PDA

View Full Version : Defense, what's its worth?



AngelAstro
04-08-04, 10:23 AM
Seems that defense is the topic of the past few days.

I've tried to put my mind around UZR, which is supposedly a very good indicator of defensive ability. I've read the articles by MGL at Baseball Primer:

http://www.baseballprimer.com/articles/lichtman_2003-03-14_0.shtml
http://www.baseballprimer.com/articles/lichtman_2003-03-21_0.shtml

This seems like a comprehensive metric. I know there are others as well, and I plan on reading about them when time permits.

http://www.baseballprimer.com/articles/lichtman_2004-02-29_0.shtml
http://www.baseballmusings.com/archives/cat_defense.php
http://www.baseballstuff.com/fraser/new/dips.htm
http://www.futilityinfielder.com/dips03.html
http://www.baseballstuff.com/tangotiger/#Dipsuzr

I opened this thread to start a good discussion on defensive metrics (which metrics are best), how important good defense actually is to a winning ballclub (is pitching dependent on defense, defense on pitching or is the difference between good and bad team defense so slim that it makes no difference) and what skills are most important for a good defender (range, hands, arm).

I would like this thread to stay away from the arguments that Cairo should be playing over Wilson, or other such arguments. I want myself and others to gain a better understanding of defense.

markp
04-08-04, 10:58 AM
baseball is divided into two parts:
1. Scoring runs
2. Stopping the other team from scoring.

Number 2 is further divided into
1. Pitching
2. Defense

Since we see different pitchers with exactly ther same defense behind them have very disparate ERAs and total runs allowed per 9 innings, the effect of defense doesn't seem to be a very big one.

As far as the forumlas used to rate defenders, until they find a way to root out subjectivity, they'll be unreliable. That certainly doesn't mean they're useless, but when ratings are done by people with an obvious axe to grind, I am suspicious of their accuracy.

AngelAstro
04-08-04, 11:04 AM
markp I agree on all points.

With that said, I still think that it is important to attempt comparing different players to determine who is a better defender, why he is a better defender and what that means to a team.

I've read in numerous places that Mike Cameron can save a few wins a year over another CF because of his glove. If defense has little to do with winning ballgames, how are statements like that possible?

Monkeyman
04-08-04, 11:39 AM
Originally posted by AngelAstro
markp I agree on all points.

With that said, I still think that it is important to attempt comparing different players to determine who is a better defender, why he is a better defender and what that means to a team.

I've read in numerous places that Mike Cameron can save a few wins a year over another CF because of his glove. If defense has little to do with winning ballgames, how are statements like that possible?

The fact that someone is making a statement like that is based completely on subjective opinion. Cameron "saving a win" is a matter of him making a good play that someone thinks someone else wouldn't have made and, in addition, they think that without it, the team would have lost. You can't statistically prove something of this nature.

I don't like most defensive stats for this reason. They all involve deciding what is or is not an error, or a ball in "your zone," or they don't take into account pitching quality.

AngelAstro
04-08-04, 11:48 AM
Originally posted by Monkeyman


The fact that someone is making a statement like that is based completely on subjective opinion. Cameron "saving a win" is a matter of him making a good play that someone thinks someone else wouldn't have made and, in addition, they think that without it, the team would have lost. You can't statistically prove something of this nature.

I don't like most defensive stats for this reason. They all involve deciding what is or is not an error, or a ball in "your zone," or they don't take into account pitching quality.

No the statement was backed up by stats such as UZR, the statement I believe was from MGL, not some hack reporter. I'll try to find it.

As far as deciding what is or is not an error, I agree that is too subjective. Since balls in play are trackable, it is easy to determine if a certain fielder should have a chance at making a play on a ball.

As far as pitching quality is concerned, I'm not convinced that it makes a difference. Once a ball is in play it is in play and at that point it is the defense that must field the ball. How does pitching quality affect the balls that are actually in play?

AngelAstro
04-08-04, 12:06 PM
Well, these weren't the exact threads I was looking for, but they may do:

http://www.baseballprimer.com/clutch/archives/00009932.shtml - scroll to post 10 specifically.

Here is another interesting discussion, though not the exact point I was trying to find:

http://www.baseballprimer.com/clutch/archives/00006590.shtml

LoneRedSeat
04-08-04, 12:53 PM
UZR, IMHO, is a fascinating metric, and easily the best defensive metric created to date. Is it subjective? Yes. However, it is still far and away the best method developed in defensive performance analysis thus far.

As an answer in regards to a player saving a win for a team based on UZR:

Monkeyman, you misunderstand. It is not based on a player making a "good play" with a win in that game as a direct result. The processes used to determine a win saved are much more complicated.

One way to do it: After a complete season, you can look at the UZR percentage, and the subsequent UZR runs, and determine how many runs a player would save the team over an average centerfielder over the course of a season.

Regarding Cameron, specifically: Take the difference in UZR runs between Cameron, and the person he was replacing (Shinto?). Plug the difference into Bill James' Pythagorean theorem of baseball (altering the number of runs allowed). Compare the projected wins/losses the Mets would have with Cameron versus the player he was replacing.

AngelAstro
04-08-04, 01:06 PM
Thanks for explaining better than I could, LoneRedSeat.

I do believe that this was done in one of the many threads on Cameron going to the Mets at Primer. If I remember correctly, the Mets actually had very good CF play last year, so comparing to the Mets may not make Cameron look great, but comparing to a league average player would.

markp
04-08-04, 04:09 PM
I think UZR tends to be better when the players don't have a reputation one way or the other, aren't high profile, or aren't playing for a high profile team. When you visit Baseball Primer, Hardball Times, all-baseball, etc. you discover a great deal of bias. While they insist on objectivity with most teams, there are some that they openly root against and/or mock everything about. The Yankees are the obvious target, but some of the bad teams tend to have their players downgraded more than the numbers show.
I'd be a lot more inclined to accept Cameron's UZR than I would Bernie William's or Robert Fick's.

AngelAstro
04-08-04, 04:22 PM
Originally posted by markp
I think UZR tends to be better when the players don't have a reputation one way or the other, aren't high profile, or aren't playing for a high profile team. When you visit Baseball Primer, Hardball Times, all-baseball, etc. you discover a great deal of bias. While they insist on objectivity with most teams, there are some that they openly root against and/or mock everything about. The Yankees are the obvious target, but some of the bad teams tend to have their players downgraded more than the numbers show.
I'd be a lot more inclined to accept Cameron's UZR than I would Bernie William's or Robert Fick's.

But how would one skew data from UZR? Either the fielder fielded a ball in his zone or he didn't or he made an error. I don't see how even if someone was rooting against a player how that would show in UZR.

I think the problems come down to errors because an error is a subjective stat. However, if anything, error rates are probably low for everyone. I can't think of how many times I watch a game see what I believe is a blatant error and then see the official scorer call it a hit.

markp
04-08-04, 10:48 PM
Originally posted by AngelAstro


But how would one skew data from UZR? Either the fielder fielded a ball in his zone or he didn't or he made an error. I don't see how even if someone was rooting against a player how that would show in UZR.

I think the problems come down to errors because an error is a subjective stat. However, if anything, error rates are probably low for everyone. I can't think of how many times I watch a game see what I believe is a blatant error and then see the official scorer call it a hit.


The subjective part is determining which balls are in a players zone and which aren't. Since in every case the decision is dependant on the observers judgement, and since the evidence that few (if any) observers are completely unbiased, their judgements are consciously or unconsciously skewed to fit their preconception and bias.

Monkeyman
04-09-04, 12:08 AM
Originally posted by AngelAstro

As far as pitching quality is concerned, I'm not convinced that it makes a difference. Once a ball is in play it is in play and at that point it is the defense that must field the ball. How does pitching quality affect the balls that are actually in play?

A worse pitcher would be more likely to, say, give up a scorcher up the 3rd base line than someone like Mariano who induces pop-ups and weak flares a lot. If you're on a team with a worse pitching staff, you'll have more balls to field, but they'll probably be harder to field.

I flatly don't agree with this idea that the only thing that matters are walks, ks, and homers. If you're giving up hard hit balls in play, you're giving up hits. If you're a pitcher getting GBs to SS and 2B, you're not. Some pitchers aren't trying to keep the ball out of play (i.e. Derek Lowe, Paul Quantrill) but still are effective pitchers.

While technically bad enough fielding could turn any ball into a hit/error, and conversely exceptional enough fielding could turn any ball into an out, its reasonable to expect that certain balls are more likely to be turned into outs, and the pitcher is probably doing that on purpose. If a guy is getting good movement on his pitches, hitters will have a harder time making good contact, but the ball can still go in play. It is more likely to be a weaker hit, and thus much easier to field.

AngelAstro
04-27-04, 12:37 PM
Originally posted by markp



The subjective part is determining which balls are in a players zone and which aren't. Since in every case the decision is dependant on the observers judgement, and since the evidence that few (if any) observers are completely unbiased, their judgements are consciously or unconsciously skewed to fit their preconception and bias.

So, I contacted STATS, Inc., the company which records the play-by-play data, about how they determine which zone a ball is hit. Their answer:

"Each reporter working a game will note where a ball is hit. We have two reporters at every game, and if there is a discrepancy we also have it on tape, and can review it if neccessary."

So, while it is possible that the opinions are biased, I find it very hard to fathom that two people would make a mistake on where a ball is hit. I tried to do this myself while I was at the game on Saturday, I did not find it all that difficult (some plays were more difficult than others), but overall I think it would be very hard to make a mistake. I wish I could compare my scorecard to that of STATS.

Vile Tom
04-27-04, 10:18 PM
Originally posted by markp



The subjective part is determining which balls are in a players zone and which aren't. Since in every case the decision is dependant on the observers judgement, and since the evidence that few (if any) observers are completely unbiased, their judgements are consciously or unconsciously skewed to fit their preconception and bias.

They're developing a video system to help evaluate defense. They'll be able to judge how hard the ball was hit, how quickly the player reacts, how close to the "ideal" route to the ball the fielder took etc.
This really isn't that difficult a project in terms of writing the software to evaluate these factors. It's just very expensive to put the cameras in.
I believe the results will look a lot like UZR.

AngelAstro
04-28-04, 07:08 AM
Originally posted by Vile Tom


They're developing a video system to help evaluate defense. They'll be able to judge how hard the ball was hit, how quickly the player reacts, how close to the "ideal" route to the ball the fielder took etc.
This really isn't that difficult a project in terms of writing the software to evaluate these factors. It's just very expensive to put the cameras in.
I believe the results will look a lot like UZR.

Is there some link to an article about this? This sounds very interesting.

WiffleWOOD
04-28-04, 08:33 AM
Originally posted by AngelAstro


Is there some link to an article about this? This sounds very interesting.

i'll try to find it. i read it months ago, too. they'll be installing a few cameras a year until 2006 or so, when they will have a bunch in each park. This could be a major step forward for defensive metrics.

justin32099
04-28-04, 02:22 PM
The effect of defense is much, much greater than people understand. Take a good defensive centerfielder, for instance, Mike Cameron. I'd say that it's totally fair to say that Cameron might catch 30-40 balls that....oh, say....Bernie Williams wouldn't be able to get to. They aren't errors on Bernie, but it's factored into UZR pretty well.

I'd say I believe in defensive value much more than most sabermetcially-minded people. If you ever have some time and you like statistics, take a few of your favorite pitchers. Graph their K/9, BB/9, and HR/9 ratios over their career. They will usually be pretty steady, or move uniformly. Then divide the number of non-home run hits (H - HR) allowed into the number of balls batted into play (AB - K - HR) and graph that. For most pitchers, this will fluctuate far more. Often what is perceived as an "off year" can be explained simply by poorer defense, or sometimes just bad luck. That's why power (K) pitchers, in general, are more steady than pitchers (see Derek Lowe, Greg Maddux) that rely on their defense to do a lot of the work.

That's why, for example, Randy Johnson's slow start (well, before his last start) means little, since K and BB ratios are still about where they have been before. But Mussina's slow start, on the other hand, really worries me, as his strikeouts are way down and his walks are way up. Something is really up with him--it isn't defense or bad luck.

BomberBrian
08-01-04, 10:39 AM
Can someone tell me where i can get UZR stats for this year and past years?

AngelAstro
08-01-04, 12:41 PM
Originally posted by BomberBrian
Can someone tell me where i can get UZR stats for this year and past years? http://www.tangotiger.net/#Dipsuzr

JeffWeaverFan
08-03-04, 10:32 AM
On ESPN.com when you look up the players profile and go to the defensive stats they have two that I have noticed: Range Factor and Zone Rating. The Zone Rating is a percentage out of 1.000 and Range Factor is a number usually around 2-3. But, I'm not really sure how they calculate those number and what exactly Zone Rating is. I was also wondering how good these are to measure a players defense. So, if you have any more info on them, please post. Thanks.

Irony Of It All
08-03-04, 10:42 AM
Originally posted by JeffWeaverFan
On ESPN.com when you look up the players profile and go to the defensive stats they have two that I have noticed: Range Factor and Zone Rating. The Zone Rating is a percentage out of 1.000 and Range Factor is a number usually around 2-3. But, I'm not really sure how they calculate those number and what exactly Zone Rating is. I was also wondering how good these are to measure a players defense. So, if you have any more info on them, please post. Thanks.

RFg = (A+PO) / G
RF9 = 9 * (A+PO) / Inn

Range factor is not a very good stat because it's largely dependent on the pitching staff. So for example, a SS on a team with a lot of groundball pitchers will appear to be better than someone on a big strikeout team since they'll have more opportunities to make plays.

Zone rating - if I recall correctly - is basically the percentage of balls that player gets to that's in their "zone."

UZR is much better than both of these metrics, but during the season I'd say ZR may be the best - at least that I know of.

JeffWeaverFan
08-03-04, 11:00 AM
Originally posted by Irony Of It All


RFg = (A+PO) / G
RF9 = 9 * (A+PO) / Inn

Range factor is not a very good stat because it's largely dependent on the pitching staff. So for example, a SS on a team with a lot of groundball pitchers will appear to be better than someone on a big strikeout team since they'll have more opportunities to make plays.

Zone rating - if I recall correctly - is basically the percentage of balls that player gets to that's in their "zone."

UZR is much better than both of these metrics, but during the season I'd say ZR may be the best - at least that I know of.
Allright, Now I get it a little better. So if you are on a team that has mostly fly ball pitchers and you are an outfielder, your range factor should be better.

I also agree about UZR being the best, but I can't find a site that has UZR for just last season and of course, they can't do it for this season until after the season is finished. Thanks for the info.

EDIT: One more question: Since players play in diferent parks would the "zone" that they have to cover (I'm talking about outfielders here) be different from other players "zone"?

AngelAstro
08-03-04, 11:02 AM
Originally posted by JeffWeaverFan
On ESPN.com when you look up the players profile and go to the defensive stats they have two that I have noticed: Range Factor and Zone Rating. The Zone Rating is a percentage out of 1.000 and Range Factor is a number usually around 2-3. But, I'm not really sure how they calculate those number and what exactly Zone Rating is. I was also wondering how good these are to measure a players defense. So, if you have any more info on them, please post. Thanks.

Zone Rating is basically a less refined version of UZR. Since UZR is only calculated after the season ends, ZR is probably the best fielding metric to look at during the season.

Irony Of It All
08-03-04, 11:05 AM
Originally posted by JeffWeaverFan

EDIT: One more question: Since players play in diferent parks would the "zone" that they have to cover (I'm talking about outfielders here) be different from other players "zone"?

I haven't read this yet but I believe this should answer your question.

http://www.baseballthinkfactory.org/files/main/article/lichtman_2003-03-21_0/