A word about those defensive metrics October 30th, 2012, 11:21 am posted by JEFF FLETCHER, OCREGISTER.COM
Since the Gold Gloves are coming out later today, we are certain to hear a whole lot about what a farce they are. Well hear how the managers and coaches are terrible judges because they dont use UZR, while the traditionalists will say the sabermetrics guys should close their laptops and watch a game.
Really, both sides are wrong.
First, there is no question that managers and coaches do a poor job at this. Ive been around when the ballots come out, and it goes like this: Coaches sit at a table and say to each other Third base? Who won last year? Oh, hes good. Or else Im voting for Player X. Hes made a couple great plays against us. Coaches just dont see enough of their opponents to make informed judgments about defense. I will accept as the gospel their opinion of their own players defense, because they see them every day, but not what they think of opponents they see a few times a year.
The problem is that eyeballs are a bad tool to analyze just about anything on a baseball field, especially when you are talking about small sample sizes. Thats why stats are so handy.
But theres another problem. The stats arent any good either.
A few years ago I was chatting with Kevin Towers, who is one of the most well-rounded GMs in terms of merging the worlds of scouting and modern metrics. We were watching the Giants Fred Lewis, who was an interesting case because the numbers loved his defense, but anyone who watched him couldnt figure out why. I asked Towers about Lewis, and he said that the defensive stats that are out there for public consumption just dont work. Thats why many teams have their own proprietary metrics.
One of the problems with things like UZR is the data going in isnt scientific at all. Many of these defensive stats are generated based on people who sit in the press box with a grid in front of them and they record the spot on the field where the plays are made. But its far from precise. They are simply estimating the location on the field with their eyes, and their accuracy depends on the location of the press box and whether they saw a replay, among other things. And they dont measure how far the player went to get there, so good positioning can masquerade as good range.
But dont worry, the answer is out there. And its coming.
Imagine if the same technology used to track the exact location and speed of a pitch was used to track the exact location and speed of a shortstop. The people who brought you Pitch f/x are working on Field f/x. Its been in development for a few years, so hopefully it wont be long before this data is out there and available.
Continue reading here:
A word about those defensive metrics
=:76 
$2?5JVD 27E6C>2E9 0 5:DCFAE65 EC2:?D[ A=2?6D[ 3FD6D 2?5 76CC:6D[ 7=@@565 3F:=5:?8D[ 3=@4<65 C@25D 2?5