't have patience for all 12 articles, Part 11 is a summary of the main findings.

Basically, what Jazayerli found is that much of the gap between college players and high school players has disappeared. Why? Perhaps teams learned from experience, and from these types of studies. Over the past several years, the availability of high-school talent has increased (Jazayerli speculates it's that large signing bonuses are convincing young players to sign instead of going on to college). But the proportion of high-schoolers being drafted has stayed the same, or even dropped a bit. That means that teams are actually less likely to draft a given high-schooler than they were before. So they're concentrating more on the better ones, which increases the returns.

In the period covered by the study, college hitters are still the best drafting bet, but not as much as they used to be. Moreover, in the years since Moneyball revealed the "draft college players" strategy, teams have drafted so few high-schoolers that Jazayerli argues that they might now be worth 40% *more* than college players!

Lots of other good stuff in these studies ... I'm not sure what to make of the breakdowns by position. Some positions seem like better choices than others, but it seems to me that the sample sizes are pretty small compared to the variances between players.

In any case, I think this series joins the Bill James study as a must-read for anyone doing research in this area.