Add new comment

Well, it means the stat is more consistent year-to-year than you'd think, and is more accurate the higher the sample size.
It doesn't neccesarilly mean that the stat accurately projects defensive ability (it means its relatively consistent in what it is measuring). UZR is less predictive than a stat like OBP or SLG, and neither of those stats are as predictable year-to-year as you'd think.
I'm a big believer in using several defensive stats+scouting to evaluate defenders. It gets rid of some of the outliers you have year-to-year when you're relying on a single stat.
If SABRMatt would make PCA public, that would help a lot in the discussion as well. Currently there aren't any metrics that evaluate individual defenders on a team basis (as oppossed to zone rating).