Wednesday, January 9, 2013

WAR is not the Answer

It would be great if we had a universal statistic that took all of the players in baseball and gave us a fair way to compare them.

Arguments about baseball would practically cease.  There would be no reason for them, as this one magical statistic would settle everything.  It might usher in a new era of understanding that could eventually lead to world peace.

Unfortunately, I don't think such a statistic is possible.  But even if it was, WAR is certainly not it.

This is not the stat you are looking for.

A quick primer for the uneducated:

WAR stands for Wins Above Replacement.  It is supposed to measure how many wins a given player has contributed to a team compared to a theoretical replacement player.  For a more detailed explanation, please check the WAR section on

I understand why people want WAR to work, and I will concede that the statistic does have some worth.  For instance, it rewards players who are versatile and help their team in many ways, as opposed to statistics like home runs, which only measure one particular skill.

But does it accurately determine a player's value to his team?  I don't think so.

Admittedly, most saberfans do not claim that WAR is the ultimate statistic, or that it is the best measure of a player's value.  They claim that it should only be used as a "conversation starter."

And yet, it seems like every time a saberfan debates a player's value, WAR is the first thing they point to.

Don't believe me?  Check out some of the things that were written during the 2012 AL MVP debate:

WAR backs up Trout's MVP candidacy

Stats Revolution doesn't have enough clout

Mike Trout is your AL MVP

It seems like every saber-minded writer was required to do two things when weighing in on the MVP debate:
  1. Make it clear that WAR is not perfect and they are not basing their judgement solely on it.
  2. Use Trout's WAR value as evidence that he deserved the MVP.
Some of these guys would make great politicians the way they endorse WAR while also refusing to fully commit to it.

On the other hand, I will take a firm stance: I don't like WAR.

I don't like it because I don't like the fundamental concept of a replacement player that the statistic is based on.

I tried to determine just what makes someone a "replacement player."  I've searched the web, and I've never been able to find an answer that isn't vague at best.  From what I can gather, a replacement player is a borderline major league player who can be easily obtained from the minor leagues.

But should a judgement be made based on a theoretical player?  If you're going to try to determine how many "wins" a player is providing over his replacement, then shouldn't you consider the actual player who would replace him?

For example, let's say that the Dodgers have Joe Shortstop playing the shortstop position. (Crazy coincidence, huh?)

Pictured: Joe Shortstop, or as some people call him, "Hanley Ramirez."

Joe puts up solid numbers both offensively and defensively.  According to the formula behind WAR, shortstops are more valuable than most of the other positions, because it is typically more difficult to find a good shortstop than a left fielder or first baseman.  Since he performs well at an important position, Joe has a high WAR.

Unfortunately, Joe suffers and injury and has to miss a few months of the season.  At first glance, it seems like the Dodgers are screwed because based on Joe's WAR, it will be very tough to replace him.

Thankfully for the Dodgers, they have their top prospect Billy Shortstop playing at AAA.  The Dodgers call up Billy, and he ends up playing about as well as Joe did.

According to WAR, Joe is more valuable simply because he plays shortstop.  But if you're going to reward him based on his position, don't you also have to deduct him some points because he doesn't actually provide much more value than his replacement?

Of course, there's no way to accurately determine just how well a given player's replacement would perform.  The best that could be done would be to project the production of all of a team's potential replacements, and make a calculation based on that.  But even that would be mostly guesswork, and nearly impossible to measure.

That impossibility is the main reason why WAR doesn't really work as a measure of true value.

As I said, I understand the motivation for coming up with a universal statistic.  But in this case, much like Marvin Gaye once said: WAR is not the answer.


  1. My problem with WAR comes from the other end -- not what consitutes a "replacement player," but the floating definition of "value."

    You can't just throw every position into a statistical woodpile and judge them all the same. One team full of heavy-footed sluggers might be sorely in need of a no-hit second baseman with good defense -- but the same player wouldn't be anywhere near as valuable on a team full light-hitting defensive wizards.

    How can you judge what "value" is when that terms means different things to different teams? The notion of establishing a fictitious "replacement player," devoid of any context, seems kind of pointless.

    1. Very good point. Sure, you might be able to say how valuable a given player is in some sort of void, but that's not how baseball actually works.

  2. Any stat that is based solely on imaginary players should NEVER be consider an official statistic, period. You may as well have given Trout 'eleventeen' as his statline if that's all you've got.