The NL MVP race has heated up lately, and I think that provides as good an opportunity as any for me to explain where I feel WAR fits on the scale of useful baseball statistics. Additionally, my father managed to write me into his healthcare blog (Yes, that picture is me. No, my dad isn't tasing me with a foam football, despite appearances), looking at how statistics like WAR can be used in other fields (teachers are already being looked at by a system called “value-added modeling,” which isn’t unlike WAR in that it compares each teacher to a baseline replacement). Every stat falls somewhere on the continuum between statistics that are completely useless (some more traditional statistics fall closer to this end than you might think) and what I’ve heard referred to as “The Truth,” a mythical statistic that will truly describe a player’s contribution to his team exactly and perfectly. The development of sabermetrics is far from over. Innovative sabermetricians across the world are undertaking pioneering research, arguments are being debated and discussed, and entirely new lenses through which we can view the game of baseball are being formed. Advances in the past few years, such as hit and pitch f/x, have brought new tools for sabermetricians to analyze the world of baseball, and upcoming field f/x (that picture is awesome) and other techniques promise to do the same, especially in the area of fielding statistics.
WAR isn’t perfect, by any stretch of the imagination. First of all, there are two different kinds. One, which you can find on Fangraphs, uses UZR as its fielding component, Weighted Runs Above Average for hitting, and a ballpark-adjusted version of FIP for pitching. For more information, check out Dave Cameron’s series on Win Values (scroll down toward the bottom). Baseball Prospectus’ metric (known as WARP-1), however, uses Baseball Prospectus’ Fielding Runs statistic for fielding. Hitting is based on Equivalent Average, and pitching on Defense-Adjusted ERA. Though good players will score well on both systems and bad players are always going to look bad, these systems can result in different scores and some changes in the leaderboards for the season. For example, Fangraphs’ WAR has the NL position-player leaderboard as Zimmerman (7.1), Votto (7.0), Holliday (6.4), Pujols (6.3), and Tulowitzki (6.1). Baseball Prospectus, however, sees the race as Pujols (7.0), Holliday (7.0), Adrian Gonzalez (6.6), Carlos Gonzalez (6.4), and Votto (6.4). Zimmerman scores comparatively poorly on Prospectus’ WARP, largely because Fielding Runs has him at 7 runs below average, while UZR scores him as a brilliant fielder, at 15.9 runs above average.
I generally believe Fangraphs’ WAR is a better indicator of overall value, so that’s what I’m going to be using on this site. However, future research could certainly prove Prospectus’ WARP to be a more effective overall value indicator, or cause us to move toward another measure. In sabermetrics, much like any academic science, we need to be ready to accept the latest research and data and base our conclusions on the best tools we’ve got. As the Zimmerman case illustrates, fielding statistics still lag behind hitting and pitching in terms of their effectiveness and preciseness, and the development and widespread use of field f/x will likely change this in the future. When it does, we have to be ready to accept that the statistics we’ve been using are no longer current. The faster these advances come, the quicker our statistics become outdated, so in a few years we may perceive people using WAR and WARP much the way many in the sabermetric community see old-school evaluators still using pitcher wins, batting average, and RBI as their main statistical tools.
So that’s where WAR and WARP stand. They’re not perfect, but they seem to be some of the best tools we have, so we have to use them until we’ve got something better. Check back tomorrow, when I’ll be using our newfound image of WAR to take a look at the race for the MVP award.