Critique of WAR

[CubinNY]

This article better summerizes my problems with WAR than I have time to articulate here.

 

http://www.realclearsports.com/articles/2013/09/30/going_to_war_with_baseball_statistics_97870.html

 

What is not mentioned is that the model is able to account for only .85 of the variance in data, meaning it has a 15% margin of error. So, although I think WAR is useful in the aggregate, the way it's used most of the time goes well beyond what the statistic can account for. For example, declaring Mike Trout was a more valuable player the Miguel Cabrerra in 2012.

[OleMissCub]

For example, declaring Mike Trout was a more valuable player the Miguel Cabrerra in 2012.

 

That's not all that hard to believe, is it?

[Guest]

For example, declaring Mike Trout was a more valuable player the Miguel Cabrerra in 2012.

 

That's not all that hard to believe, is it?

It's not hard to believe at all. The issue is using WAR to support the claim.

[Hairyducked Idiot]

The problem is that those supporting the reverse claim are using even worse stats.

[Guest]

People that criticize WAR in this way are essentially applying an idea that everyone agrees with, but only doing so for WAR, which is why there's conflict.

 

WAR isn't infallible, especially when you start dealing with decimal places. It's an approximation of player value, and a good one. The dirty little secret here is that applies for pretty much every statistic. For example, OPS is a good offensive measure, but it has blind spots. That doesn't mean it's trash or that we shouldn't use it for anything, but there's times where it's the best representation of what we want to measure, so we accept it. Not as gospel, but as the best we can tell.

 

The same applies to WAR, even more so than most metrics actually. WAR is our best representation of total player value, especially since most of the time when people criticize the outcome(i.e. Trout v. Cabrera) with WAR, the criticisms are more "that can't be right" than "here's why it should be this instead". The correlation data should be enough for anyone to not reject it out of hand. But for whatever reason, people ascribe the support of WAR as a metric as support of WAR as canon, immune to any further improvement or insight. People use WAR because it's often the best we can come up with, but even so there's still compelling arguments that it's not accurate in specific cases. "That can't be right" isn't compelling.

[Derwood]

Most criticisms I hear for WAR are from sabermetric skeptics who dismiss it because "no one can tell me how it's calculated"

[Sammy Sofa]

People that criticize WAR in this way are essentially applying an idea that everyone agrees with, but only doing so for WAR, which is why there's conflict.

 

WAR isn't infallible, especially when you start dealing with decimal places. It's an approximation of player value, and a good one. The dirty little secret here is that applies for pretty much every statistic. For example, OPS is a good offensive measure, but it has blind spots. That doesn't mean it's trash or that we shouldn't use it for anything, but there's times where it's the best representation of what we want to measure, so we accept it. Not as gospel, but as the best we can tell.

 

The same applies to WAR, even more so than most metrics actually. WAR is our best representation of total player value, especially since most of the time when people criticize the outcome(i.e. Trout v. Cabrera) with WAR, the criticisms are more "that can't be right" than "here's why it should be this instead". The correlation data should be enough for anyone to not reject it out of hand. But for whatever reason, people ascribe the support of WAR as a metric as support of WAR as canon, immune to any further improvement or insight. People use WAR because it's often the best we can come up with, but even so there's still compelling arguments that it's not accurate in specific cases. "That can't be right" isn't compelling.

 

I love you.

[Sammy Sofa]

Most criticisms I hear for WAR are from sabermetric skeptics who dismiss it because "no one can tell me how it's calculated"

 

Or it feels like a play on the "well, it's called the 'THEORY' of evolution"-argument. Just because some people act like it's set in stone doesn't negate it's value or mean that the people that use it correctly don't understand that it's not perfect.

[Guest]

Lol. Carry on.

[Sammy Sofa]

Lol. Carry on.

 

So do you disagree with the idea that this type of criticism is applicable to plenty of other stats besides WAR?

[SABR Gamer]

Lol. Carry on.

 

Lol. Okay, Mr. Average.

[Geech]

Lol. Carry on.

 

I'm genuinely curious: what are you laughing out loud at?

[jersey cubs fan]

Most criticisms I hear for WAR are from sabermetric skeptics who dismiss it because "no one can tell me how it's calculated"

 

Or it feels like a play on the "well, it's called the 'THEORY' of evolution"-argument. Just because some people act like it's set in stone doesn't negate it's value or mean that the people that use it correctly don't understand that it's not perfect.

Not at all like that. The problem with war is it's attempt to quantify defense and baserunning. Hitting and pitching stats work because of what they measure, one on one matchups. Football stats don't work because of what it attempts to measure, and the same is true for defense in baseball.

[Guest]

Lol. Carry on.

 

I'm genuinely curious: what are you laughing out loud at?

I don't think the article was read by most of the people posting.

 

The point wasn't WAR SUCK! Far from it.

[Guest]

Most criticisms I hear for WAR are from sabermetric skeptics who dismiss it because "no one can tell me how it's calculated"

 

Or it feels like a play on the "well, it's called the 'THEORY' of evolution"-argument. Just because some people act like it's set in stone doesn't negate it's value or mean that the people that use it correctly don't understand that it's not perfect.

Not at all like that. The problem with war is it's attempt to quantify defense and baserunning. Hitting and pitching stats work because of what they measure, one on one matchups. Football stats don't work because of what it attempts to measure, and the same is true for defense in baseball.

 

This is just a variation on "that doesn't seem right". If the baserunning and defensive components were as shoddy as you're suggesting, then WAR wouldn't add up as well as it does.

 

I don't think the article was read by most of the people posting.

 

The point wasn't WAR SUCK! Far from it.

 

The article was equal parts "this is what WAR is" and "here's some random FUD about defensive metrics". Brady Anderson hit 50 home runs one year, never cracking 24 the rest of his career. By the article's logic, we should question whether or not home runs are a good measure of counting home runs.

[Sammy Sofa]

Most criticisms I hear for WAR are from sabermetric skeptics who dismiss it because "no one can tell me how it's calculated"

 

Or it feels like a play on the "well, it's called the 'THEORY' of evolution"-argument. Just because some people act like it's set in stone doesn't negate it's value or mean that the people that use it correctly don't understand that it's not perfect.

Not at all like that. The problem with war is it's attempt to quantify defense and baserunning. Hitting and pitching stats work because of what they measure, one on one matchups. Football stats don't work because of what it attempts to measure, and the same is true for defense in baseball.

 

But everyone who should know that already knows that; WAR is used as the best metric of overall player value because that's what it is right now. It's not perfect and it's not set in stone, but it's the best available, and the caveats (especially in regards to defense) have been obvious from the get-go; it's very much an evolving thing. The article is just incredibly redundant.

[Derwood]

Most criticisms I hear for WAR are from sabermetric skeptics who dismiss it because "no one can tell me how it's calculated"

 

Or it feels like a play on the "well, it's called the 'THEORY' of evolution"-argument. Just because some people act like it's set in stone doesn't negate it's value or mean that the people that use it correctly don't understand that it's not perfect.

Not at all like that. The problem with war is it's attempt to quantify defense and baserunning. Hitting and pitching stats work because of what they measure, one on one matchups. Football stats don't work because of what it attempts to measure, and the same is true for defense in baseball.

 

But everyone who should know that already knows that; WAR is used as the best metric of overall player value because that's what it is right now. It's not perfect and it's not set in stone, but it's the best available, and the caveats (especially in regards to defense) have been obvious from the get-go; it's very much an evolving thing. The article is just incredibly redundant.

 

to me, the biggest thing hurting WAR's credibility among the common fans and media is that there are several versions of it that are sometimes at odds with each other. If there was just WAR, fine, but with bWAR and fWAR, etc., it muddles what should be a commonly used evaluator

[Sammy Sofa]

to me, the biggest thing hurting WAR's credibility among the common fans and media is that there are several versions of it that are sometimes at odds with each other. If there was just WAR, fine, but with bWAR and fWAR, etc., it muddles what should be a commonly used evaluator

 

I don't think it "muddles" so much as expands on what is very much a work in progress. I mean, do we really care about its credibility with meatballs? All of those variations are just efforts to better refine what WAR can tell us, which is infinitely more valuable/important than whether or not the common whatever embraces it.

 

I mean, wanting it to be "just WAR" seems completely ass-backwards.

[Derwood]

to me, the biggest thing hurting WAR's credibility among the common fans and media is that there are several versions of it that are sometimes at odds with each other. If there was just WAR, fine, but with bWAR and fWAR, etc., it muddles what should be a commonly used evaluator

 

I don't think it "muddles" so much as expands on what is very much a work in progress. I mean, do we really care about its credibility with meatballs? All of those variations are just efforts to better refine what WAR can tell us, which is infinitely more valuable/important than whether or not the common whatever embraces it.

 

I mean, wanting it to be "just WAR" seems completely ass-backwards.

 

It muddles it when neither site will reveal their calculations. Yeah, we know one weighs defense or baserunning or whatever more than the other, but when certain players vary wildly in their fWAR/bWAR, it makes it hard for some people to trust. Add in the somewhat nebulous definition of "replacement player", and it's just not something that will ever catch on with the masses.

 

Not that any of that makes in invaluable, of course. People who know how it works will continue to get use out of it

[Guest]

Simple. Just look at fWAR.

[MWV]

The article was equal parts "this is what WAR is" and "here's some random FUD about defensive metrics". Brady Anderson hit 50 home runs one year, never cracking 24 the rest of his career. By the article's logic, we should question whether or not home runs are a good measure of counting home runs.

If no one else is going to acknowledge how hilarious/spot-on this comment is, I guess it falls to me.

 

And much like many other critiques of WAR, this is (as TT observes) largely a critique of defensive metrics which--like WAR--are admittedly imperfect. That being said, the degree of correlation is pretty outstanding considering the the breadth of performance that WAR encapsulates.

[sneakypower]

It muddles it when neither site will reveal their calculations. Yeah, we know one weighs defense or baserunning or whatever more than the other, but when certain players vary wildly in their fWAR/bWAR, it makes it hard for some people to trust. Add in the somewhat nebulous definition of "replacement player", and it's just not something that will ever catch on with the masses.

 

Not that any of that makes in invaluable, of course. People who know how it works will continue to get use out of it

the calculations are publicly known

http://www.fangraphs.com/library/misc/war/

[Guest]

The article was equal parts "this is what WAR is" and "here's some random FUD about defensive metrics". Brady Anderson hit 50 home runs one year, never cracking 24 the rest of his career. By the article's logic, we should question whether or not home runs are a good measure of counting home runs.

If no one else is going to acknowledge how hilarious/spot-on this comment is, I guess it falls to me.

 

And much like many other critiques of WAR, this is (as TT observes) largely a critique of defensive metrics which--like WAR--are admittedly imperfect. That being said, the degree of correlation is pretty outstanding considering the the breadth of performance that WAR encapsulates.

That's the point. Defense and base running suffer from inherent problems of reliability of measurement. Since they go into the calculation, WAR suffers and it's construct validity is questionable. The extent to which a player's WAR is derived from defense and base running the more suspicious the value. In other words, the fact that WAR correlates in the aggregate has little bearing when one is talking about an individual performance. That's an axiom of group data.

 

But that's not really the main issue I have with it. It is obstenibly used to take context out performance and try as one might that cannot be accomplished, it is an impossible task. Now does that make it useless? No. It's just one of many measures of performance, not THE measure of performance.

[Sammy Sofa]

But who is saying that it is? That whole idea is that it's working off of quantifiable data, and the obvious reality is that not everything about the game falls under that umbrella. WAR is just the best effort.

 

I guess I really just don't understand what exactly you're taking issue with; people who use it like it's perfect? Do you think it's been presented that way all along?

[KingCubsFan]

 

I guess I really just don't understand what exactly you're taking issue with; people who use it like it's perfect?

That's my issue with it. It's an interesting statistic, and certainly the best at attempting to quantify all the ways a player may impact the game. But I feel like people have come to lazily rely on it as the only measure of a player's worth. Discussion of players has essentially been boiled down to "[X Player] had a higher WAR than [Y Player]! Of course he's better!"