Posted by bad_luck on 4/10/2014 3:55:00 PM (view original):
I don't think it's ridiculous. Who is this "everyone" that you know personally?
A) It's obviously ridicuous. WAR is a statistic that defines what happens on the field - ie, how many "wins" did a player contribute to his team. Not "how many wins would he have contributed if luck were not involved." That's an absurd level of speculative, really. In general, jumping to FIP or xFIP or ERC (or, for hitters, RC, defensive runs saved, etc.) involves some degree of speculation, specifically invoking a number of assumptions about various results having "average" probabilities that may or may not have been reflected in one season's worth of sampling. Jumping from stats to WAR suggests that there is a simple mathematical relationship between production and wins, which again involves assumption of some averaged results which don't always correspond to what happens in a limited sample size. Which of course is why teams with lower aggregate WAR regularly win more games than teams with better aggregate WAR. The 162-game schedule may be the most grueling in American professional sports, but it's still fairly small, and statistical anomalies will happen.
FanGraphs WAR compounds the uncertainty involved in normalizing simple outcome-based statistics into cumulative value-based or predictive statistics with the uncertainty in converting player performance to team wins. That's too damn much uncertainty to make sense. And the fWAR results for Tillman and Kazmir bear that out in spades. Tillman threw more innings, in a tougher division, in a much more hitter-friendly ballpark, and has a lower WAR? How can you throw more innings, allow fewer runs per inning in a substantially more favorable run-scoring environment, and somehow provide your team with fewer wins? That doesn't make sense. At best fWAR is telling you how many wins somebody SHOULD have provided, outside the influence of luck, assuming all the assumptions are right.
B) You're setting up a ridiculous and unnecessary standard. Why should I have to know these people personally? In general, people who write/appear for the national sports news media - IE ESPN, MLB Network, MLB.com, NBC Sports, FOX Sports, Yahoo Sports, etc. - refer to bWAR when discussing WAR the overwhelming majority of the time. I don't have to know them to be aware of that. I just have to watch, listen to, or read their work. Sure, if you get your baseball writing on FanGraphs.com - which I sometimes do - people will tend to use fWAR. But outside of that, you see bWAR virtually everywhere. And if writers or on-air personalities just refer to WAR, and you look it up - which I used to do, out of interest, when I had too much free time - the number is ALWAYS a bWAR. So yeah, I don't know Buster Olney or Ken Rosenthal or Jon Heyman or Brian Kenney personally. But I know that when they talk about WAR, they talk about bWAR. And you can verify that in about 5 minutes.