San Francisco Giants v Philadelphia Phillies

Last week, we talked about hitters that had problems with particular pitches. The quintessential Pedro Cerrano / Jobu / Curveball situation. “He can’t hit that.”

Then SABR threw a party called the Saber Seminar in Boston and all sorts of nerdery broke loose. One presentation was by SABR president Vince Gennaro. He put forth the idea that lineups could be better constructed if we knew how players did against similar pitchers. By using pitcher similarity scores, we could group certain pitchers by things like pitch repertoire, fastball velocity, release point, swinging strike percentage and zone percentage.

To some extent, this is already in practice today. Dan Straily admitted to me in an interview that his hitting coach finds him similar pitchers so that he can see how hitters did against other starters with his sort of arsenal. (He wouldn’t tell me which pitchers those were.) Just this week, Justin Masterson admitted to me that he watches A.J. Burnett starts sometimes to see how hitters fare against someone with his sort of (three-pitch) arsenal, and how Burnett attacks those hitters back. So the pitchers know about these groupings, at least.

Gennaro’s point was that we can use this idea to help managers better set their lineups. The flip side of the same coin, perhaps.

And though Gennaro’s numbers are way more advanced than what I can replicate for this week’s article, it did jog my memory back to last week, when we were talking about using pitch type values to find players with bad results against a certain pitch.

The argument against pitch type values, to recap, is that you don’t know what the pitch *before* the pitch in question looked like. Chris Archer‘s changeup is not very good, but his fastball is sexcellent and his slider breaks bats. His changeup probably gets a few people looking at strike three, or out in front on a ground ball even if it isn’t very good. That might be why Archer’s changeup had a (small sample) positive pitch type value last year. And it’s also why pitch type values rate R.A. Dickey‘s 81 mph fastball as his best pitch — and, in some years, one of the best fastballs in the game.

But zoom out, and say “how does this guy do against all changeups,” and you might still get some interesting — and believable — values. We learned, for instance, that Pedro Alvarez is not that bad against changeups (he hits them far when he hits them), but that maybe Gerardo Parra is still not great against changeups.

How bad does that make Gerardo Parra against a pitcher with a great changeup? Even the heaviest changeup users throw the pitch less than 40% of the time, giving a player like Parra plenty of chances to swing at something that isn’t a changeup.

So: take the top 20 pitchers by changeup pitch type values. These are your Good Changeup Pitchers, for better or for worse. It’s okay, Cole Hamels is at the top of the list with Felix Hernandez, Hyun-Jin Ryu and David Price, the list passes the sniff test. Now take the 20 worst hitters by changeup pitch type values.

How do those hitters fare against those pitchers?

This is a surprisingly hard query to run, so I had to get a little help from my friend Jeff Zimmerman. And then I had to run some pivot tables, and then I had to actually comb through 2000+ lines of data one last time. And the results are so tiny:

Group AVG OBP SLG
Bad Changeup Hitters, vs All Pitchers, This Year .257 .324 .410
Bad Changeup Hitters vs Good Changeup Pitchers, This Year .248 .311 .398

I’ll hold for the applause.

Considering that the league is hitting .254/.318/.398, this isn’t a terrible finding. Across more than 3400 plate appearances this year, a good changeup pitcher can turn a bad changeup hitter who is otherwise better than league average into a worse-than-league-average hitter. It’s how Alex Gordon — normally a league average whiffer — has struck out 31 times this year against Justin Verlander and Chris Sale combined (and five times against Ubaldo Jimenez). It’s the kind of thing that’s worth pursuing in order to get that extra two percent that wins you games.

And it’s also a minor score for pitch type values. We used something that was pitch-specific, but by zooming back out to all at-bats between these two groups, we avoided some of the pitfalls of sequencing. Many of the outs in the batting lines above were on fastballs, and many of the hits were on changeups.

But that doesn’t matter. A bad changeup hitter against a pitcher with a good changeup takes a serious ding. As every old-school scout could have probably told us without the hours of SQL work and the eye-crossing excel sheets.