At what point do we take a good long look at WAR and realize it needs fixin'

[Mike Sixel]

I agree with Jay, read that article. It is very good.

[Oxtung]

Fangraphs also happened to take a look at the exact impetus of this thread -- WAR for Marte vs Harper season-to-date.

 

I agree in full that the defensive metrics need to continue improving, but I think there's also some validity to the point that it's much easier for us to quantify offensive contributions in our minds than it is to assess defense or baserunning (which is how Marte is so close to Harper in WAR).

 

Here's what Dave Cameron published yesterday looking at removing various defensive components to see if it made WAR more accurate (it didn't): WAR: Imperfect but Useful Even in Small Samples | FanGraphs Baseball

 

That is an interesting article and helps add to the overall WAR dialogue however, it has no bearing on the original question of Marte vs. Harper or more specifically how defensive metrics should be considered at the player WAR level. To put this in simpler terms think about this example:

 

I want some scrubway but realize I'm short a dollar so I look around my house. I find 27 cents on my night stand, 72 cents on under my couch cushions and 13 cents in my car. By adding all these up I know I have $1.12. Or in other words we can use the individual stacks of money to determine a total. However, when I get to scrubway to meet my buddy and I showed him that $1.12 he would have no idea how the money was distributed throughout my night stand, couch and car. You can't use a sum to determine how the individual addends were originally divided.

 

Likewise in Cameron's piece he shows that full team WAR (the sum) with the defensive metrics added is highly correlated to team winning percentages however, that says nothing about individual player WAR (the addends).

 

Of course there are other problems too with Cameron's analysis vis a vis Marte v. Harper. The biggest being that by adding all the player's WAR to create team war he significantly increased the amount of defensive data being processed.

[Mike Sixel]

How does the eye test work, if each individual scout sees a player a handful of times? they might not get more than a dozen chances or so for that one scout......and UZR and the other defensive stats ARE eye tests, they are just eye tests where they eyeball every single play, chart it, analyze it, add it all up. Isn't UZR just an eyetest, but with a lot more observations? I've never understood how looking at 10-15 plays tells you more than looking at Every Single Play.

[diehardtwinsfan]

What you're referring is the "range factor" stat. Which not only has its own problems with sample size, but it compounds them by including the ginormous problem of pitching staff composition.

 

Low-k staff = more balls in play = more outs for the fielders = better fielders?

 

 

Just my guess, but that would be dramatically worse. I'd even bet on it.

 

You can normalize that stat using the team's average K rate pretty easily...

[The Greatest Poster Alive]

I really don't understand the compulsion to define a statistical category for defense.

 

Defense is just not something that can be defined on stats alone. Certain stats can get your a vague picture of what a player is like on the field... but it really doesn't do any justice to watching a player on the field.

[snepp]

You can normalize that stat using the team's average K rate pretty easily...

 

And a groundball staff vs flyball?

 

Different/odd outfield dimensions? (Fenway, TF, etc)

 

Accounting for errors and arm?

 

 

Why go through the trouble to make a lousy measure even moderately useful, when you can just use something superior right from the start?

[snepp]

I never understood the caveman's compulsion with learning. What was wrong with their dank, wet caves? Wheels? Fire? Who needs 'em!

 

 

The "compulsion" is that people like expanding their views on things, like learning new things, enjoy trying to improve things. No one is forcing you to use them, so why piss on those that do?

[cmathewson]

I never understood the caveman's compulsion with learning. What was wrong with their dank, wet caves? Wheels? Fire? Who needs 'em!

 

 

The "compulsion" is that people like expanding their views on things, like learning new things, enjoy trying to improve things. No one is forcing you to use them, so why piss on those that do?

 

I don't piss on those that do. It just bugs me when people use them wrongly to justify their overconfident opinions. To me, the 2010 Young/Punto example constitutes a reductio ad absurdum on the brokenness of WAR. And that doesn't even touch how much more broken it is for pitchers and catchers than it is for guys who make plays on live baseballs.

 

But we see the use of WAR to "price" contracts all the time, or to argue MVP awards or to do 100 things in isolation of context that are just wrong. It's a crude implement to begin with. With the wrong usage, it can justify all kinds of mistakes.

[LaBombo]

I really don't understand the compulsion to define a statistical category for defense.

Two words:

 

The.

 

Captain.

 

http://0.media.sportspickle.cvcdn.com/92/65/aa9848dc08470b64c3cf231ad0bb37cd.jpg

Defense is just not something that can be defined on stats alone. Certain stats can get your a vague picture of what a player is like on the field... but it really doesn't do any justice to watching a player on the field.

Nobody said anything about defining defensive ability with stats alone. But after years of Yankee propagandists talking up Jeter's mantle full of Gold Gloves, it would be nice to point to a reliable metric which makes a meaningful argument that they are full of crap.

[YourHouseIsMyHouse]

There's this thing called OPS, that is almost like WAR, but without the defense. I think that there should be a stat that includes SBs in OPS. I've never found anything of the sort, but perhaps it already exist. Like I was saying though, go OBP+a modified slugging (TBs+SBs/ABs)= True offensive worth. Caught stealing could also be included if you wanted more of a net stolen base deal. But I think that little things speed guys can do (or maybe assume they can), like scoring from 1st on a double or bunting, negate that.

[Kobs]

it would be nice to point to a reliable metric which makes a meaningful argument that they are full of crap.

Rafael Palmeiro was awarded a Gold Glove in 1999. You don't need any more metrics to discount that designation.

[Kobs]

I've never really been a fan of any of the more convoluted regression based stats. I'd rather the number have some actual meaning.

[Otto von Ballpark]

To me, the 2010 Young/Punto example constitutes a reductio ad absurdum on the brokenness of WAR.

 

Really? You watched Delmon Young for four seasons here and you're going to hitch yourself to his wagon?

 

I assume your beef here isn't with Delmon. He had an above-average offensive season in 2010 but played the field like he should have been a DH. And sure enough, his OPS+/WAR are almost a perfect match for full-time DH Hideki Matsui that year. I think WAR pegs him pretty accurately in that regard.

 

Your beef seems to be with Punto, who accumulated defensive WAR at a career-best rate in 2010. But the rate itself doesn't seem out of whack -- in 2010, per 1200 innings, he saved 28 runs at 3B and 19 runs at SS, but at other times in his career he posted rates of ranging from 19-24 over similar samples at 2B, 3B, and SS. Can a defensive player not have a career year like that? In fact, isn't Punto's career best defensive rates in 2010 not materially different than Young's career best offensive rates that same season?

 

And can a defensive player, having a career year like that at tough defensive positions, not have a half-season equal in overall value to a full season from a modestly above average hitting DH? Some of it could certainly be small sample -- Young certainly would have been better off in 2010 if his season had ended after July when he was OPSing .916, so why couldn't the same apply to a sample of Punto's defense?