Monday, January 24, 2005
SPEAKING FOR THE DEFENSE
One of the more alarming circumstances this off-season for the baseball blogosphere is that Mitchel Lichtman's consultancy with the St. Louis Cardinals precludes him from publishing complete Ultimate Zone Rating figures for 2004. It's not that UZR is perfect, but it always seemed closer to perfect than what Davenport and Bill James have made of traditional fielding data. So an important piece of the puzzle is missing for looking at player defense.
We do have raw Zone Rating, of course, from ESPN.com, so that's a good thing. But, as is obvious, not every ball hit into a player's "zone" is equal, and every player at a position does not get an equal number of opportunities in each part of his zone. A good defender might have an artificially low ZR just because more difficult balls were hit into his zone than another player saw.
MGL, with UZR, attempts to rectify that to some degree, and also attach run values to each event. Without his figures, is ZR the only thing to look at?
Thanks to David Pinto, the answer is "no." Last year he unveiled a system somewhat similar to UZR called "The Probabilistic Model of Range." Pinto takes actual play-by-play data to determine the probability of a ball being fielded, i.e. a ball hit at such-and-such a speed to such-and-such a place at such-and-such a height will be caught 10% of the time, and this fielder had 100 balls hit there, and he caught 15 of them, so he made five more outs than the average.
Now, it's not converted to run figures like UZR, but that's still pretty good, and it will be interesting to see how these numbers look over time.
To fill the UZR Gap, Pinto is beginning to publish his PMR for 2004. He has started with team totals, and his calculations (which represent practically, but not quite all, of 2004 at this time) reveal what many of us witnessed and noticed in 2004: the Angel defense was fairly lousy. In fact, by this measure, the Angel defense was the worst in the game, making 90 less outs than would be predicted by where the balls were hit. I'd bet that a lot of these outs were hit to center, left, and third, so it will be interesting to see as Pinto publishes individual ratings in coming weeks.
90 outs is no small deal, by the way. An out is worth about -.28 or so runs against average, so that's about 25 runs given away right there. And that's not even considering that each non-out becomes a hit that has some degree of value, so it's even more. That's probably two or three wins we didn't have because of our fielding.
You can read all of Pinto's writings on defense here for more detail on his system, with more essential posts coming here and here. In the middle of that comes this entry on Father Time, Steve Finley, from over a year ago. Like UZR and Davenport, Pinto sees Finley as being below average (this is for 2003), and it actually breaks down where he wasn't catching balls that season. It's pretty interesting. Pretty much, don't hit line drives right at Steve Finley.
I'm certainly not saying this is perfect, but I appreciate any work that adds to the discussion, and with UZR out of the picture for now, Pinto's PMR might be the best thing (read: least imperfect thing) we can get our hands on for awhile.
One of the more alarming circumstances this off-season for the baseball blogosphere is that Mitchel Lichtman's consultancy with the St. Louis Cardinals precludes him from publishing complete Ultimate Zone Rating figures for 2004. It's not that UZR is perfect, but it always seemed closer to perfect than what Davenport and Bill James have made of traditional fielding data. So an important piece of the puzzle is missing for looking at player defense.
We do have raw Zone Rating, of course, from ESPN.com, so that's a good thing. But, as is obvious, not every ball hit into a player's "zone" is equal, and every player at a position does not get an equal number of opportunities in each part of his zone. A good defender might have an artificially low ZR just because more difficult balls were hit into his zone than another player saw.
MGL, with UZR, attempts to rectify that to some degree, and also attach run values to each event. Without his figures, is ZR the only thing to look at?
Thanks to David Pinto, the answer is "no." Last year he unveiled a system somewhat similar to UZR called "The Probabilistic Model of Range." Pinto takes actual play-by-play data to determine the probability of a ball being fielded, i.e. a ball hit at such-and-such a speed to such-and-such a place at such-and-such a height will be caught 10% of the time, and this fielder had 100 balls hit there, and he caught 15 of them, so he made five more outs than the average.
Now, it's not converted to run figures like UZR, but that's still pretty good, and it will be interesting to see how these numbers look over time.
To fill the UZR Gap, Pinto is beginning to publish his PMR for 2004. He has started with team totals, and his calculations (which represent practically, but not quite all, of 2004 at this time) reveal what many of us witnessed and noticed in 2004: the Angel defense was fairly lousy. In fact, by this measure, the Angel defense was the worst in the game, making 90 less outs than would be predicted by where the balls were hit. I'd bet that a lot of these outs were hit to center, left, and third, so it will be interesting to see as Pinto publishes individual ratings in coming weeks.
90 outs is no small deal, by the way. An out is worth about -.28 or so runs against average, so that's about 25 runs given away right there. And that's not even considering that each non-out becomes a hit that has some degree of value, so it's even more. That's probably two or three wins we didn't have because of our fielding.
You can read all of Pinto's writings on defense here for more detail on his system, with more essential posts coming here and here. In the middle of that comes this entry on Father Time, Steve Finley, from over a year ago. Like UZR and Davenport, Pinto sees Finley as being below average (this is for 2003), and it actually breaks down where he wasn't catching balls that season. It's pretty interesting. Pretty much, don't hit line drives right at Steve Finley.
I'm certainly not saying this is perfect, but I appreciate any work that adds to the discussion, and with UZR out of the picture for now, Pinto's PMR might be the best thing (read: least imperfect thing) we can get our hands on for awhile.
Comments:
Post a Comment