Archive for the ‘rants’ Category

In Defense of Compassionate Sabermetricism

May 23, 2009

If I’m going to have a horribly unhealthy, gut-busting, productivity-killing Friday lunch, I’m a big fan of Panda Express’ Orange Chicken. And there’s a decent copycat place a couple blocks from the office, but it was a nice day yesterday, and I was up for a walk, so I went for the real thing. To get that, you have to head to the James R. Thompson Center, a big gathering point for a lot of Chicago that, as I understand it, houses some government offices and whatnot. The Panda Express is really all I’m interested in.

So I get there, and there’s this big protest going on right outside the building. Up close, people are waving signs about the right to life and how gay marriage is destroying our families, milling about in the general neighborhood of someone who is speaking ineffectively into a megaphone, while across the street is another group of people doing their best to drown out this first group with shouts like “What do we want?” “Abortion rights.” “When do we want it?” “Now!” and “Fascists go home!” and I’m thinking to myself, what are these people (any of them) doing here, really? Do they expect to convince anyone by labeling the other side murderers or fascists, or by just being louder? Or do they just like to hear themselves talk? Is there just nothing better to do on a pleasant Friday leading into a holiday weekend?

That’s basically how tHeMARKsMiTh sees the world of baseball fans and writers: the internet-savvy sabermetric crowd against the talk-radio-and-newsprint traditional crowd, both sides trying to shout each other down, never getting anywhere. (Of course, that doesn’t even remotely do justice to his post. Read it yourself; I’ll still be here when you’re done. Ready now? Good.) A couple basic things to get out of the way:

  1. I agree with most of his main points. There’s a lot of shouting into the abyss that goes on on both sides, a lot of name-calling and making fun, and it’s hard to see how any of it does anything at all other than making people on the same side feel smug and superior at the other side’s expense. (Okay, I have to make an exception for these guys, who were just too funny. And JoePoz, who’s kind of a fence-straddler, anyway. But otherwise, I don’t see the point.)
  2. I don’t think traditional stats (or most of them, anyway; sorry, Holds and Fielding Percentage) are completely worthless. You’ve seen me use HR and RBI a bunch of times already. Stats like those give context; even if you believe that VORP or WAR or Win Shares are a perfect measure of player value, think of the traditional stats as the splash of color in the crystal-clear black-and-white picture. They tell the story: what kind of hitter he is, where he likely hit in the lineup, and so on. WAR will tell you that Mark Teixeira and Carlos Beltran were almost exactly as valuable as each other in 2008, but don’t you want to know a little more than that? That’s where I think runs, RBI, HR, SB, and so forth come in handy.
  3. Another main point of Mark’s is that neither side has it completely right. I agree with that, too: there’s not much “right” about picking an MVP based on who has the most HR or RBI or Saves, and sabermetric analysis is certainly far from perfect as well — all you need to do is look at how much the various metrics (WARP vs. WAR, plus/minus vs. UZR) disagree with each other.

But where I disagree with Mark is: I don’t see this as being like the abortion or gay marriage debate at all. In those debates, like in the “dialectic” Mark envisions, there are really only three plausible truths: (a) one side is correct; (b) the other side is correct; or (c) the answer is somewhere in the middle. If you have one side that believes that abortion should be legal in all circumstances and one side that believes it should be banned in all circumstances, that’s as far as it goes; it can’t be more legal than the first side wants it, and it can’t be more illegal than the second side wants it. So the one true “right answer” has to be either one of those extremes or something between them.

Not so here. Our advanced metrics are flawed, but the answer isn’t some compromise between them and the traditional stats; the answer is more research, and more metrics. The metrics we have have grown out of the more traditional statistics. Saying you prefer HR and RBI to VORP and WAR isn’t at all like saying you prefer “Choice” to “Life” or vice-versa; rather, it’s like saying you prefer Betamax to Blue-Ray.

Here’s how Mark defends the traditional crowd:

Those who follow counting numbers have a point (among many). Baseball revolves around the run. It determines who wins and who loses. Therefore, should you not pay attention more to runs, RBI’s, and home runs? Home runs automatically score a run (making them slightly important) and bring in whoever is on base (making them more important). If the point of the game is to score runs than the other team, home runs and RBI’s are awfully darn important, which gave Howard the edge [over Pujols for 2008 NL MVP].

But this ignores the critical weakness of run and RBI totals (and this isn’t a criticism of Mark, who I know understands this: it’s just that I don’t think there’s any way for anyone to successfully defend this position), which is that, in every instance in which you don’t hit a home run, your runs and RBI are totally dependent upon your teammates either getting on base for you or driving you in.

This doesn’t work well for the NL race, because Howard actually did do a phenomenal job of knocking runners in in 2008 (Pujols was still the clear MVP for other reasons), but take a look at this list (I hope). In 2008, Justin Morneau finished 2nd in the AL MVP voting, while his teammate Joe Mauer finished a distant 4th, based largely (or rather, entirely) on the fact that Morneau had 129 RBI and Mauer managed just 85. If that link went to the right place, though, you’ll see that when they batted with runners on base, Mauer and Morneau drove in those runners at almost exactly the same percentage: 19.0% to 18.6%. Morneau gets that huge edge in RBI because he batted with 151 more runners on base than did Mauer. Morneau actually batted with the most runners on base of anyone in the league. Part of that, of course, is because he’s not a catcher, and thus got to play every day. But a huge part of that is that he got to hit behind Joe Mauer, and his 2nd-in-the-AL OBP!

So the RBI stat tells you who was at the plate for the final event resulting in the creation of a run, but it can actually distort your sense of how that run was created. Mauer was, hands down, a better hitter than Morneau in ’08, and played a much bigger part in how the Twins’ runs were scored. When you add in defense and adjust for position scarcity, it’s not even close. They’re very nice complementary pieces, but Morneau is the Scottie Pippen to Mauer’s MJ.

So, yeah, runs are awfully important. On the team level, you could almost say they’re all-important (almost). But to look at the HR, runs or RBI a single player has as a way of judging that player’s value is never a good idea. Even with Howard: make him the MVP because he drove a bunch of guys in, and you’re ignoring Pujols’ 100+ points of OBP and 100+ points of SLG, amounting to 100+ fewer outs and many more runs for Pujols’ team, and Pujols’ vastly superior defense, all for the sake of (a) Howard’s good fortune of having 50 more runners on base during his PAs than Pujols had in his and (b) a 2% edge in his success at driving those runners in. It doesn’t add up, or even come close.

More to the point, every one of those traditional stats is totally encapsulated in some more advanced metric or other. Whatever skills you think RBI measures, that’s also measured, and better, in SLG; or, if you think hitting with runners on base or “in the clutch” is a skill that’s worth measuring, stats like WPA/LI do a better job with that. Batting average is a fun little stat for what it is, but OBP tells you the same thing and more. Fielding Percentage is totally encapsulated by all advanced fielding metrics, like UZR and Plus/Minus.

You might think that these things (well, save OBP) are less-than-perfectly accurate, but that’s not an argument in favor of going back to the old things; it’s an argument in favor of doing more research and finding better new things. UZR may not be perfectly accurate, but it’s always, in every possible instance, going to do a better job of telling you who is the better fielder than fielding percentage will. FIP may not be perfect, but it’s better than just comparing two players’ ERAs. There may be slightly different ways to measure OPS+, but it’s always going to be better than not adjusting for era or ballpark factors at all. And so on. We can argue about how good the new stuff really is, but it’s just plain better than the old stuff (the well-grounded stuff that gains some level of acceptance, that is, not just any old thing someone thinks up).

So that’s the point: I’m not going to use the term “flat-earthers” around here. I try to avoid mudslinging of all types. I have nothing against people who rely solely on traditional stats, and I think those stats have their place. But their place isn’t in player analysis, not anymore. If you’re going to argue something like that Howard was the 2008 NL MVP and base it on traditional stats, you’re going to be wrong — simply, objectively, obviously wrong. And I’m sincerely sorry to say that. But I’m not trading in my DVD player for a VCR, and I’m not giving up my numbers for a set that tells me the same stuff, but less of it, and with more static.

Advertisements

Errors in Judgment

May 16, 2009

As Rob Neyer brought to your attention yesterday, a five-member panel appointed by MLB granted the Royals’ appeal, overturning the Angels’ official scorer’s decision granting Howie Kendrick an inside-the-park home run on what was really a routine popup down the right field line that was completely misplayed by Jose Guillen.

I happened to be watching that game live, and, I mean, this was a terrible decision by the scorer. No two ways about it. You can click on the second link above and watch the video clip for yourself. If that’s a home run, we should go back to ’86 and give Mookie an RBI single.

So, I’m glad they overturned it. And Rob’s post and some of the comments beneath it raise some good questions about the quality and motivations of (team-employed) official scorers.

But I think this touches on an even bigger issue. The only reason this was any kind of controversy is that the ball just barely grazed Guillen’s glove on the way by. Consider this same result in a couple alternate universes:

  1. Guillen gets a great jump on the ball and camps under it, but he pulls his eye off it too early and it pops right out of his waiting glove, and then he kind of head-butts it all the way to the fence, resulting in Guillen being featured prominently in blooper reels for the rest of the year.
  2. Guillen takes his eye off of it on his way over, so he takes a slightly wrong angle whereby he comes too far in on the ball, and then watches helplessly as it bounces six feet beyond his reach.
Is there any question in the world that (1) is scored an error, (2) a home run? Yet, isn’t Guillen exactly equally culpable in both scenarios? And in the third scenario, the one that happened back in reality? In all three cases, he should’ve made the play, but didn’t. Why (at least for purposes of fielding and pitching analysis) treat the three cases any differently?

If you can watch the play and read the accompanying story and not come to the conclusion that “errors” and “fielding percentage” are utterly useless as tools for measuring defense, I’d really love to hear your argument in their favor. (Well, read the rest of this, then let me have it in the comments.)

Properly evaluating defense, at its core, requires you to ask one question, and it has nothing to do with whether or not the guy got a glove on the ball. Whether the fielder caught the ball, or dropped the ball, or ended up thirty feet away from the ball, the question should be exactly the same: should we have expected a dude in that position to make the play that that dude just made (or didn’t make)?

The Twins provide another convenient vehicle for making this point. Most days, as I’ve discussed here before, they start one of the worst left fielders in baseball (Delmon Young, or occasionally Jason Kubel); on the other days, they start one of the best (Denard Span sliding over from CF when Carlos Gomez plays). Now, Young and Span may end up with essentially the same number of “errors” over the course of the season, but if you watch them every day, you’ll routinely see Young come up ten feet or more short on fly balls hit at the exact same angle and speed as balls that Span catches with no difficulty. And when Span does make an “error,” odds are it’ll be on a ball just like that: one that Delmon could have been expected to play into a double. See, this works both ways. If Span’s legs and instincts get him to a ball that only one or two other guys in baseball could’ve hoped to, it doesn’t make a ton of sense to punish him if he bobbles it.

Turns out, most MLB clubs already have, internally, done away with fielding percentage and errors. Most teams (not the Twins, clearly; get Go-Go back in the damn game already!) employ some kind of sophisticated system of defensive analysis using tools — like my oft-cited favorite, UZR — that really do nothing but attempt to answer that one simple question (albeit in a slightly more sophisticated way than the way I just posed it).

But how long do you think it’ll take before this straight-forward, common-sense, weirdly counterintuitive idea takes hold among the media and public at large?

Or, to pose the same question in a different way: how many times must the author hear Joe Q. Colorcommentator cite errors made or fielding percentage as evidence that a team is first or last or sixth in “team defense” before he experiences some sort of cataclysmic psychic event?