July 1, 2013
Felix Hernandez, Lance Lynn, and a Peripherals Paradox
I remember once that there was an article (not written by me, but it might as well have been; I’ve certainly written a version of this article before) that looked at a batter’s increased walk rate and concluded that it was due to... not swinging at as many pitches outside the strike zone. Colin Wyers tweeted something in response that went something like, but not exactly like, this: “uh no doy.” I try to keep that tweet (or at least something like that tweet) in mind, because it’s easy to find explanations that are already embedded in that which you seek to explain. Baseball generally obeys its own physics. Player is struggling because his heat map looks awful. Fielder’s numbers are down because fielder isn’t making plays in front/in back/whatever of him. Pitcher is walking more batters because pitcher is throwing fewer pitches in the zone.
But what about when that last one isn’t true? There are 179 pitchers who threw at least 1,000 pitches last year, and have thrown at least 500 pitches this year. The correlation between year-to-year changes in zone rate and changes in unintentional-BB rate is fairly modest: about .4. That means there must be a lot to not walking batters other than throwing pitches in the strike zone.
Among the 179, Lance Lynn has had the biggest increase in pitches thrown in the strike zone. (In the strike zone, for this piece, refers to BP’s adjusted strike zone.) Last year, 49 percent of his pitches were in the zone, 124th in the sample of 179; this year, more than 55 percent have been, 10th most. And yet this year Lynn has unintentionally walked 9.2 percent of batters he has faced, compared to 8.2 percent last year.
It’s not that he’s not getting more strikes on all those pitches in the zone; he is, on 65 percent of his pitches compared to 63 percent last year. There’s the dog-bites-man story: more pitches in the zone lead to more strikes. Furthermore, it’s not that batters are taking more pitches, or making less contact, and therefore working deeper into counts. They’ve actually swung more (for obvious reasons) and made contact on more of their swings, both of which should lead to fewer walks. So, the deal?
It seems to be all in the timing. Lynn is throwing more strikes overall, but he’s not throwing more strikes in every situation. Here's his strike rate, broken down by balls in the count:
There’s a lot in this piece that’s obvious, but I’m going to assume a lot of people haven’t thought about it in just this way, and say the obvious stuff anyway. One obvious thing is that the count matters a lot when a pitcher throws a bad pitch. Throw ball one and it leads to a walk 10 percent of the time; ball two leads to a walk 18 percent of the time; ball three, 41 percent; and ball four, well, shoot. That one leads to a walk 100 percent of the time. A ball thrown on 3-1 costs a pitcher about seven-tenths of a walk, compared to the expectation after a strike in the same situation; a ball thrown on an 0-2 count costs a pitcher only about 1/40th of a walk, relative to a strike. So a pitcher who is wild with three balls is going to walk a lot of batters.
Lance Lynn hasn’t had as many three-ball counts this year, but he has been, relatively speaking, wild when he does. About 45 percent of his three-ball counts have led to walks; last year, when he wasn’t nearly the pound-the-zone pitcher he is now, 38 percent of those three-ball counts did. If he had the same walk rate on three balls this year that he did last year, he’d have five or six fewer walks. And, of course, we’d expect him to have a lower walk rate this year, considering that he throws more strikes overall.
Now look at Felix Hernandez. He’s throwing fewer pitches in the zone this year; he’s getting fewer strikes this year; but he’s walking fewer batters. As with Lynn, the obvious explanations come up empty:
But, again, we see the significance in count timing. Here's his strike rate in various counts:
Felix has been wilder overall, but with three balls his control has been incredible. Thus, no walks.
The question is which way this should make us lean on each player. My instinct is to say that the larger sample—all pitches thrown—is more significant than the splits. (Keep in mind that the three-ball splits are considerably smaller than the no-ball and one-ball splits.) My gut is that this falls in the category of various expressions of clutchness: an instance where noisy, unsustainable, disproportionately important splits lead to skewy results. This is where we see baseball disobey its physics, for short periods of time. If that’s true, then you’d expect Lynn’s three-ball strike rates to regress, and for his walk rate to come down, and for his FIP to improve. Buy buy buy! And you’d expect Hernandez’s three-ball strike rates to regress, and his walk rate to go up, and his FIP to get worse. Sell sell sell everything must go!
But pitchers have different repertoires, different strengths and weaknesses, and presumably different abilities to throw strikes (and get strikes) in certain counts. If you told me Felix Hernandez is better than almost any pitcher in the game at throwing strikes on three-ball counts, I might believe you; he’s Felix, after all. If you told me that Lynn, who throws more fastballs than any starter in the game except one, has a hard time getting batters to chase when they’re sitting on his one pitch, I’d believe that. I think there’s something here, but also think I’m a ways away from getting all the way to the bottom of it.