"You're insane." That's generally the response I get when I present the information you're about to read. I've been accused of being the "epitome of 'pseudo-stat fan' gibberish." I've even been accused of being Aaron Sele writing under a pseudonym... My belief? Well, simply that hits allowed are not a particularly meaningful statistic in the evaluation of pitchers.
-- Voros McCracken
The above opened an article written for Baseball Prospectus in 2001 by McCracken, which has to count as one of the most groundbreaking in the history of sabermetrics. Most numbers the field produces don't tend to challenge deeply-held beliefs: WAR tells us, for example, that Gerardo Parra and Paul Goldschmidt have been pretty good players for the Diamondbacks this year: while it's nice to be able to put a number to that, this hardly counts as some kind of shocking revelation. However, the conclusion McCracken reached was something which, at first glance, seems entirely counter-intuitive.
The issue into which he was looking, was how truly to evaluate a pitcher's performance - and a pitcher's performance alone. The problem with ERA, WHIP, etc. is that the defense playing behind him plays a significant part in this: if you have eight Didi Gregorius clones, you're likely going to concede fewer runs than if you have eight Jason Kubels out there - not just through errors (which are taken into account in ERA), but simply because the latter won't get to as many balls in play. The resulting higher ERA is not the pitcher's fault.
So what McCracken did was to concentrate on the aspects of the game where the defense was out of the equation, where it's just the pitcher vs. batter. Those are strikeouts, walks and home-runs (essentially - park factors obviously play into those, to some extent). But it was when McCracken looked at everything else - what happens when the defense got involved - that, as he put it, "The trouble really started. I swear to you that I did everything within my power to come to a different conclusion than the one I did. I ran every test, checked every stat, divided this by that and multiplied one thing by another. Whatever I did, it kept leading back to the same conclusion:"
There is little if any difference among major-league pitchers in their ability to prevent hits on balls hit in the field of play.
The best way to measure that is through BABIP., batting average on balls in play, which is the rate at which balls in play i.e. excluding home-runs, walks and strikeouts, become hits. The formula for this is:
What McCracken found, is there seemed little skill involved in a pitcher's BABIP, which goes against 'common sense'. You'd think that the Justin Verlanders of this world would be harder to put "good wood" on than, say, Eric Stults, and so would have a much lower BABIP. That's not the case: over his career, Stults is at .291 while Verlander is all the way down at...er, .288. That difference works out at maybe two fewer hits per season for Verlander. It turns out that what makes him great are the strikeouts, lack of walks and fewer home-runs. If you can put the ball in play, the results are almost identical for both men. Verlander just has fewer balls in play, which leads to fewer hits.
McCracken instead found that there was almost no correlation between a pitcher's BABIP one season, and the figure the next: "The pitchers who are the best at preventing hits on balls in play one year are often the worst at it the next." That's strong evidence there's little or no measurable skill involved, because if there were, you'd tend to see the same pitchers doing well from year to year. Instead, as time goes on, everyone tends to gravitate towards somewhere about .295. If you look at the 104 pitchers with 600+ IP since the start of 2008, about 80% had BABIPs which were within just 15 points of that, over the whole period.
BABIP is a little like gravity: you can defy it for a while, but it's always going to start pulling you back to ground eventually. If you look at last year's qualifying pitchers, the range was broader, as you'd expect from a smaller sample size: only about 55% had a BABIP within that 15-point span. The rest varied from .241 by Jared Weaver, to .345 from Rick Porcello - unsurprisingly, Weaver was third in the Cy Young, while Porcello's ERA was 6.28. But even there, both pitchers had numbers more than 30 points outside their career BABIP figure, so you can make the case that both their seasons were, to a degree, propelled by luck of one kind or the other.
I should mention, there is evidence that some pitchers do tend to have lower BABIP. High strikeout pitchers form one such group: Clayton Kershaw, for example, has a lifetime BABIP of .276. Also, extreme groundball pitchers tend to have a lower BABIP on those groundballs, leading to a lower overall figure. But the effects are generally relatively small and far from guaranteed: Brandon Webb, for example, had a career BABIP of .291. Also, hitters have more input into their BABIP: one obvious factor, can they leg out infield squibs, and turn those from outs into hits? See Ichiro in his prime: from 2001-10, he had a .357 BABIP.
But again, this isn't to claim that all pitchers are somehow "equal". Just that the differences among them are better seen in things like strikeouts, walk-rate and home-run rate, rather than what happens to balls in play, where things like defense and, yes, pure luck, increasingly muddy the water. Looking at a pitcher's season total, comparing it to his career numbers, can give you clues as to their future results. You've all seen games where a pitcher will get every ball hit right at the defense, or conversely, the same guy will have a night where they keep finding gaps and dropping in. Same pitcher, same skill-set: but very different results.
That's what you're looking at and measuring with BABIP, and it can vary quite dramatically, especially in small samples. We saw this last night against Stults, where he initially seemed to have turned into Tom Glavine: The first two times through the order, we had 18 PA's, with one walk and two strikeouts, so there were 15 balls put in play. Only one of those was a hit, a BABIP of .067. The rest of the way? 14 balls in play, seven hits, and I think we all know the game went quite a bit better for the Diamondbacks. Overall on the night, our BABIP was .276, not far off what you'd expect.
What does this mean for the 2013 Diamondbacks' pitchers? Here are some numbers to look at:
The last two columns are a player's career BABIP, and an arrow (shamelessly copy-pasted from 'charmer's Trends!) which indicates whether I'd expect their performance to improve, stay bout the same, or get worse going forward, based entirely on how this season's BABIP compares to their career one: if they are outside the 15-point spread mentioned earlier, they get an up or down arrow, depending on the direction. As you can see, the team total BABIP is virtually bang on the league average, but there are wide variations among the individual players.
Bell, in particular, seems to have been BABIP'd to death thus far, and I'm optimistic that .407 figure won't be sustained for the long-term. But the poster-child is undoubtedly McCarthy. His April ERA of 7.48 was connected to his April BABIP of .404. This month, his ERA has collapsed to 1.59, and that has certainly been helped by a May BABIP of .259. It's also interesting to note that Reynolds' low ERA is not the result of a flukey BABIP. Key factors in his success are more that he hasn't allowed a home-run yet, and has stranded 85% of base-runners allowed, compared to a league-average of 73%. [Both of those may regress, but that's a subject for another day...]
At the bottom end, we see Cahill, Corbin and Ziegler, all sporting 2013 BABIPs significantly lower than their career numbers [albeit not excessively so in the first-named's case]. However, Ziegler's overall figure may be skewed by the fact that he has only recently become the extreme ground-ball pitcher we know and love. Through 2011, his GB/FB ratio was 1.65; since then, it has ballooned to 3.21, and he is likely an example of the lower BABIP enjoyed by that kind of pitcher, as discussed above. His 2012 BABIP was .267, so this year's figure is much closer to that than his career number.
With Corbin, you are still dealing with a relatively-small sample-size, less than 170 career innings, so it's hard to say where his "true" BABIP will end up. Like Reynolds, he has also benefited from a very low-rate of home-runs, allowing just two in 62.1 innings, and an even more excessive strand-rate (88.2%). So, yes, some regression is certainly to be expected. To which revelation, the correct response would be, "Well, duh" - unless you expect him to have a 1.44 ERA for the rest of his career. Nice though that would be.... We'll revisit the numbers above around the two-thirds point (if I remember!), and see if my predictions have been accurate.