WAR! What Is It Good For?

            Being officially a member of the Old School when it comes to baseball statistics, raised strictly on a diet of BA, ERA, OBP and SP, I struggle when it comes to understanding some of the new-fangled stats favored by the sabermetric crowd.


            Part of my problem is strictly understanding how one gets from Step A to Step B. So, while it’s easy for me to embrace a stat like WHIP (walks + hits per inning pitched) and OPS (On-Base Plus Slugging Pct.), it is more difficult to embrace stats like FIP (Fielding Independent Pitching), UZR ratings (a defensive metric) and WAR (Wins Above Replacement).


            The latter three are harder to understand because there are more complex formulas being used to come up with the figure and, in some cases, I don’t understand the rationale behind the why certain figures are used in the calculations, like why does this number get multiplied by 13? Where does that number 13 come from?


            Still, I plod on and attempt to understand what the metric means, even if I don’t understand how the metric comes about. Thus, I understand what FIP, UZR and WAR are meant to measure while not necessarily knowing how accurate a measurement it truly might be.


            It is with the understanding that I don’t understand the mathematics behind it that I present this column on a difference of opinion I have with the WAR stat as it relates to pitchers. And, since this blog has always dealt with the Texas Rangers in particular, I will use members of the Rangers as an example.


            I have seen on Rangers fan blogs and other MLB blogs in general the argument put forth that, using WAR as a benchmark, a team’s relief corps is more easily replaceable than the starting staff. In other words, a closer like Neftali Feliz, with 40 saves in 2010, still only graded out to a 1.0 WAR, while a starter like CJ Wilson was a 4.4 WAR (according to FanGraphs.com). Thus, CJ is the more valuable pitcher to have. These ratings indicate how many more wins a player is worth relative to your average minor league call-up replacement player or minimum contract major leaguer.


            The argument has also been made that the Rangers are foolish to have Feliz in the bullpen when his talent indicates he would be much more valuable as a starter from the perspective of WAR. This despite the fact Feliz has never once started a major league game. There is absolutely no evidence to indicate what his WAR would be as a starter. Yet these fans argue there are any number of pitchers who could replace Feliz as a closer and be just as effective as Feliz in this role.


            No matter how I’ve tried to wrap my brain around this, there’s nothing I can see that makes this argument seem right. I’m not saying Feliz wouldn’t be more valuable as a starter. I have no evidence to indicate that one way or the other. Yet, with Feliz just having spent two weeks on the DL, the results of the Rangers bullpen would certainly back up my view that replacing Feliz as the closer would NOT be easy, at least based on this WAR statistic alone. In the time Feliz has been out, Texas has blown as many saves as they’ve earned, if not more. Quality pitchers like Darren Oliver and Arthur Rhodes have had equal amounts of success and failure in the role during Feliz’ absence.


            I’m sure a true sabermetric fan would tell me the sample size is too small for me to come to the conclusion I’ve come to and maybe it is, but I would think sabermetrics would also strive to have statistics that mirror what most of us as fans see on the field. In other words, it’s easy to see through sabermetrics why Felix Hernandez deserved a Cy Young Award in 2010 despite a pedestrian 13-12 record. Just about every statistical category, both well-known and less-known, backed it up. I just don’t see how WAR backs up a relief pitcher and, more particularly, a closer.


            Again, I’m sure a true sabermetric fan will tell me how the formula for devising WAR for relief pitchers is different than starting pitchers because the devisers recognized the difference. Still, I don’t think they recognize the difference enough.


            First, take the statistic itself: WAR stands for Wins Above Replacement. In my brief research, I noticed that, in 2009, roughly 70% of a team’s wins were credited to a starting pitcher. Thus, a relief pitcher’s WAR would understandably be consistently lower than a starter’s. But that in itself can’t mean they’re easier to replace, can it? Maybe the WAR stat is accurate as far as Wins above replacement, but relievers also have stats like holds and saves that starting pitchers don’t have and WAR does NOT take these things into account.


            Certainly one can debate the criteria for the Save and Hold stats. There are camps that say a save should only go to a pitcher who has the tying and/or winning runs on base when he nails down the final out. Conversely it seems pretty silly to give a reliever a blown save when he comes in during the 7th inning of a one run game and gives up the tying run, when these days it’s very rare for a pitcher in that type of situation to pitch the final three innings of the game.


            Still, those statistics exist and I defy you to show me any team who has had any long-term success (ie more than one consecutive season) with the “Closer By Committee” approach. A player who can close out ball games is valuable to his team, certainly much more valuable than his WAR being listed as a mere 1.0.


            At the very least, I would argue one cannot compare the WAR of starters to the WAR of relievers and come to the conclusion that relievers are “easier” to replace than starters. Starters’ WARS have to be compared to starters and relievers to relievers, with one WAR not correlating with the other.


            Meanwhile, I would further challenge the sabermetric community to try to devise an alternative method of better showing the value of relief pitchers that take more of their role into account than the current WAR system does.

One comment

  1. Jeffster

    I agree completely. Personally, I think stats are overrated in general, as they have little to do with how entertaining a baseball game is. But the one constant in baseball is that you can’t really predict anything based on past performance. Alexi Ogando had never been a starter, but he was thrust into the rotation and dominated his first couple starts. Feliz might have done the same, but on the other hand, he might have flamed out terribly, and at that point, if you put him back in the bullpen with his confidence shaken, does he regain his form as the dominant closer? Maybe, maybe not. No stat you can pull out will truly tell the future, no matter how many times you run the numbers.