Sunday, May 19, 2013

A Not-Quite-Half-Hearted Defense of Wins as a Statistic



Despite what you may have heard, THE BEST BLOG IN AMERICA IS NOT DEAD. We've just been busy, okay guys? Can't you appreciate that? Can't you understand that? No? Well screw you, it's not like you're paying us for this. Sheeeeeeet.

The writers on this blog are all proponents of sabermetrics, or advanced baseball statistics. We are proponents of sabermetrics because sabermetrics are like...right about stuff. They measure good things. They tell you that traditional stats like batting average and RBI and wins for pitchers are fairly stupid and not good indicators of a player's overall performance. It makes total sense. But I just wanted to share a couple of little thoughts about the value that wins can have in determining a pitcher's value.



Basically, the reason why wins generally suck is because they don't measure anything useful. It's kinda like measuring the ability of a pitcher to keep a lead when his team gets one. It's a "right place at the right time" stat, kind of like RBI. You can go 5 innings and give up 8 earned runs and get a win, or you can pitch a no-hitter and take a loss. It doesn't measure anything real. But it is, in many ways, a halfway (or maybe quarterway) decent indicator of success.

This thought occurred to me when I was looking at Red Sox pitcher Felix Doubront's stats this year. As of right now, Doubront has a 6.03 ERA, a 1.82 WHIP, and an ERA+ of 73. All of these are pretty terrible. But trust me when I tell you that Doubront has not been as bad as those numbers indicate. Here are a couple of ways to show this.

The first is his advanced pitching stats. His FIP and xFIP are 3.32 and 3.79, which are actually pretty good. FIP and xFIP should at least come close to resembling your ERA, and in Doubront's case, they don't. His BABIP is also .400. This partially means that, in the grand scheme of things, part of Doubront's struggles really aren't his fault, and he's kind of been unlucky. So there's that.

But here's another stat we haven't talked about. Doubront has a record of 3-1. In 6 starts, Doubront has gotten the win in half. So even though most of his raw stats (ERA, WHIP, ERA+) are bad, he's been the winning pitcher in half his games. Surely, this is an example of wins being a dumb stat, right? In a way, yes. But there's another way in which that record is more of an indicator of how well he's pitched this year than ERA and other things. We can demonstrate this by looking at each of Doubront's six starts.

Doubront has had five starts in which he's pitched between 5 and 6.2 innings and given up 3 runs or fewer. His record in these starts is 3-0. He's also had a start in which he pitched 3.2 innings and gave up 6 runs. That was his loss. The other start in question was a 5.1-inning performance in which Doubront gave up 6 runs, taking a no-decision.

So basically, Doubront has had about 5 pretty good games and 2 pretty bad games. His record is 3-1. That kind of makes total sense. The 3-1 record makes more sense than the 6.03 ERA when you look at each start individually.

Now, I know that wins still don't MEASURE anything. Like, Doubront could have gotten the win in his 5.1-inning, 6-run start, and taken the loss in his 6.2-inning, 3-run start. I absolutely get that, and because of that reality, I really take wins with a grain of salt. Trust me. I do. BUT, where they lack in MEASURING something, they make up for a bit in INDICATING something that ERA and FIP and WHIP can't. They take you game-by-game. Stats like ERA and FIP and WHIP give you a snapshot of the season on a large scale, looking at the pitcher outside the context of his team. This is generally a GOOD thing in determining an individual player's performance. But they don't tell you what wins can, and are supposed to, indicate. They indicate the game-by-game success and failure of a pitcher, which is really really valuable to know. A pitcher's ERA might be 8.00, but he may have just had a start where he gave up 12 runs in 1/3 of an inning, and the rest of the year he went 7 innings and gave up 2 runs. Y'know what I'm getting at here? Y'see what I'm saying?

So are wins still a bad statistic? YES, because they don't measure anything real, and they don't even necessarily indicate what I just talked about. But they're usually okay at indicating the thing they're supposed to indicate (2012 Cliff Lee notwithstanding). So if you see a guy like Doubront, who is 3-1 with a 6.03 ERA, don't scoff at the win total. Take a closer look, and you might find that win total to be a bit more accurate than you might imagine.

Let the hate mail from fellow disciples of sabermetrics commence.

1 comment:

  1. And Doubront's start last night is a perfect example of why wins should be taken with the tiniest grain of salt imaginable.

    ReplyDelete