Insert sweeping statement about the analytics “movement” in baseball here. Similar ones have been made countless times and from countless angles. Some talk about the progress the movement has made with regards to its acceptance by the mainstream baseball community. Others say almost the exact opposite. That seeming dissonance is a near perfect metaphor for a smaller discussion in the same vein. When we discuss WAR, we’re still talking past each other.
Over the past few days, we’ve gotten a few solid articles about the much-maligned and much praised statistic. Sam Miller at ESPN penned this in defense of WAR and Michael Hurley of CBS Boston turned in this piece as a response to Miller’s. Overall, I don’t have many problems with either piece. I think each one is more or less on point: WAR is simultaneously over-bashed and over-hyped, but each article has one thing I take issue with.
Starting with Miller’s piece, I will say it was long and fairly comprehensive. However, the thing I noticed most was something that was missing. In his musings on WAR, Miller failed to mention anything about the dubiousness of any and all defensive stats that go in to the calculations of WAR. As for Hurley’s piece, most of it was okay, but the part at the end regarding RBI rubbed me the wrong way. No one says that driving in runs in meaningless. It’s a misrepresentation of the argument; the rest of Hurley’s piece is pretty well worded, so I’ll chalk it up to a slip up in phrasing.
Briefly, this presents a problem. One side is mischaracterizing the opposition’s argument and the other has a hard time admitting its own flaws. I’ve shied away from using WAR over the last year or so, but it still has its merits, as do traditional stats. It seems like we’re getting close to having a legitimate conversation about these things, but both sides need to make ‘concessions.’ Those on the traditional side need to realize that those who espouse advanced stats aren’t trying to turn the game into a spreadsheet based computer simulation meant to rob joy and emotion from the game. If anything, I’ve become more passionate a fan because of my exposure to advanced stats. And on the other side, those who embrace advanced analytics need to admit that MAYBE their methodologies aren’t perfect and that decimal point differences in WAR aren’t all that significant. Until we do those things, we’ll just be talking past each other and talking in circles ’til we’re blue in the face.
If you enjoyed this article, please consider sharing it!
6 Responses to Still talking past each other when it comes to WAR
Leave a Reply
TIQ IQ
I took umbrage with this statement:
“There’s a reason companies interview candidates for open positions rather than just make a decision based on a resume.”
The thing is, the research suggests that doing open-ended interviews results in biased and sub-optimal selection decisions (i.e., basing decisions on whether I “like” you, rather than whether you are the right person for the job). So, yes – people still do these interviews, but it tends to screw up selection decisions.
If you want to roll this out to baseball – yes, people still evaluate a pitcher’s success based upon how many wins he achieved, but that is not an optimal way of judging whether the pitcher was better at his job than someone else. And if I was a GM, I would look at a lot of other factors instead of wins to decide on whether to offer that pitcher a big contract.
I’m and “old stat” guy but to tell you the truth, I’m really not sure why wins are awarded to pitchers anymore. When I was a kid, you just accepted stats…wins, losses, ERA…BA etc As kids,we never questioned why certain stats were kept. i remember thinking there should be a game winning hit stat kept….then they did keep that stat and then decided it was not important. Wins is a pretty poor stat for a pitcher if you ask me….and NOT because it is not indicative of a pitchers skill. (I leave that argument to others) i just think it is a team game and the pitcher often has little to do with the win.(and often they are) Quarterbacks don’t get wins…why should pitchers? Teams get wins.
The problem is not the stat, but rather, how much weight we put on it. Did the player who posted a 4.6 WAR really have a better year then the guy who posted a 4.4 WAR?
Maybe.
But WAR (and many advanced stats) are a best guess numerical representation of what went on on the field, averaged over the huge sample size of every play by every player. Matchups, situations and other factors can’t be taken into account. We already know WAR doesn’t really evaluate Catchers well, doesn’t take a lot of what 1Bman do into account, and of course relies on flawed defensive data.
WAR is a great stat. But like all stats, it has it’s limitations. In evaluating a player, I would look at as many different stats as possible, and try to come up with an overall conclusion.
It is an evaluation tool, not an absolute statement.
No! I’m right and you’re wrong!
Well, ever since WAR started, MGL, “inventor” of UZR model for defense, claimed that UZR and, following, WAR can’t be accurate to a precision greater than 0.5 wins (that is, 5 runs). So, WAR should be expressed as 2.5, 3, 3.5… or even 2, 3, 4, 5…
As far as I know, plenty of the best people in stats evaluation (like MGL, Tango, Rally (the guy who did what became B-R WAR model)) agree that defensive stats are nowhere near precise enough to judge players at scale more precise than -10, -5, 0, 5, 10,… runs. But, people who are the most visible in WAR presentation do it as precise as possible, without mentioning how accurate it is.
Edit: inserted missing space, edited commas.
Very well said, Matt. Couldn’t agree more.