16 Days Till Opening Day: Back to BABIP

Way back at the start of this project I wrote a bit of a primer to some advanced pitching statistics, centered around BABIP and the discovery that pitchers have much, much less control over what happens to a ball in play than previously thought. After reading that the scales surely fell from your eyes, and like Saul, you were baptized into the Church of Sabermetrics. But just in case you need a little more convincing, let’s talk a bit more about why ERA is, if not useless, deeply flawed.

Things get instructive at the extremes. Imagine a pitcher starts the game by walking the first three batters of the first inning, then the fourth batter hits into a triple play. The pitcher then does the same for the next five innings (assuming the manager allows this), finishing the game with 6 innings pitched, 18 walks, no strikeouts, six triple-plays induced, and no runs allowed. His ERA is 0.00, and he’s credited with a win.

No one would say this pitcher did a good job, even if the outcome of his efforts was positive for his team. It’s simple to imagine the opposite extreme, where a dominant pitcher is blooped and bled to death, ultimately giving up a ton of runs despite not pitching poorly. It happens all the time. Most fans will allow for the vagaries of luck over the course of a game, maybe even two. But here’s the thing: these streaks of good luck and bad luck can extend for entire seasons, or even longer. Think Ryan Vogelsong in 2011 with good luck, or James Shields in 2010 with bad luck. There were major corrections for both pitchers the following years, as the luck evened out.

The way to more accurately evaluate a player’s talent is to strip luck out of the equation as best you can. Here’s what pitchers can mostly control: strikeouts, walks, and home runs. Good pitchers do at least two of those three things well; great pitchers do all three. There’s an easy ERA analogue that more accurately represents a pitcher’s talent level, and that’s FIP. It stands for Fielding Independent Pitching and it takes into account strikeouts, walks, and home runs, then scales the number to resemble ERA, so a good FIP is 3.00, a bad one is 5.00, and we can all relate to it easily.

Fangraphs.com also shows xFIP, which is a park and league-adjusted version of FIP, but which also corrects for fluctuations in HR/FB rate. HR/FB is as it says, the rate at which the flyballs a pitcher surrenders turn into home runs. Pitchers can experience good or bad luck with that, and unless a pitcher (like Kershaw mentioned in the prior article) has demonstrated a long track record of suppressing home runs, xFIP assumes a pitcher’s HR/FB rate will regress to the league average, which is a little over 10%.

We can all recall times when a pitcher gives up a loud out to the warning track, or a long fly ball that just barely scrapes over the left field fence. Those different outcomes don’t demonstrate a difference in pitcher skill; Ian Kinsler’s miracle double in Game 2 of the 2010 World Series didn’t stay in the park because Matt Cain was six inches a better pitcher. I rarely use xFIP, because there really are pitchers who have true talent HR/FB rates significantly above or below the league average and I don’t think it’s accurately representing them. Also, it’s more of a “what should’ve happened” stat than a “what did happen” one, and I prefer the latter.

Still, if you sort 2014’s pitchers by xFIP, you really do get the best pitchers who had the best seasons at the top. It works.

Evaluating pitching is really tricky, and the best method is to be holistic. Look at strikeouts and walks, look at groundball rate, look at strand rate, home run rate, ballparks, left/right splits, PitchF/X data like swinging strike rate and velocity, etc. It’s much easier to evaluate offense, which I will do in a future post.

Comments Off on 16 Days Till Opening Day: Back to BABIP

Comments are closed.