The Simple Questions Thread

[WavelandAve]

Sorry for asking so many, maybe obvious, questions, but I am just trying to get a good understanding of some of this stuff. For something like WAR, I understand its concept and meaning but I don't know how to determine what a good WAR is. Is there a way to see/determine what point is above average, below average, etc...

[C1Bman88]

Sorry for asking so many, maybe obvious, questions, but I am just trying to get a good understanding of some of this stuff. For something like WAR, I understand its concept and meaning but I don't know how to determine what a good WAR is. Is there a way to see/determine what point is above average, below average, etc...

The average WAR is about +2 WAR. This is working under the assumption that a replacement level player costs your team 20 runs (2 wins) over the course of a full season. +3 WAR is solidly above average, +4 is very good, +5 is excellent, and anything beyond that is simply outstanding.

To be honest (and this is something of a digression- so you don't need to read this unless you absolutely want to), most people might be better suited to simply use a baseline of "average" rather than "replacement level," because 1) it's easier to comprehend, and 2) because the definition of "replacement" is not a static one.

There are various forms of replacement level. Some people use a win percentage of .350 (so they'd be expected to win 57 games), others use .290 (47 wins). Some believe the replacement level should be set at the most recent season's lowest winning percentage, and some people will tell you that a replacement level team is a .450 (73 win) team. There's a number of definitions for replacement level. But if we use "average," the baseline is exactly .500- no ifs, ands, or buts.

[Hustla23]

C1Bman88, I hope you're free sometime during the day to take a look at my question.

I was debating a poster and I used WAR to show that David Wright and Albert Pujols possessed the same value the last two years when they posted similar WAR's.

The other poster stated that he does not credit WAR as a useful statistic because it utilizes UZR and UZR in itself is flawed.

He stated that the UZR for first baseman is useless because many balls are not put into play for them so range isn't a huge factor.

So I have two questions. Is UZR, in fact, something not worth mentioning when comparing the defensive abilities of players because they might be accurate for one position but not for another ? If this is true, what is the best defensive statistic out there ?

And : How is a first baseman's defensive ability measured best in ? I guess UZR doesn't take the cake because it doesn't measure scoops or stretching out to catch a throw or something like that.

Thanks.

[C1Bman88]

C1Bman88, I hope you're free sometime during the day to take a look at my question.

I was debating a poster and I used WAR to show that David Wright and Albert Pujols possessed the same value the last two years when they posted similar WAR's.

The other poster stated that he does not credit WAR as a useful statistic because it utilizes UZR and UZR in itself is flawed.

UZR is absolutely flawed, and this poster is completely and totally correct in his assertion. It's even more flawed on a single-season basis, because the amount of chances a player receives is extremely small (especially at the corner positions) and because there is variation in the baseline.

I would be interested to see what this poster DOES use in terms of player evaluation. We can talk all we want about the flaws of UZR, but we have to realize that there are a number of flaws in run estimators as well, and that includes linear weights as well.

He stated that the UZR for first baseman is useless because many balls are not put into play for them so range isn't a huge factor.

This isn't true at all. A first baseman generally receives 250 BIP at his position, which is nearly half the amount a shortstop receives. So while it is true that there are less BIP hit to the position, it doesn't mean that it's not important to take their range into consideration- it just means that we have to look at multi-year data instead of a single season to get an accurate view of their range. I can't imagine a single person that would want David Ortiz playing defense at first base over, say, Casey Kotchman.

So I have two questions. Is UZR, in fact, something not worth mentioning when comparing the defensive abilities of players because they might be accurate for one position but not for another ? If this is true, what is the best defensive statistic out there ?

There is a much higher degree of certainty when measuring the defensive value of a shortstop, second baseman or center fielder than with the corner positions. It has absolutely nothing to do with the formation of the metric- rather, it has to do with the simple fact that corner positions receive less balls in play than up the middle positions do.

UZR is the gold standard for publicly available defensive metrics because of all the adjustments it makes. It doesn't mean that it's not flawed, but what this poster (and a lot of other people) don't realize is that UZR (or ANY defensive statistic, for that matter) isn't meant to measure a player's ability; it's meant to measure their value. Having a good or bad UZR doesn't mean you're a good or bad defender (unless we're using regressed, multi-year data), it just means you provided more or less value to your position than the average fielder based on how well you do in particular sub-zones.

We can't say that Albert Pujols is a +1.3 defensive player because that was the number he posted in 2009, or that Mark Teixeira was a -3.7 or whatever he was this year either. You have to look at multiple years to get their actual talent level. For first basemen, that's going to be about four years worth of data.

And : How is a first baseman's defensive ability measured best in ? I guess UZR doesn't take the cake because it doesn't measure scoops or stretching out to catch a throw or something like that.

UZR is still the best. In terms of measuring "scooping" ability, you're best suited to use a With Or Without You (WOWY) approach to see how many throwing errors your infield commits with and without the first baseman at the position. You're looking at a maximum of around +4 runs saved.

Since most people don't have the database available to do it, you can try something like:

(3B.a + 2B.a + SS.a)/(3B.a+3B.e+2B.a+2B.e+SS.a+SS.e)

where the denoted letter "a" stands for "assists" and "e" is for "error." That'll give you a very rough estimate. Compare this percentage to the league and you'll find +/- throws saved.

I hope I've answered your questions, and if I haven't, let me know and I'll do what I can to help out.

[jetsfan89]

what kind of defensive metric can we use for catchers?

[C1Bman88]

what kind of defensive metric can we use for catchers?

This is JinAZ's:

http://www.basement-dwellers.com/200...-catchers.html

And this is mine- it's an aggregate system. I use different methodologies (including J's) and data to come up with an estimate:

http://triplesalley.wordpress.com/20...ll-underrated/

[Hustla23]

UZR is absolutely flawed, and this poster is completely and totally correct in his assertion. It's even more flawed on a single-season basis, because the amount of chances a player receives is extremely small (especially at the corner positions) and because there is variation in the baseline.

I would be interested to see what this poster DOES use in terms of player evaluation. We can talk all we want about the flaws of UZR, but we have to realize that there are a number of flaws in run estimators as well, and that includes linear weights as well.



This isn't true at all. A first baseman generally receives 250 BIP at his position, which is nearly half the amount a shortstop receives. So while it is true that there are less BIP hit to the position, it doesn't mean that it's not important to take their range into consideration- it just means that we have to look at multi-year data instead of a single season to get an accurate view of their range. I can't imagine a single person that would want David Ortiz playing defense at first base over, say, Casey Kotchman.



There is a much higher degree of certainty when measuring the defensive value of a shortstop, second baseman or center fielder than with the corner positions. It has absolutely nothing to do with the formation of the metric- rather, it has to do with the simple fact that corner positions receive less balls in play than up the middle positions do.

UZR is the gold standard for publicly available defensive metrics because of all the adjustments it makes. It doesn't mean that it's not flawed, but what this poster (and a lot of other people) don't realize is that UZR (or ANY defensive statistic, for that matter) isn't meant to measure a player's ability; it's meant to measure their value. Having a good or bad UZR doesn't mean you're a good or bad defender (unless we're using regressed, multi-year data), it just means you provided more or less value to your position than the average fielder based on how well you do in particular sub-zones.

We can't say that Albert Pujols is a +1.3 defensive player because that was the number he posted in 2009, or that Mark Teixeira was a -3.7 or whatever he was this year either. You have to look at multiple years to get their actual talent level. For first basemen, that's going to be about four years worth of data.



UZR is still the best. In terms of measuring "scooping" ability, you're best suited to use a With Or Without You (WOWY) approach to see how many throwing errors your infield commits with and without the first baseman at the position. You're looking at a maximum of around +4 runs saved.

Since most people don't have the database available to do it, you can try something like:

(3B.a + 2B.a + SS.a)/(3B.a+3B.e+2B.a+2B.e+SS.a+SS.e)

where the denoted letter "a" stands for "assists" and "e" is for "error." That'll give you a very rough estimate. Compare this percentage to the league and you'll find +/- throws saved.

I hope I've answered your questions, and if I haven't, let me know and I'll do what I can to help out.

Thanks so much, once again

So UZR is still the best metric to use, even though you mentioned it is very flawed ?

So, in order to calculate a first baseman's total defensive value, would you add the WOWY and UZR to find how many total run's they saved ? And then figure that into WAR ?

Also, the poster mentioned that a better metric to use would be Bill James's + / - rating which apparently also gives you the number of runs saved but takes into account a first baseman's ability to catch scoops, an infielder's handling of bunts and their abilities to turn the double play, as well as other aspects.

Are you familiar with this ? I am not so I am not in a position to make a judgment as to whether this would be more insightful or more worthy of being integrated into WAR than UZR.

[C1Bman88]

Thanks so much, once again

So UZR is still the best metric to use, even though you mentioned it is very flawed ?

Exactly. It's the most accurate defensive measurement we have available to the public. It's possible that Major League organizations have more refined data, but it's hard to believe that it could be that much better than UZR.

So, in order to calculate a first baseman's total defensive value, would you add the WOWY and UZR to find how many total run's they saved ? And then figure that into WAR ?

Pretty much, yeah. But you're not going to get a good WOWY rating unless you've got the retrosheet data all set up.

Also, the poster mentioned that a better metric to use would be Bill James's + / - rating which apparently also gives you the number of runs saved but takes into account a first baseman's ability to catch scoops, an infielder's handling of bunts and their abilities to turn the double play, as well as other aspects.

Are you familiar with this ? I am not so I am not in a position to make a judgment as to whether this would be more insightful or more worthy of being integrated into WAR than UZR.

This poster is wrong. Plus/Minus does NOT rate a first baseman's ability to handle scoops. At all. BIS tracks the data, yes, but neither James nor Dewan have implemented it into the Plus/Minus system. They will most likely try and use it in the future (and if they don't, they should). As for bunt plays and such...UZR handles all of these things. UZR rates a player's ability to turn the double play, handle bunts, and so on and so forth. UZR also makes adjustments for the handedness of the pitcher and the tendency of the pitching staff in terms of inducing ground balls or fly balls. Plus/Minus does not.

Here are some of the fundamental differences between the two systems, as described by Mitchel Lichtman (the creator of UZR).

The Plus/Minus system gets a lot of positive press because it has James' seal of approval. While that might be all nice and dandy, James is not infallible and the system he tends to get a lot of credit for (it's essentially John Dewan's) is a step behind UZR right now.

Something tells me this poster has no actual knowledge of UZR and the way in which it works, and how it compares to Plus/Minus. Now, this isn't to say that Plus/Minus isn't bad- it's not. In fact, it's the second best system behind UZR. I like to blend the two- 75% UZR, 25% Plus/Minus- to get a full idea of a player's defensive value in a given year.

[Hustla23]

Exactly. It's the most accurate defensive measurement we have available to the public. It's possible that Major League organizations have more refined data, but it's hard to believe that it could be that much better than UZR.



Pretty much, yeah. But you're not going to get a good WOWY rating unless you've got the retrosheet data all set up.



This poster is wrong. Plus/Minus does NOT rate a first baseman's ability to handle scoops. At all. BIS tracks the data, yes, but neither James nor Dewan have implemented it into the Plus/Minus system. They will most likely try and use it in the future (and if they don't, they should). As for bunt plays and such...UZR handles all of these things. UZR rates a player's ability to turn the double play, handle bunts, and so on and so forth. UZR also makes adjustments for the handedness of the pitcher and the tendency of the pitching staff in terms of inducing ground balls or fly balls. Plus/Minus does not.

Here are some of the fundamental differences between the two systems, as described by Mitchel Lichtman (the creator of UZR).

The Plus/Minus system gets a lot of positive press because it has James' seal of approval. While that might be all nice and dandy, James is not infallible and the system he tends to get a lot of credit for (it's essentially John Dewan's) is a step behind UZR right now.

Something tells me this poster has no actual knowledge of UZR and the way in which it works, and how it compares to Plus/Minus. Now, this isn't to say that Plus/Minus isn't bad- it's not. In fact, it's the second best system behind UZR. I like to blend the two- 75% UZR, 25% Plus/Minus- to get a full idea of a player's defensive value in a given year.

Once again, wonderfully stated and extremely helpful.



Just curious, are you a professional statistician, work in baseball, ... ?

[jetsfan89]

Once again, wonderfully stated and extremely helpful.



Just curious, are you a professional statistician, work in baseball, ... ?

nah he's just really educated on sabermetrics

[C1Bman88]

Once again, wonderfully stated and extremely helpful.



Just curious, are you a professional statistician, work in baseball, ... ?

Nope, just an avid fan.

[Hustla23]

Nope, just an avid fan.

Sure haha

Sorry to keep pestering, but WAR for catchers cannot be determined because it doesn't integrate a catcher's fielding.

EDIT:

I just saw the link you posted for Jetsfan. I'll read it and ask a related question. Thank you.

[jetsfan89]

sorry if this has been asked before but what does the WPA stat illustrate?

also i have another question. UZR is obviously the most commonly used defensive metric. but i find it fluctuates too much. is there any other defensive metric that could be used other than UZR?

[C1Bman88]

sorry if this has been asked before but what does the WPA stat illustrate?

also i have another question. UZR is obviously the most commonly used defensive metric. but i find it fluctuates too much. is there any other defensive metric that could be used other than UZR?

WPA is the "story telling stat." A full explanation can be found here.

As for your second question- all defensive metrics fluctuate from year to year. If you're looking for "consistency," then that implies you're looking to measure a player's ability; not his value. If that's the case, all you need is multi-year UZR/150.

If you're looking for another defensive metric just to use comparatively, you could mess around with THT's Revised Zone Ratings, Dave Pinto's Probabilistic Model of Range, Total Zone, or (if you have a subscription to BJOL) John Dewan's Plus/Minus. But all of those metrics are inferior to UZR, and none of them really offer anything that UZR doesn't.

There is one system out there that has the potential to surpass UZR, and that's Shane Jensen's Spatial Aggregate Fielding Evaluations (SAFE), which uses a smoothing function instead of arbitrary zones. SAFE is still lacking in some aspects though, so I'd still recommend UZR for any and all purposes of defensive evaluation. You can regress it to the Fans' Scouting Report, if you'd like. Some people like to do that to get a better grasp on "ability."

[CUBDOM4life]

So I'm in the middle of an intense battle with a couple buddies on why stolen bases are extremely overrated. They have no idea what WAR is unfortunately. I explained how a sb is worth approximately .22 runs and a caught steal is -.36 runs. They don't give a flying ****. I made it clear that a caught steal hurts you more than a succsessful steal helps you.

I don't know how to make them understand why I'm right because they don't know about WAR.

What are some other points I can make so that they can understand?

Oh and sorry if this has been previously discussed.