Because of the work I've done putting together my Minor League Splits site, I'm one of the few people who has complete play-by-play data for the 2006 Minor Leagues. There are many things I could be doing with it that I haven't done yet...but I just made a big leap on one of them.
Using a stat called Range (described here), I've figured out how to use my data and come up with defensive stats on minor leaguers. Basically, the stat uses a number of variables to estimate how many chances each position should have on a given team, then compares that to actual chances coverted. (For infielders, assists; for outfielders, putouts.)
As with any defensive stat, there plenty of limitations--small sample sizes, having to estimate the number of chances a player "should've" gotten, knowing nothing about positioning, etc.--but some data is definitely better than none. I've only looked at a few of the numbers so far (I haven't set things up for easy browsing yet), but they seem to line up reasonably well with scouting data.
On to the topic of the post. As you might imagine, the minor leaguer I was most interested in was Ryan Braun, since he's the Brewers uber-prospect, and his defense may have a lot to do with what position he plays in the majors. And, that position will have something to do with where Bill Hall might play in 2008.
Brevard County: In 58 games in the Florida State League, Braun's Range was +1. Basically average. The only worrisome thing: Alcides Escobar, Brevard County's shortstop, came in at -27. That's about as bad as it gets. Escobar has (I thought) a good defensive rep, so it could be that Escobar and Braun were positioned to make up for Braun's lackings...I don't know. But for whatever reason, the Manatee infield was cumulatively well below average.
Huntsville: In 57 games with Huntsville, Braun's Range was -28. That's shockingly bad, record-settingly bad over a full season. If taken literally, that means that an average 3B (not even a good one) would've turned one more grounder into an out every other game than Braun did. Ouch.
I have a theory, though. After the All-Star break, the Stars pitching staff had some really mature pitchers: Yovani Gallardo, Tim Dillard, and Steve Hammond. Range assumes that if, say, Southern League RH batters hit 25% of their grounders to third base, then Stars pitchers allowed the same amount. However, it seems plausible that overpowering (or very deceptive) pitchers don't allow as many balls to pull, so maybe Braun didn't get the chances because balls weren't hit to him, not because he's bad.
Scottsdale: Finally, in 14 or 15 games in the AFL, Braun has been at -3 plays. That's not very good, but in that size sample, who knows.
Putting it all together, Braun's at 30 plays below average in 130 games at 3B. Pretty bad, especially if you figure that Double-A average is worse than MLB average. And the whole story is a case study in the limitations of defensive stats. But it is a start.