5.03.2012

The War Against WAR

Baseball is a stat driven game, well past a time when success was measured by home runs and RBIs.  It's the era of OPS, where stats require equations that the average fan doesn't begin to understand or attempt to use.  And at the helm of the modern day statistical ship is the big one. WAR.  A stat that can actually tell you how many wins a player adds to your teams total.

It's also one of the most overused and values-less stats that floats around the baseball world.

What is WAR?

Wins Above Replacement, a stat that is intended to tell you exactly how valuable a player is to a team by measuring how many wins a player adds over a replacement bench or Triple-A player.  For example Matt Kemp, last years runner-up to the NL MVP award, had a WAR of 10 (by the baseball reference formula) for the Los Angeles Dodgers.  What that means is he was ten wins better than a player that the Dodgers could take off the bench or out of the minor leagues 2011.

One thing to remember about WAR is not that it's wins provided above a player that put up the major league average at his position, it's win's provided above a replacement player.  That replacement player is someone who is coming in off the bench or someone who is called up from the minor leagues.

The stat is put together by combining a number of different statistics, I won't try to sell you on it but this is a better write up than I could come up with.

In practice, this stat allows a user to say player X is worth more wins to his team than player Y.  A cut and dry measurement of how important a player is to his team.

So why doesn't it work?

No Standard Formula

To find out the WAR number of a player you can go to a website like Baseball Reference and find it in their stat line. Or you can go to Baseball Prospectus and find it in his stat line.  Or you can go to Fangraphs and find it in his stat line.  There's only one problem, all three numbers will be different.

Take Jose Bautista for example, according to Baseball Reference his WAR was 8.5 in 2011.  On Baseball Prospectus it was 10.3 and on Fangraphs it was 7.1. Which of these stats is right? Is one better than the other?  If all three websites use a different formula to come out for the same stat how is any one of them valid?

There a simplicity to measuring things like batting average or earned run average because every time you measure it your numbers are going to be the same.  The WAR of a player cannot be three different numbers and still be taken seriously.  If one site values one aspect of their WAR equation more than another then the stat becomes subjective.  Stats can't be subject to a value judgement like that.

Defensive Misgivings

Measuring defense in baseball is by far one of the most difficult things to do.  It's been widely accepted that errors and fielding percentage tells us very little beyond how many majorly botched plays a defender makes in a season.  There have been numerous attempts to accurately quantify defense but thus far nothing has taken a lead.

WAR uses these less than sound defensive metrics, be it one of the many attempts at a zone defensive ratings or otherwise, in its formula.  This is unsound.  Until there is a more widely accepted measurement of defense all of the WAR ratings will be based on ratings that are, in and of themselves, unreliable.

Baseball will find a valid system of measuring it's players defensive capabilities but until they do, basing a statistic on the current flaw-ridden defensive metrics only serves to poke more holes in a stat that has as many holes in it as swiss cheese.

The Replacement Player

Assumption.  That is what the presumed replacement value is.  It is an assumption about what the replacement player can do, for good or ill.  When a player has a negative WAR, like Casey McGehee did last year, can we say that the replacement player would have been better than he was?

Perhaps, but that also draws a conclusion that this replacement player will perform at a certain level.  WAR generalizes.  One teams replacement option is different then the next team's.  If you have a top prospect waiting in the wings, like the Brewers did with Taylor Green, it probably would have been a safe bet to say Green could have done a better job.  But if you didn't have a good option?  If your replacement level player has a negative WAR?

That's the problem of WAR, a replacement player can have a negative wins above replacement value.  So then what? Another replacement player with an even lower value?  The stat becomes ineffectual.  Even more to the point, you just can't measure a real players actual performance against a generalized, assumed performance of the so-called replacement player.

Winning the Battle

The fact of the matter is, WAR isn't going away.  As long as performance can be measured by a variety and combination of statistics then new statistics will be derived to measure what a player actually contributes to his team.  It's a noble effort, to establish how many wins a single player can add to his team's performance over the marathon that is the professional baseball season, unfortunately it's also one that's inherently flawed.

Should the metric world cease it's attempts at a better measuring stick for Major League players? Not at all. But for those that consider themselves statistically enlightened they must also recognize the problems that an all-encompassing stat like WAR can bring.  You can use statistics to prove anything and WAR gives credence to that old standby.

The truth of the matter is that quoting a player's WAR is about as convincing as quoting his RBIs.  To measure a players performance you must use several statistics, not several statistics mashed together in some sort of sabermetric stew.

Maybe some day baseball's all-consuming and all-enveloping stat will come to fruition but until that day having more than one way to measure a player is not a bad concept.