Thursday, December 12, 2013

Preaching to the Choir: A Handy Guide to Why Advanced Stats Aren't Stupid

I read an article recently that redirected me to the blog of Murray Chass, the former New York Times baseball writer who is famously against the phenomenon known as "the passage of time." The fact that he has a blog is kind of amazing. He's in the "computer geeks in their mothers' basements" crowd, the sort of person who reads Parade Magazine every week. This also puts him squarely on the side of people who HATE advanced stats. He's an anti-sabermetrics extremist, the kind who would characterize sabermetrics as "well he hits fly balls 20% of the time on Tuesday and 30% of the time on Wednesday, and goobledee gobbledee I don't know what they're tryin' to say!" You know the type.

The article that I read was on Murray's fervent support of Jack Morris as a Hall of Famer. While it would be great to go through and do a FJM-style takedown of this piece, I want to address some larger points that anti-"advanced stats" people have, and perhaps clear up some misconceptions. Maybe we can ease a little hostility between the Murray Chassosaurs of the world and the charts 'n' graphs HOTSHOTS.

Firstly, let's address the anti-advanced stat people's concern about the lessening prevalence of wins as a stat for pitchers. Here's a quote from Murray Chass's piece on Jack Morris's Hall of Fame candidacy:

The stats geeks judge him on statistics that didn’t exist when he pitched. In another example, the new-age guys discount Morris’ 254 wins, saying wins are the least meaningful statistic for pitchers.

Yes, fellow dinosaurs, that’s what they say these days, even though when Morris pitched, wins were still meaningful

Speaking of his time and the matter of wins, Morris said, “You gotta ask, what is important to get to the World Series?” The answer was wins.

There is a fundamental misunderstanding of advanced statistics' discounting of wins. Assigning a pitcher a "win," and then saying that is the most important part in evaluating a pitcher's performance, is misguided. A win does not measure anything. It's an indicator of success, and it's usually not a bad one, but it doesn't measure anything. A pitcher gets a win because his team had the lead when he left the game and never relinquished that lead. A starter could go 5 innings and give up 4 runs and get a win, or he could go 8 shutout innings and not get a win. Who pitched better? Does wins tell you, or do other statistics?

I think where the Chasses of the world get hung up is that WINNING is the whole point of the game. And that's absolutely true. But just because you're calling the stat "wins" doesn't mean they're the same as actual wins. Like, wins in the standings. Those two things are different. You can discount pitcher "wins" without discounting actual "wins." And this leads to a larger point about sabermetrics: they are all (more or less) designed to measure how likely a player is to help his team win actual games. They're not just numbers that are used to measure arbitrary things. These statistics are all meant to correlate to winning, and how each player helped his team win.

There were stats that were taken seriously before the advent of advanced stats (baseball statistics didn't appear out of thin air in 1985). Batting average is one. Home runs is another. Hits, ERA, wins, RBI, stolen bases, and many others were very much a part of evaluating talent in the pre-sabermetrics era. You can make charts and graphs, assumptions, decisions, and evaluations based on these statistics. Sabermetrics points out the deficiencies in these statistics, and looks for ways to evaluate performance even more accurately. The Murray Chasses of the world have a few problems with this: they've never heard of these stats, and they're sometimes a step more complicated than simple algebra.

Murray Chass refers to "meaningless initials" in his post, and he includes WAR (Wins Above Replacement), WHIP (Walks + Hits per Innings Pitched) and WAA (Wins Above Average). By calling them "meaningless initials," what is Chass trying to accomplish? I assume he thinks they're meaningless because he doesn't know what they mean. But they mean something, and often, they mean something very simple.

Let's look at WHIP, for example. WHIP is about as simple as it gets: how many walks and hits did this guy give up per inning pitched? There's really nothing advanced about it. It's very straightforward. It's as straightforward as batting average. It's actually one fewer algebraic step than ERA! And it measures something useful. Seeing how many baserunners a pitcher gave up (seeing as how it is the pitcher's job not to allow runs, and runs come from baserunners) is useful in seeing how well he pitched. This is a stat that could have existed and been important in 1950, but it just wasn't. That's why Murray Chass doesn't like it.

WAR and WAA are more complicated, and are "made up" in certain ways. They use formulas that are still algebraic, but perhaps with more steps than, say, WHIP or batting average. They're more holistic, and they try to boil down a player's performance into one nice number.

Murry Chass and co. could argue that you can't do this, and shouldn't try. While there's some merit to that, and while WAR is far from a perfect statistic, it's certainly worth trying. It's also interesting, if you're a baseball fan, to think about everything that goes into evaluating a player's performance compared to another player. For instance, the era in which a player played, the ballparks in which they played, the quality of their teammates, and all other factors that go into a player's performance are part of what makes WAR so interesting and so complicated. But trying to figure this out is probably better than just ignoring all of these factors, and basing player evaluation on good ol' statistics that happened to exist during a period in which you were an active writer and fan.

Another reason Chassites claim not to like these stats is that it takes the "human element" out of the game. For instance, this is a quote from Chass's pro-Jack Morris article, in which he's responding to an anti-Morris emailer:

“Did you ever see him pitch, or do you base your view on numbers?” I asked. “A number did not pitch a shutout in Game 7 of the 1991 World Series. Morris did.”

This is a classic pro-Morris argument. Chass's emailer responded that Don Larsen is not in the Hall of Fame, at which point Chass retorts that this game was just one moment in Morris's otherwise fine statistical career (the main statistic he cites is wins). I just thought I'd add that little tidbit because it shows that even Chass doesn't believe in purely anti-stats evaluations of players, and when faced with the logical conclusion of his argument, he reverts back to using stats.

Anyway, Chassites argue that sabermetrics would ignore Morris's game 7 heroics as a statistical anomaly. He's not entirely wrong about this, but he's not entirely right. I think most sabermetrics adherents (at least I can speak for myself) would see Jack Morris's performance in that game as commendable, brilliant, and an example why we're fans of baseball in the first place. Watching Jack Morris pitch that game must have been truly thrilling, and I wish I had been old enough to appreciate it. It is a remarkable and fantastic high point of an otherwise good career. But as even Chass argues, that's not enough on its own to merit Hall of Fame induction. You have to look at other things, like a player's entire body of work, to figure out whether he's a Hall of Famer. You can appreciate both the "human element" of baseball and the value of sabermetrics. Those two aren't mutually exclusive. You can play fantasy sports and still enjoy watching football on Sundays. You don't have to be a nerd in your mother's basement to find advanced stats interesting. In fact, for me, they enhance my experience as a fan because they're interesting to think about. They supplement rather than replace the "human element" of being a fan.

Sabermetrics exist to try to evaluate players in an objective way, and all of them are geared toward figuring out what helps teams win ballgames. If you're really into baseball, like Murray Chass is, they can be really interesting. If you're not really into baseball, they can certainly be boring and irrelevant to you. But in no way are they "bad." You can appreciate sabermetrics while also appreciating the game of baseball. I do it all the time. J.D. Drew may have double the career WAR of Trot Nixon, but I'm always gonna love Trot Nixon more. That doesn't mean I can't acknowledge that J.D. Drew was a better player.

So the real reason Murray Chass and co. are against sabermetrics? They like things they can remember. If it wasn't around in 1960, it's not good. New things threaten the Murray Chasses of the world because they were so invested in old things. So instead of seeing the value in new things, or challenging their merits in a thoughtful way, they reject them out of hand because they're new, and they'll find any excuse to disregard them. For an old guy, that's pretty immature.

1 comment:

  1. about time you posted something. what am i supposed to do during the day if you aren't writing things?