Posted by bad_luck on 2/29/2016 4:48:00 PM (view original):
The biggest problem with UZR (and DRS) is sample size. As long as you average out multiple years of data, it's pretty reliable. It's definitely better than range factor and fielding percentage (which suffer from the same sample size limitations).
But my example still stands as a potential flaw. A great CF will take balls away from his COFs and could lower their UZR, even though they may still be above average-to-elite defenders.
And I'll be honest, I have Markakis in mind here. He's a guy I've always felt advanced metrics have gotten wrong. I don't think he has elite range by any means, but watching him for years, I never recall thinking "so-and-so would have gotten to those balls". He got to everything, in my opinion, he should have gotten to. With Jones beside him, there was no reason for him to range into the gaps to make as many plays, but I still think he was capable of doing it.
As for fielding percentage - it shows you how reliable a player is on balls he gets to. I never once said it was the whole package. I would never take it in a vacuum, but it's useful.
2/29/2016 5:11 PM (edited)