By Jason Wojciechowski on September 2, 2011 at 8:30 PM
Bill Simmons wrote about the MVP in baseball at Grantland. I of course object to pretty much everything he says. To keep myself organized and focused, I'm going to go in a very linear fashion here, paralleling his line of argument, but I'm not going to do an old-school fisking with blockquotes and all that jazz. (I also stop midway through his article, because the rest of it is unreadable number soup, and because, as you'll see in point seven below, he said something at that point that made me question whether this is a man even worth engaging with.)
First thing's first: stop saying "WAR." Not to channel Tom Tango too closely here, but WAR is a construct, not a statistic. Fangraphs and Baseball-Reference have done us no favors by having different statistics both labeled "WAR" but many of us have figured out that we should, on our own, label these things "fWAR" (for Fangraphs) and "rWAR" or "bWAR" (for Baseball-Reference; the "r" stands for "Rally," an alias of Sean Smith, the creator of the WAR calculation that Baseball-Reference favors). This is, perhaps, sub-ideal, but it's also not hard to understand.
This doesn't actually matter that much, but it's potentially a signal of something that does matter. Simmons spends the entire piece citing "WAR," but never explains what calculation he's using. I could figure it out by comparing rWAR to fWAR for one of the seasons he cites and seeing which agrees with the Simmons-quoted number, but I should not have to do this. The more important thing that the inability to distinguish WARs tells us is that Simmons has not actually thought about what WAR means, how it is calculated, and so forth.
I'm not proposing that everybody who wants to use a statistic be required to take a six-hour course on its calculation and use, getting into the heavy weeds of all the little pieces that go into it. What I am proposing is that people who want to do analysis think a little bit about the tools they are using to perform their analysis. When I am discussing a player's overall value, I do not cite batting average. Do you know why that is? It is not because it is hard to find on Fangraphs and sabermetricians make fun of how crummy a statistic it is for measuring a player's value. It is because I understand that it is crummy for measuring a player's value -- I understand that it does not contain walks, a significant method of getting on base, and that it does not differentiate between home-runs and singles. I started reading blogs and Baseball Prospectus when I was in college and I learned these things from them. I read the reasons why they used VORP and (what was then called) EqA and I thought about it. I decided that I believed them. They taught me, but I was not a passive receiver of new statistics with which I could impress my (very nerdy) friends. I was learning. I would like to think that it's not too much to ask the proprietor of one of the most-read and best-capitalized sports-analysis outlets do the same if he desires to write about baseball.
Now, if you had to guess, based on reading this piece, whether Simmons knows what goes into WAR and how it works, would you guess that he does? I wouldn't. So what should we think about an analyst who blindly quotes statistics that other people have told him to use? Why should we listen to him?
In one paragraph: Simmons brings up steroids in one sentence, dismisses them in the next, predicts that people will think a thing in the third sentence, and says that they shouldn't think that thing in the fourth. What does that add up to? Bupkis. Why even mention this? Because he just can't bear to write a Bautista piece without mentioning steroids? I don't know. Pointless paragraph. (Not unlike this section.)
In the next paragraph, Simmons calls BABIP a statistic that has "gotten so good ... [that] you can predict with shockingly decent accuracy when certain players might streak or regress ... ." I am not sure that Simmons understands what BABIP is. It is not a statistic that has "gotten good." It is as purely descriptive a statistic as exists -- take balls in play and ask how many of them fell for hits. Don't adjust for park. Don't adjust for whether the balls were line drives. Don't weight some balls over others. Don't do anything but say "yes" or "no" for every single ball in play, add up the yeses and then divide by the total. Does Simmons get this? It's weirdly unclear (weirdly because of how simple BABIP really is).
Also, if sabermetricians are actually predicting "when certain players might streak," then jiminy, we're all doing better than I thought.
I'm being more sarcastic than I need to, but I think I know what Simmons thinks, and I think I understand the misimpression he's under -- Simmons thinks that we can look at a player who is doing poorly, examine some of their statistics and determine whether that player is likely to do better in the future. And we can! (Sorta.) But I think Simmons further thinks that when we do those analyses, we are saying that the player is due for a hot streak to return his numbers back to normal. This is not what any analyst does, of course, because that's a completely fallacious understanding of regression to the mean. (So is thinking that "regress" only means "do worse," but that's a quibble.)
Simmons says this, which Colin Wyers brought to Twitter's attention earlier today: "They believe every act on a baseball field can be measured, and that baseball is an individual sport that masquerades as a team sport." No. Most certainly not on the first count, and probably not on the second.
Every single "new schooler" (his phrase) worth his or her salt can tell you that we have trouble measuring all sorts of things that very likely do matter: fielder positioning, how hard and where a ball is hit in play, catcher framing, coaching, emotion, nerves, and on and on. Some of these we have no idea how to measure (and thus also have no idea whether they're meaningful) and others we're making progress on, but there is very likely not a single person in the world who thinks we're accurately measuring every act on a baseball field.
Now, the "individual sport" thing is more troublesome, and it was the focus of the Twitter discussion today.1 Still, I think we can agree that a team interacts on defense, that, for example, the feed from the second baseman to the shortstop on a double-play attempt matters. I cannot imagine a "new schooler," if pushed on this point, would insist that defense is an entirely individual endeavor.
Sixth, Simmons finally gets down to the point and engages in the mind-twisting on the definition of "value" that characterizes every article of this type since Bill James was mimeographing articles in a sausage factory. The core of Simmons's misstep on this point is jumping from the individual to the team without justifying that move. Here's the key quote: "'valuable' suggest pretty strongly that something positive happened." Yes! (Well, in conjunction with "most," anyway, since "value" is actually a content-neutral word. Whatever.) Right? What's the problem? Just that Simmons, based solely on his idea that "something positive happened," believes that the "something positive" must have been a team event. Jose Bautista made many positive things happen, but those don't count because his team didn't win enough. Adele made many positive things happen, but those don't count because the VMAs sucked. (He uses Adele at the VMAs as an analogy. I won't comment on this.)
Where in the piece does Simmons support this key jump from individual to team on "something positive"? Nowhere. (But you knew that, or I wouldn't be writing this.) Simmons simply believes that value refers to the team doing well, but is unable to support that belief with citations to any dictionary definitions of value, common usages of the word in other areas of life, or any other evidence that might help me take his argument seriously.
Let's take Simmons's argument further. Shouldn't we wait to award the MVP to the best player on the team that reaches the World Series? He argues that nobody cares whether the Blue Jays win 80 games or 70, which might be true. But in ten years, do we remember who lost in the Division Series? In other words, how do we determine what a good team is? What is "success"? These questions have to be answered if you want to take the position that team success matters; without answering them, you've created a system whereby you can argue for whichever candidate you want, not an independent framework that you can apply to the particular situation of the season at hand. Anyone who wants their opinion on baseball awards taken seriously must have a defensible system. Otherwise they're just saying words.
This is the worst thing I've ever seen someone who's supposedly serious about analysis say: "it's my interpretation, which means that I am neither right nor wrong". Watch how easy this is to disprove: I hereby interpret the word "idiot" to mean "Bill Simmons." Am I right or wrong?
"You're right!" says the crowd. You guys are a hoot. But no, really, I'm wrong. The word "idiot" can not mean "Bill Simmons." This is a change in the established English language that cannot be supported by logic or evidence. It is entirely possible for interpretations to be wrong.
If Bill is actually reading this (he's not), I hope he takes a second to let this sink in. Forget everything I said above. It's all baseball. This part actually matters: opinions can be wrong. If it is your considered opinion that Derek Jeter is the best player on the Yankees, if your interpretation of the word "best" is that Jeter is that, then you are wrong. If your opinion is that the sun orbits the earth, then you are wrong.
In something as subtle and shifting as language, we can argue about reasonable interpretations of words. Lawyers do this all the time about things like "commerce" or "violence" or "intent." But there are interpretations of those words that are wrong! If someone physically grabs my arm, puts a knife in my hand, and shoves that hand into another person's chest, then I did not intentionally stab that person. An interpretation of "intent" that calls this intentional would be wrong.
We can argue about interpretations of the word "value." If we wanted, we could take an extreme position and ask which player created the most revenue for their team. I don't know how we'd measure that, but we could try. That would be an interpretation of "value" about which I would simultaneously say that we should not use it and that it is not a wrong interpretation of the word "value." Having established that there is more than one reasonable definition of "most valuable," however, does not establish that any interpretation must be taken seriously, that no answer can be considered wrong.
No, what Simmons is doing here is transparent and intellectually bankrupt. Rather than fully argue his position and support it with the best evidence that can be mustered, he claims, "Well, this is my opinion, man, and you can't tell me my opinion is wrong!" The absurdity of such a claim is apparent, or should be to anyone with a modicum of the integrity that Simmons so sorely lacks.
Look, lots of people are wrong on the internet, especially old people writing about baseball. It takes a special personality, though, to tell you right to your face that you're simply not allowed to say "that's wrong." Congratulations on all your success, Bill.
You could go check it out if you want -- find Colin Wyers and look in his timeline. Mike Fast and Dave Metz were also involved. So was I, of course, taking my rightful place as "the guy who chimes in on all the philosophical issues of sabermetrics despite doing no research of my own." It's a heavy burden to bear, I tell you.) ↩