Why do people value WAR/Arbitrary formulas.

[glenn__davis]

This has been a great thread. Good arguments from all sides.

Pickles is being unfairly piled up on here, IMO. He most definitely is not a troll, and the suggestion that he is was very silly.

He is correct in that if someone makes an argument using batting average they are often derided, but if WAR is misused it is often not called out. I have seen many posters here make arguments along the lines of "Player X is worth 3 WAR, Player Y is worth 2 WAR, therefore the Orioles would be adding 5 wins". National columnists have begun to make the same misinformed arguments.

For me personally there is no question that the misuse of the stat is more egregious than the stat itself, but that misuse has begin to get so widespread that it does get annoying.

[Lucky Jim]

This has been a great thread. Good arguments from all sides.

Pickles is being unfairly piled up on here, IMO. He most definitely is not a troll, and the suggestion that he is was very silly.

He is correct in that if someone makes an argument using batting average they are often derided, but if WAR is misused it is often not called out. I have seen many posters here make arguments along the lines of "Player X is worth 3 WAR, Player Y is worth 2 WAR, therefore the Orioles would be adding 5 wins". National columnists have begun to make the same misinformed arguments.

For me personally there is no question that the misuse of the stat is more egregious than the stat itself, but that misuse has begin to get so widespread that it does get annoying.

Perhaps. Let's set aside whether what he was doing fits the definition of a troll. As BD pointed out, at best, Pickles was indulging in pretty "egregious" sophistry (making his self-designation as "Socratic" all the better) and he was either (i) being intellectually disingenuous, or (ii) failed to understand his own argument at the beginning of the thread. His final position (i.e., "people misuse the statistic") is a far cry from his repeated assumption of the OP's conclusion (i.e., that WAR is speculative and arbitrary). Treating inconsistent lines of argument as if they are consistent is a Treaian maneuver. The issue was largely settled by post 24 of the thread:

How exactly should you use a flawed, subjective statistic? What's the best way to do that?

How is it not? Are there not multiple WAR numbers? Do the individual systems not change frequently?

Seriously? It's the definition of subjective.

All statistics are "flawed." You use them properly by knowing their limitations and not exaggerating what they tell you.

Right. If the inputs are constant, and the data entry is rigorous, then it's neither arbitrary or subjective. There's no prelapsarian statistic that didn't involve "subjective" or "arbitrary" choices of what to value.

Pages and pages later, he's now arguing, not that the OP was valid, but that:

...my real objection, is the way people use WAR #s to try to end any real discussion of player value, as if we already have something handed down from Sinai. I'm not saying you are guilty of this, but surely you've seen it happen time and time again on this board and others.

Well, which is it? Are all stats subjective? Are all stats therefore "unusable"? What are the "degrees" of subjectivity? And why would you argue that something is "subjective" and "arbitrary" if you are later going to concede that everything is "subjective" and "arbitrary." To me, disingenuously arguing something that you don't believe is trolling. Having defended myself, I'll concede that I was too harsh, and those kinds of posts short-circuit conversations. There was some value to be had in here, if only in terms of entertainment.

[DrungoHazewood]

This has been a great thread. Good arguments from all sides.

Pickles is being unfairly piled up on here, IMO. He most definitely is not a troll, and the suggestion that he is was very silly.

He is correct in that if someone makes an argument using batting average they are often derided, but if WAR is misused it is often not called out. I have seen many posters here make arguments along the lines of "Player X is worth 3 WAR, Player Y is worth 2 WAR, therefore the Orioles would be adding 5 wins". National columnists have begun to make the same misinformed arguments.

For me personally there is no question that the misuse of the stat is more egregious than the stat itself, but that misuse has begin to get so widespread that it does get annoying.

So let's try to frame the arguments as "people don't know how to use the tools" as opposed to finding fault with the tools.

JTrea claiming you can just add up your favorite version of WAR to predict wins with high probability in coming years is no different than someone cutting their leg off with a chainsaw. It's not the chainsaw's fault any more than it's WAR's fault.

[Frobby]

Well, which is it? Are all stats subjective? Are all stats therefore "unusable"? What are the "degrees" of subjectivity? And why would you argue that something is "subjective" and "arbitrary" if you are later going to concede that everything is "subjective" and "arbitrary." To me, disingenuously arguing something that you don't believe is trolling. Having defended myself, I'll concede that I was too harsh, and those kinds of posts short-circuit conversations. There was some value to be had in here, if only in terms of entertainment.

It seems to me that the defensive component of WAR has a degree of subjectivity, despite the best efforts to make it as objective as possible. Either that, or it remains a flawed objective measure because it doesn't take all variables into account in way that consistently works. But, maybe that is my "subjective" opinion.

[Mizerooskie]

Well, which is it? Are all stats subjective? Are all stats therefore "unusable"? What are the "degrees" of subjectivity? And why would you argue that something is "subjective" and "arbitrary" if you are later going to concede that everything is "subjective" and "arbitrary." To me, disingenuously arguing something that you don't believe is trolling. Having defended myself, I'll concede that I was too harsh, and those kinds of posts short-circuit conversations. There was some value to be had in here, if only in terms of entertainment.

Stats are not subjective. They are facts/points of data.

WAR is not a stat.

[Lucky Jim]

Stats are not subjective. They are facts/points of data.

WAR is not a stat.

As I noted earlier, WAR is a model. Whatever you want to call the individual numbers that are derived from that model, the argument is largely semantic at this level of discussion.

That said, is batting average a stat?

[Lucky Jim]

It seems to me that the defensive component of WAR has a degree of subjectivity, despite the best efforts to make it as objective as possible. Either that, or it remains a flawed objective measure because it doesn't take all variables into account in way that consistently works. But, maybe that is my "subjective" opinion.

Sure it does. Absolutely. But, as made clear in this thread, discretion and subjectivity are endemic to any kind of statistics/statistical modeling in baseball.

[Mizerooskie]

As I noted earlier, it's a model.

Is batting average a stat?

Yes, batting average is a stat. It is a factual calculation of hits/at bat. The scoring of a batted ball that puts a player on base is subjective. The statistic is not.

WAR, and other models are subjective, because the creators decided how much to weigh certain statistics/other models in the calculation.

[Mizerooskie]

Sure it does. Absolutely. But, as made clear in this thread, discretion and subjectivity are endemic to any kind of statistics/statistical modeling in baseball.

False. Subjectivity and discretion are never found in statistics, else they wouldn't be statistics.

[TerpsBaseball26]

I think it's also important to keep in mind that in any statistical modeling procedure, there will be some sort of variability around the solution. In the case of WAR, there is a good chance that the variability around the solution estimated for a player in any given season will be somewhat higher due to inconsistent defensive metrics used. Thus, when someone reports a single WAR number for a given player, they fail to take into account the variability around this solution. So, a 2.0 WAR player for a season may actually be a 0.5 WAR player or a 3.5 WAR player. This may add to the illusion that the model is subjective, but rather it is normal variability inherent in statistical modeling. As the defensive metrics (in particular) continue to improve, so too will WAR modeling.

[Lucky Jim]

Yes, batting average is a stat. It is a factual calculation of hits/at bat. The scoring of a batted ball that puts a player on base is subjective. The statistic is not.

WAR, and other models are subjective, because the creators decided how much to weigh certain statistics/other models in the calculation.

But the choice of what to exclude or include in a calculation is its own form of weighting, no?

[Mizerooskie]

I think it's also important to keep in mind that in any statistical modeling procedure, there will be some sort of variability around the solution. In the case of WAR, there is a good chance that the variability around the solution estimated for a player in any given season will be somewhat higher due to inconsistent defensive metrics used. Thus, when someone reports a single WAR number for a given player, they fail to take into account the variability around this solution. So, a 2.0 WAR player for a season may actually be a 0.5 WAR player or a 3.5 WAR player. This may add to the illusion that the model is subjective, but rather it is normal variability inherent in statistical modeling. As the defensive metrics (in particular) continue to improve, so too will WAR modeling.

There's no illusion that the model is subjective. It is subjective. That's clearly evidenced by the fact that it's not standardized across all publications. Different statisticians have different opinions as to how it should be calculated.

[Mizerooskie]

But the choice of what to exclude or include in a calculation is its own form of weighting, no?

No, it's not. The definition of batting average is hits/at bat. There's no choice there. It is what it is.

[Lucky Jim]

False. Subjectivity and discretion are never found in statistics, else they wouldn't be statistics.

My point was that the determination of a "hit" is, itself, subjective and discretionary at times. And, in a different fashion, choosing to calculate BA by using at-bats instead of plate appearances is both subjective and discretionary. Or calculating all "hits" equally. There's no objective reason for making that determination. I'm not sure how that choice (or even the component points - how we define an AB and how those ABs came to be) can't be considered fundamentally similar to any sort of discretion that creeps into models.

I do understand your point, though. But it's been gotten at above in this thread, perhaps best by Drungo. The point is that the statistical purity you identify in "batting average" is self-limiting.

[Mizerooskie]

My point was that the determination of a "hit" is, itself, subjective and discretionary at times. And, in a different fashion, choosing to calculate BA by using at-bats instead of plate appearances is both subjective and discretionary. There's no objective reason for making that determination.

The determination of a 'hit' has absolutely nothing to do with the statistic. And choosing to calculate using at-bats instead of plate appearances isn't discretionary or subjective. The choice not to count walks and reaching on errors as at-bats was.