Why do people value WAR/Arbitrary formulas.

[MrOrange82]

Don't you think that's a little extreme? Agree or disagree with it, it offers a measure of total player value. That's information people want. That's why people value it.

WAR...huh, yeah, [but] what is it good for?

[Lucky Jim]

And the inputs are constant?

In each batch of modeling? Yes. They may change over time, but in each calculation, they apply the same model to each data set. None pretend to be the only model. And none pretend, at the time, toward perfection.

I guess we could throw out all statistics, which may be the only cure for the epidemic of subjectivity.

[sangar]

Is WAR a characteristic of perception relating to a subject? Yes, it is subjective.

Is WAR dependent upon individual discretion and based on individual preference? Yes, it is arbitrary.

Is WAR a useful statistical model? Yes, if used correctly.

Everyone is right, give it a rest guys.

[Fired-Up]

Don't you think that's a little extreme? Agree or disagree with it, it offers a measure of total player value. That's information people want. That's why people value it.

Alright. I guess I'm just going to have to live with the fact that some people value a statistic that is created arbitrarily by someone else. Slugging percentage for example is concrete. It doesn't attempt to be anymore than it is. WAR tries to weigh the values of power, speed, defense, etc. Just because player X has Y amount of power doesn't mean that he creates Z amount of wins.

[Flash- bd]

I think the critics of the stat itself are really, deep down, more concerned with how it is used.

First of all, WAR is an attempt to take as many components possible of what a player does, and actually match that to winning games. That is an EXTREMELY ambitious thing to try to do -- if it were ever fully accomplished you could have a perfect evaluator for all trades, at least after the fact, and could in fact compare apples and oranges. And WAR, as it exists, is as close as any stat has come to achieving that.

But it is STILL a work in progress. As some have pointed out, there are multiple WARs out there. And one that tries to include defense and uses UZR, which is known to be very questionable when applied to small and even medium sample sizes. Even a full season UZR is suspect. So when you use a stat like that to determine the defensive component of WAR, the resulting value is going to be very questionable even when describing a full season of a player's performance.

OK, it's still alright to use the stat as long as you understand those limitations. But because the ambition of the stat is to equate a players' production with actual WINS, and the name of the stat even states that, people tend to use it as a conclusive argument ender. Oh, that trade will result in a net -1.5 WAR, so it's a bad trade. After all, we want to win more games.

If it is used as a rough yardstick ("he's a 1 WAR player, he's a 2 WAR player), it is fine. But to conclusively use it to evaluate a deal, or to determine how much to pay a free agent -- I don't think it's there yet. It's a valiant attempt, and better than any other stat that has yet to be created, but I don't think it has yet reached its lofty goal of correlating player performance to wins, especially the defensive component. Yet a lot of people toss it around as if it has succeeded in conclusively tying performance into actual wins. I read a lot of arguments that use WAR that, at least they seem to, imply that they are ending the particular argument and putting in the FINAL WORD because they have invoked WAR to evaluate the trade, or contract, or decision. And I don't believe that the stat is "ready for prime time" enough to be used that way.

Yeah, you've basically expressed--far more eloquently--what I was trying to get at earlier in the thread. The name itself is a problem, at least as far as the impact it has psychologically speaking. Like I said, Runs Above Replacement or Value Points Above Replacement would be a lot better.

[Pickles]

Don't you think that's a little extreme? Agree or disagree with it, it offers a measure of total player value. That's information people want. That's why people value it.

I think it has value as a conversation piece and as a kind of "baseline" so I don't think it's worthless; I just find it annoying, and frankly, against everything sabermetrics is supposed to stand for, when people use it as some kind of definitive proof.

[Flosman]

I just created this thread so I could ask a simple question. I thought there would be a few answers. I thought People would explain the value of an arbitrary statistic which is what WAR is. I didn't think I'd open such a can of worms. If you guys want to lock the thread then lock the thread. If not, then perhaps somebody can explain to me why this horrible, disasterous thing called WAR is valued by some.

I think WAR, and more directly the WAR components, do a fair job of evaluating the relative value of a players production. I think the measuring of the defensive contributions is very problematic but that is the case for any measure of defensive contributions. I think some of your problem with the stat is that it is relative and that leads to identical numbers producing different numbers because of changes in the production environment.

[24fps]

I guess we could throw out all statistics, which may be the only cure for the epidemic of subjectivity.

Solipsism, the true dismal science.

[crissfan172]

I wouldn't call WAR "arbitrary". I'm sure a lot of statistical analysis went into making the stat as accurate as it possibly can be. That being said, I think it's a lazy statistic. It gives people a simple number to compare players at different positions, rather than making people really dig into the numbers themselves. Plus, I feel like it bases way too much of its value on advanced defensive metrics, which still have plenty of room to improve upon themselves. For that reason alone, I feel that the stat can be flawed or misleading, especially when you consider that UZR is (supposedly) even more questionable with smaller sample sizes (like...say...only 1 season). I don't mind BB-reference's oWAR, though I find it much more useful (and satisfying) to look at a player's slash-line and other various offensive metrics to gauge a player's value, rather than just looking up their WAR.

[Flacco Machado]

Too many big words. Head hurt. Me go sleep now. :002_ssleepy:

[Pickles]

In each batch of modeling? Yes. They may change over time, but in each calculation, they apply the same model to each data set. None pretend to be the only model. And none pretend, at the time, toward perfection.

I guess we could throw out all statistics, which may be the only cure for the epidemic of subjectivity.

So you honestly don't recognize the difference between the way ERA or Batting Average (for all their flaws) are calculated and the way WAR is calculated?

[Lucky Jim]

The inputs are far more subjective. What is an earned run, and why does it matter? Who decides when a run is earned and where did that rule come from? Who determines an error, and how is "subjectivity" controlled for?

[Flash- bd]

YES! This thread has the potential to plumb depths of philosophy and psychology I never thought we'd get to here on the OH. I was planning on putting on a movie but I'm going to have to see where this leads.

[sangar]

The inputs are far more subjective. What is an earned run, and why does it matter? Who decides when a run is earned and where did that rule come from? Who determines an error, and how is "subjectivity" controlled for?

I believe this is exactly what the OP was getting at.

[Pickles]

The inputs are far more subjective. What is an earned run, and why does it matter? Who decides when a run is earned and where did that rule come from? Who determines an error, and how is "subjectivity" controlled for?

Yes, knowing all the "subjectivity" that goes into something fairly pedestrian like ERA, and then imagine all the "subjectivity" which goes into WAR, how can you, w a straight face mind you, claim that WAR isn't subjective?

I guess we're arguing about degrees.

I will concede that most baseball stats partake of a certain subjectivity, but that doesn't mean they are all equally so.