Nick's defense..UZR

[Mackus]

If what Drungo is referring to is correct, then in '08, the Fielding Bible had him at a+10, +4 medium, and -4 deep. With the exception of his first season he has been consistently worse deep than in his other numbers. If the wall is affecting his numbers that would account for it. This season he is improved shallow, the same medium -4, and the same deep -8 as last. His throwing numbers are the best so far of his career.

But how can the shallow wall effect the numbers so much?

As I understand it, the plus / minus system takes every ball hit, and determines how many players would get to that (somehow). If a ball hits off the short wall in Camden Yards, that's gotta be a ball that 0% of RF get to, so Nick not getting to that would be a +/- 0 for that play, since he didn't make an out that 0% of RF get to. Unlike if a routine pop up to RF that 99% of RF catch he happens to lose in the sun and miss, that would be a -.99 since he didn't get an out on a ball 99% of RF do get an out on.

Or is it based on the fact that Nick doesn't record outs on any fly balls to RF that go 330 feet since they hit off the wall, where in lets say 60% of other parks those are outs. So if a ball hits off the wall, Nick gets an (unfair) -.6 for that play because he hasn't made an out on a ball that is caught 60% of the time (without considering the park)?

[Mackus]

Actually, everything is opinion. I would take a scouting report over a defensive metric any day of the week.

A scouting report from a experienced scout who's watched the guy play a lot and watched lots of other guys play. Not the casual or even high level fan's eyes.

Defensive stats aren't randomly made up. There is a significant process to them. They still definitely have their weaknesses, but they paint a large portion of the picture.

Read this article that mweb posted in the MLB forums a little while ago. Its very interesting.

http://joeposnanski.com/JoeBlog/2010/08/11/defensive-numbers/#more-3796

[ledzepp8]

I have made this argument a lot. 'Great plays' on highlights are usually the result of an initial below average read. How many times have we seen Jim Edmonds on highlight reels making 'great plays'. His circus catch is a ball that (vintage) Andruw Jones gets to, settles under and makes it look routine.

Well how do we know that his highlight plays are on plays that he made a bad read on? How do we know that they aren't on plays that most OFers wouldn't get to anyhow?

[El Gordo]

But how can the shallow wall effect the numbers so much?

As I understand it, the plus / minus system takes every ball hit, and determines how many players would get to that (somehow). If a ball hits off the short wall in Camden Yards, that's gotta be a ball that 0% of RF get to, so Nick not getting to that would be a +/- 0 for that play, since he didn't make an out that 0% of RF get to. Unlike if a routine pop up to RF that 99% of RF catch he happens to lose in the sun and miss, that would be a -.99 since he didn't get an out on a ball 99% of RF do get an out on.

Or is it based on the fact that Nick doesn't record outs on any fly balls to RF that go 330 feet since they hit off the wall, where in lets say 60% of other parks those are outs. So if a ball hits off the wall, Nick gets an (unfair) -.6 for that play because he hasn't made an out on a ball that is caught 60% of the time (without considering the park)?

I think it's something like this. From what I understand UZR and the FB use the same data but crunch it differently. Also a ball hit off the wall in OPACY at 360' would be a HR in a park with a normal height fence at 360'.

[AgentOrange]

A scouting report from a experienced scout who's watched the guy play a lot and watched lots of other guys play. Not the casual or even high level fan's eyes.

Defensive stats aren't randomly made up. There is a significant process to them. They still definitely have their weaknesses, but they paint a large portion of the picture.

Read this article that mweb posted in the MLB forums a little while ago. Its very interesting.

http://joeposnanski.com/JoeBlog/2010/08/11/defensive-numbers/#more-3796

I already read the article. I know the stats are not made up, but there are a lot of unanswered questions from the article. They also assign number values according to the play that way made. A guy with good positioning can make a very hard play look easy. On the flip side, bad positioning could make a very easy play look hard.

I will start believing defensive statistics more when they implement the camera system that BTerp showed at the beginning of the year. That will actually measure how much ground was made up by the player, how quickly they responded to the hit, and the route they took. Until then, the scouting reports will trump number valued defensive metrics in my eyes.

[El Gordo]

I already read the article. I know the stats are not made up, but there are a lot of unanswered questions from the article. They also assign number values according to the play that way made. A guy with good positioning can make a very hard play look easy. On the flip side, bad positioning could make a very easy play look hard.

I will start believing defensive statistics more when they implement the camera system that BTerp showed at the beginning of the year. That will actually measure how much ground was made up by the player, how quickly they responded to the hit, and the route they took. Until then, the scouting reports will trump number valued defensive metrics in my eyes.

They do use cameras. The FB people view video of every play made by every player at a given position and compare them.

[AgentOrange]

So what do you do when scout says he's good, another says he's okay and another says he's not good?

You do the same thing you do when you get all the defensive metrics that don't agree with each other.

[AgentOrange]

Being properly positioned is part of what makes a fielder good or bad.

That is my point.

[AgentOrange]

Which is exactly my point.

I feel like scouting reports go into more depth than raw numbers do, which was my point.

[Mackus]

I already read the article. I know the stats are not made up, but there are a lot of unanswered questions from the article. They also assign number values according to the play that way made. A guy with good positioning can make a very hard play look easy. On the flip side, bad positioning could make a very easy play look hard.

Why shouldn't positioning be part of defense?

The only possibility for that to be a negative is if the reason for the poor positioning is because the coaching staff is forcing a guy to play somewhere, like with a weird shift or something.

I mean, everyone always talks about how Cal Ripken was so good defensively mainly because of his positioning. We've had numerous threads on here this year about Jones' positioning. Positioning is definitely a part of it.

It'd be great to have at hand both the overall data taking positioning into account and the camera, Field FX type stuff about how much ground certain guys cover, who's the quickest, and stuff like that. All of that stuff is useful. You've gotta know how to process that information, but it all helps paint a better picture of a guy than simply looking at him and watching him play. They don't replace that by any means, just like offensive stats don't replace scouting, but they paint a large part of the picture.

[AgentOrange]

I'm not sure I agree. If you know what you are looking for (not that you don't) the defensive numbers tell you quite a bit, especially if you don't rely on just one stat.

I suppose I am not comfortable assigning numbers to defense. I feel the margin of error is very high when you do that because in order to assign the number, it takes a huge human element.

[AgentOrange]

Why shouldn't positioning be part of defense?

The only possibility for that to be a negative is if the reason for the poor positioning is because the coaching staff is forcing a guy to play somewhere, like with a weird shift or something.

I mean, everyone always talks about how Cal Ripken was so good defensively mainly because of his positioning. We've had numerous threads on here this year about Jones' positioning. Positioning is definitely a part of it.

It'd be great to have at hand both the overall data taking positioning into account and the camera, Field FX type stuff about how much ground certain guys cover, who's the quickest, and stuff like that. All of that stuff is useful. You've gotta know how to process that information, but it all helps paint a better picture of a guy than simply looking at him and watching him play. They don't replace that by any means, just like offensive stats don't replace scouting, but they paint a large part of the picture.

I hope you understand that I am making a point of positioning being a huge part of being a good defender.

From how I read the article, the guys are assigning "skill levels" to plays that are made. I am saying that if a guy is very well positioned, he might get penalized because he just made a hard play look easy, thus getting a lower value for what should be a high value play. This also works in reverse.

That was my point. If I read it wrong, point it out to me so I can understand better. But I am not comfortable with the fact that actual numbers are applied to each play made.

[baltfan]

They do use cameras. The FB people view video of every play made by every player at a given position and compare them.

It is amazing to me that the defenders of UZR get involved in discussions such as this since it it universally said that year to year UZR is basically useless. Of course, the issue that I have is how is it that we know that UZR over three years is accurate but over 1 year tends to not be? And don't just tell me it is because the sample size gets bigger. If there are errors in the method in which UZR is calculated they could very well be getting multiplied as the number of years increase.

It is very hard to take this stat very seriously given the obvious errors that it makes. It would be one thing if the stat sometimes didn't properly credit a player but last year Franklin Guiterrez was supposedly the greatest defensive outfielder by a long shot. His UZR was 31. This year it is 7.5.

This year Carl Crawford is the OF UZR leader at 22.1. Yet in 2007, when I am sure he was faster, his UZR was -2.5.

Marco Scutaro, who the Red Sox signed because they said he is a great defender (Yes those Red Sox that are universally lauded as the franchise that is among the most stat conscious and best evaluators of talent), has a UZR of -2.4. Oh, that is only one year you say. Well the year before it was -.9. the year before that it was 6.6, but that and 2007 (1.1) were the only years the guy was ever positive. Are the Red Sox just stupid when they said he was a great fielder?

For those of you that still defend this stat, please explain to me how you know that it is accurate? When you look at the list of players when ranked by UZR, how do you know that the stat is properly evaluating defense? Are you just depending on someone telling you that it works? If not, are you depending on your own eyes and looking at a list to see if it seems accurate? And if you are doing the latter, how is that any different than using your own eyes in the first place?

[Mackus]

I hope you understand that I am making a point of positioning being a huge part of being a good defender.

From how I read the article, the guys are assigning "skill levels" to plays that are made. I am saying that if a guy is very well positioned, he might get penalized because he just made a hard play look easy, thus getting a lower value for what should be a high value play. This also works in reverse.

That was my point. If I read it wrong, point it out to me so I can understand better. But I am not comfortable with the fact that actual numbers are applied to each play made.

I think you read it wrong.

The guys mark down where a ball was hit to and how hard it was hit there (the "how hard" part will hopefully be replaced with exact timing as technology allows). They then have, more or less, a database of every single play. They figure out how often a RF gets to a ball that was hit to a certain spot in a certain amount of time. Say its a play that is made exactly 50% of the time. If a guy makes that play, he gets a +.5, if he missed the play, its a -.5. If it is a play that's made only 10% of the time, he would get a +.9 for making it and a -.1 if he misses it. All of a guys plays are summed up over the season, and you figure out the total plus/minus.

There is definitely still subjectivity involved, but the hope is that as more technology is put in place, that can be eliminated.

[makoman]

But how can the shallow wall effect the numbers so much?

As I understand it, the plus / minus system takes every ball hit, and determines how many players would get to that (somehow). If a ball hits off the short wall in Camden Yards, that's gotta be a ball that 0% of RF get to, so Nick not getting to that would be a +/- 0 for that play, since he didn't make an out that 0% of RF get to. Unlike if a routine pop up to RF that 99% of RF catch he happens to lose in the sun and miss, that would be a -.99 since he didn't get an out on a ball 99% of RF do get an out on.

Or is it based on the fact that Nick doesn't record outs on any fly balls to RF that go 330 feet since they hit off the wall, where in lets say 60% of other parks those are outs. So if a ball hits off the wall, Nick gets an (unfair) -.6 for that play because he hasn't made an out on a ball that is caught 60% of the time (without considering the park)?

I think it does consider the park, so he's not getting a -0.6 for that play, the problem is he's not getting the +0.4 that he normally should get in generic big stadium. The person just says "it hit the wall, no one would get it, a 0 play." But somewhere else it is an easy fly ball, so maybe he gets it for +.2 or something. In other words, for half of his games, he is getting far fewer routine chances cause the 335 foot lazy fly ball doesn't turn into an out. Those easy +0.3's and .4's probably add up quite a bit over 81 games...even one 0.1 chance per game could be a +8 that he isn't getting.

Now, do they account for this somehow? Maybe they do, I have no idea how this works. But even if they do, maybe they don't account for it enough. As was pointed out in some other thread, everyone on the O's OF seems to have poor UZR numbers at home, maybe they aren't accounting for the Camden Yards dimensions properly.