Nick's defense..UZR

[Frobby]

I have a hard time with the idea that a full season is not long enough for a stat to do a decent job of measuring how well a player has played defense that year. I don't really buy the "small sample size" excuse for UZR. Sure, a player's defense can have some volatility in a small sample, but over 162 games, there's plenty of basis to judge how good a fielder is. And UZR just departs from what I see too often.

I think this problem with measuring park effects seems like the biggest problem with UZR. However, it's a little hard to be sure since there is almost no information available about how these are done.

[DrungoHazewood]

I have a hard time with the idea that a full season is not long enough for a stat to do a decent job of measuring how well a player has played defense that year. I don't really buy the "small sample size" excuse for UZR. Sure, a player's defense can have some volatility in a small sample, but over 162 games, there's plenty of basis to judge how good a fielder is. And UZR just departs from what I see too often.

I think this problem with measuring park effects seems like the biggest problem with UZR. However, it's a little hard to be sure since there is almost no information available about how these are done.

Brooks Robinson had as many as 736 plate appearances in a season, almost all of which were against legitimate major league pitchers doing their level best to get him out.

He never had more than 563 defensive chances in a season, and I'd guess than the majority of those were easy two-hoppers that he threw the guy out at first by seven steps.

You really think measuring those two things should have the same level of uncertainty?

[Frobby]

Brooks Robinson had as many as 736 plate appearances in a season, almost all of which were against legitimate major league pitchers doing their level best to get him out.

He never had more than 563 defensive chances in a season, and I'd guess than the majority of those were easy two-hoppers that he threw the guy out at first by seven steps.

You really think measuring those two things should have the same level of uncertainty?

No, you are misunderstanding me. Obviously with fewer occurences there is a bigger chance that the player's performance won't reflect his true ability. But, I think a good statistic still should be able measure what the player did with the opportunities he had with reasonable accuracy.

By the way, plate appearances and total chances are not analogous. If a ball gets past the 3B and it's not an error, there is no "chance." I don't want to get side-tracked on that, but Brooks probably had a lot more than 563 balls come his way the year he had 563 "chances."

[DrungoHazewood]

No, you are misunderstanding me. Obviously with fewer occurences there is a bigger chance that the player's performance won't reflect his true ability. But, I think a good statistic still should be able measure what the player did with the opportunities he had with reasonable accuracy.

I think that what's reasonable for one metric and one set of abilities may not be reasonable for another. What's reasonable for batters, where it's you against a wide variety of pitchers, may not be reasonable for defense, where you're one of nine guys interacting with the pitcher and the park and your teammates and all the biases they bring.

We have a lot of faith in offensive measurements, and expect everything else to fall in line in terms of consistency and repeatability. But it may be that defense just doesn't fit in the same box. And right now it looks like the uncertainty on defense is about twice that of offense. That's just how it is.

By the way, plate appearances and total chances are not analogous. If a ball gets past the 3B and it's not an error, there is no "chance." I don't want to get side-tracked on that, but Brooks probably had a lot more than 563 balls come his way the year he had 563 "chances."

Sure, it was just something pulled out of the air to make a point.

[Frobby]

I think that what's reasonable for one metric and one set of abilities may not be reasonable for another. What's reasonable for batters, where it's you against a wide variety of pitchers, may not be reasonable for defense, where you're one of nine guys interacting with the pitcher and the park and your teammates and all the biases they bring.

We have a lot of faith in offensive measurements, and expect everything else to fall in line in terms of consistency and repeatability. But it may be that defense just doesn't fit in the same box. And right now it looks like the uncertainty on defense is about twice that of offense. That's just how it is.

I don't really disagree with anything you say here. I still think that UZR isn't doing a good job with park effects, but since I really don't know what they do or how, it's hard to evaluate. I hope more information becomes available on that subject.

[El Gordo]

No, you are misunderstanding me. Obviously with fewer occurences there is a bigger chance that the player's performance won't reflect his true ability. But, I think a good statistic still should be able measure what the player did with the opportunities he had with reasonable accuracy.

By the way, plate appearances and total chances are not analogous. If a ball gets past the 3B and it's not an error, there is no "chance." I don't want to get side-tracked on that, but Brooks probably had a lot more than 563 balls come his way the year he had 563 "chances."

In the case of a 3B like Brooks e.g., who is playing 1B could have a huge effect on his number of assists. The quality and landscaping of his home field. The nature of his pitching staff(LH/RH, FB vs GB) for any given season. Luck. Plenty of variables difficult to measure.

[KAZ97]

I think that what's reasonable for one metric and one set of abilities may not be reasonable for another. What's reasonable for batters, where it's you against a wide variety of pitchers, may not be reasonable for defense, where you're one of nine guys interacting with the pitcher and the park and your teammates and all the biases they bring.

We have a lot of faith in offensive measurements, and expect everything else to fall in line in terms of consistency and repeatability. But it may be that defense just doesn't fit in the same box. And right now it looks like the uncertainty on defense is about twice that of offense. That's just how it is.

In the end, what we want to know is how different the outcome would be if player X is in the field instead of player Y and the same ball is hit towards the position. The classic problem of identifying a counterfactual.

UZR appears to use some "average" over a large number of balls hit to that zone to approximate the counterfacutal. That is one approach; the beneift being its easy, the downside being its precision.

IMO, alot of the sabermetrics crowd does statistics a dis-service by using "consistency and repeatability" as the litmus test of a statistic. Why should a metric be "repeatable" if the underlying events are rarely ever the same?

Brooks Robinson taught us that.

(re: his defensive performance in the 1970 world series):

You just never know. I just happened to be in the right spot in that series. I tell people that I played 23 seasons and I never did have five games in a row like I did in that World Series. As an infielder you can go a week or two and never get a chance to do something spectacular. In this series, every game I had a chance to do something outstanding defensively

[El Gordo]

In the end, what we want to know is how different the outcome would be if player X is in the field instead of player Y and the same ball is hit towards the position. The classic problem of identifying a counterfactual.

UZR appears to use some "average" over a large number of balls hit to that zone to approximate the counterfacutal. That is one approach; the beneift being its easy, the downside being its precision.

IMO, alot of the sabermetrics crowd does statistics a dis-service by using "consistency and repeatability" as the litmus test of a statistic. Why should a metric be "repeatable" if the underlying events are rarely ever the same?

Brooks Robinson taught us that.

(re: his defensive performance in the 1970 world series):

I watched Brooks like a hawk during the his career. The same way I watch Nick and Pie today. Every play he made in that WS, I saw him make frequently during a number of his seasons. Those plays simply managed to demonstrate the extent of his range, but over an unusually short period of time. As to Player X vs player Y on the same type ball hit to them, the FB tells me that.

[Frobby]

UZR doesn't say that Adam Dunn is better than Nick. It says that based on 2010 data Adam Dunn has been worth more (compared to an average first baseman) than Nick has (compared to an average right fielder).

On the subject of Dunn, almost every time I watch him play he does something poorly. Last week I went to a game and Strasburg fielded a dribbler and made a high throw that would have required Dunn to jump perhaps 2 inches off the ground to make the play. Instead the ball ticks off his glove and the play ends up with runners on 2nd and 3rd. I know it's anecdotal, but the play was so poor as to be astonishing. And I see this stuff all the time from him. He has no reaction time, and no throwing skill. For example, he's played 118 games and has started 3 DP's all year on grounders hit to him.

[El Gordo]

On the subject of Dunn, almost every time I watch him play he does something poorly. Last week I went to a game and Strasburg fielded a dribbler and made a high throw that would have required Dunn to jump perhaps 2 inches off the ground to make the play. Instead the ball ticks off his glove and the play ends up with runners on 2nd and 3rd. I know it's anecdotal, but the play was so poor as to be astonishing. And I see this stuff all the time from him. He has no reaction time, and no throwing skill. For example, he's played 118 games and has started 3 DP's all year on grounders hit to him.

The Fielding Bible has Dunn at a -12 at 1B up from a -24 last season. By comparison Wiggi this season is a +2. But this isn't taking into account their relative abilities to make difficult picks.

[DrungoHazewood]

On the subject of Dunn, almost every time I watch him play he does something poorly. Last week I went to a game and Strasburg fielded a dribbler and made a high throw that would have required Dunn to jump perhaps 2 inches off the ground to make the play. Instead the ball ticks off his glove and the play ends up with runners on 2nd and 3rd. I know it's anecdotal, but the play was so poor as to be astonishing. And I see this stuff all the time from him. He has no reaction time, and no throwing skill. For example, he's played 118 games and has started 3 DP's all year on grounders hit to him.

I saw a Nats-Tigers spring training game this year, and he played a ball off to his side that any halfway decent fielder would have gotten in front of. It ticked off his glove rolled into right field. Again, just ancedotal, but adds to the mountain of evidence.

Not to mention his play in the last World Baseball Classic in the outfield. There were routine balls hit down the right field line Nick would have probably thrown the guy out at second, but with Dunn there I was astonished the batter didn't get an inside-the-parker.