from Amazon
Friday, December 05, 2008
The state of fielding sabermetrics in MLB
There’s a lot of discussion about defense and defensive statistics coming into vogue, especially with Peter Gammons stating that teams are realizing that the negatives of defense take away from even big positives on offense. Manny Ramirez, he’s looking at you. Adam Dunn? You too.
But while people are crowing about Zone Rating or even Dave Pinto’s Probabilistic Model, they’re missing something. If you think teams don’t read the studies and stats out here, you’re wrong. They’ve done that for years. What most people don’t realize is that while there’s some great minds writing about baseball, there’s some great minds inside the game too. There are the names you know — Eddie Epstein, Ben Baumer, Keith Woolner — and more names that you don’t.
There have been occasions where I’ve been privy to some of that work that’s going on in front offices and it simply blows away things you’d call the state of the art in sabermetrics. Not a little ahead — a lot. Some of it leaks out, some doesn’t. Some is absolutely horrible too, while some is ignored. Just don’t think that teams are using the same things you are. Anyone that thinks that, isn’t thinking.
One of his readers replied:
Your post just said, “There are some smart people in baseball, TRUST me. I can’t say anymore, but it would BLOW YOUR MIND.”
One of the best sabermetrician out there, if not the best, is Tom Tippett (with the Redsox). And his fielding work is on par with MGL. One of the other best is Dan Fox (Pirates), and his published fielding work is below that of MGL (and only because of the limited data he uses; presumably he has access to the same data as MGL and Tom now, and so, his fielding work will soon be/is on par with theirs).
Fielding systems, once they all have access to the same data, are all pretty simple. It’s all a matter of certain presumptions you make of that data, and how you decide to apply those adjustments. I’ve always advocated the “SAFE” approach (continuous function), similar to what Gabe did at ProTrade, combined with a WOWY approach. You put all of us in a room, and in a matter of days, we’ll all come out with practically the same system. And when HITf/x comes out, we’ll be in even more agreement (after two weeks). Really, half our job is trying to infer something from the data that hasn’t been recorded. HOW that inference is done is what differentiates MGL from Pinto from Jensen et al. But, with the eventual HITf/x (and GPS of fielders), practically ALL that inference goes away.
So, I don’t accept Will’s perspective on this issue at all. Sorry Will.
Recent comments
Older comments
Page 1 of 344 pages 1 2 3 > Last »Complete Archive – By Category
Complete Archive – By Date