The ground that sabermetrics broke wasn’t about newfangled statistics for matters nobody previously thought important. Instead, it was in providing evidence and language to things always thought significant, but without a way to show or express why. Take, for example, pitch framing. Catchers have been doing it for decades, but only recently have baseball watchers and analysts managed to capture its effects. Framing metrics didn’t invent framing (though they might have consigned the advantages of framing into irrelevance). A couple new statistics found at Baseball Prospectus continue the tradition of giving shape to what everyone knows is there; they help pin down each pitcher’s ability to control and command his pitches.
Two strike-oriented metrics do this: Called Strike Probability (CS Prob) and Called Strikes Above Average (CSAA). CS Prob deals mostly in control—the ability to throw strikes—as it estimates the likelihood for each of given pitcher’s pitches to be called a strike. Rather than working from the rulebook definition of the strike zone, CS Prob “reflects what [the strike zone] is: a set of probabilities that depends on batter and pitcher handedness, pitch location, pitch type, and count.” In other words, A pitcher with a high CS Prob has good control because the ball is usually somewhere in the de facto strike zone. Notably, that in itself is not a good thing. Batting practice pitchers probably have phenomenal CS Prob ratings.
The aim of CSAA is to capture command—locating pitches without making them hittable—by identifying “the additional called strikes created by the pitcher.” The premise is that staying away from the heart of the plate and grabbing extra strikes from the edges of the strike zone demonstrates command of pitches. This should sound like pitch framing, because CSAA was initially developed to measure just that. As opposed to CS Prob, which accounts for all pitches, CSAA only considers pitches taken, when the umpire has to level a judgment.
What do these statistics say about the four Rockies’ starters—Jon Gray, Tyler Anderson, Chad Bettis, and Tyler Chatwood—who logged more than 100 innings in 2016 and, mother earth willing, will do so again in 2017? Great question! It’s almost as if you’re here to read about the Rockies. Let’s find out.
Jon Gray
Gray’s CS Prob in 2016 was 48.2 percent and his CSAA was -1.62 percent. For context, the season’s high man in CS Prob was Bartolo Colón at 52.2 percent, and Wily Peralta was the low one with 41.5 percent CS Prob. There was about ten percentage points of difference between the high and low, and Gray finds himself more toward the high. Note that I’m not referring to these figures as “best” and “worst.” It’s not entirely clear what the marks mean.
Gray’s CSAA, however, is in the negative, which is something to note. Not only did Gray not add a lot of strikes around the edges of the strike zone, but his CSAA was the fourth lowest mark in baseball in 2016 (Jimmy Nelson’s -2.79% was the lowest). But again, “low” does not mean “bad” here. Gray’s not the type of pitcher who needs to locate his pitches at the edges, and that’s on account of his ability to get batters to swing at his slider, which isn’t designed to nibble. His stuff is good enough that he doesn’t need to command the edges of the zone and gain extra strikes there. That doesn’t necessarily mean he can’t. It just means that’s not the type of pitcher he is right now.
It’s also useful to see who Gray was like in 2016—who else had a pretty high CS Prob and a truly low CSAA? Gray’s closest comparison from 2016 is White Sox lefty Carlos Rodon. Rodon’s CS Prob was 48.3 percent and his CSAA -1.55 percent. Gray and Rodon also threw almost the exact same number of pitches and otherwise had similar seasons. They’re also two young starters with excellent pedigrees. For Gray and for Rodon, high control and negligent command around the edges works.
Tyler Anderson
Left-handed rookie Anderson posted the highest CS Prob of the four pitchers examined here, as well as the highest CSAA. The high CS Prob was 48.5 percent, a touch higher than Gray’s, but his CSAA was 1.29 percent. This is a profile of someone who exhibits control as well as command. Anderson doesn’t have the powerful arsenal Gray does, so he needs those extra strikes at the edges of the zone, and he was able to get them in 2016.
Of the 144 pitchers who threw at least 100 innings in 2016, Anderson’s CSAA is sixteenth highest in baseball. Because of the type of pitcher Anderson is, we can conclude that he demonstrated “good” command and that it benefitted him. Notably, however, it’s hard to find another pitcher with this combination of high control and high command. Among the 15 pitchers with a higher CSAA than Anderson, only Aaron Nola (1.8 percent CSAA) had a CS Prob within one percentage point of Anderson (48.1 percent to Anderson’s 48.5).
It could be telling that two young pitchers, Nola and Anderson, are somewhat distinct with their high CSAA coupled with a high CS Prob, whereas most of the other pitchers with a high CSAA have a much lower CS Prob are veterans, such as Zack Greinke, Jon Lester, and Justin Verlander. Twenty-four-year-old Zach Davies is the exception that at least supports the rule, as he led baseball with a 3.5 percent CSAA and had an extremely low 42.8 percent CS Prob.
In other words, when looking for Anderson’s maturation and improvement as a pitcher, look for him to throw fewer probable strikes and to earn more strikes from edges.
Chad Bettis
Bettis’s combination of control and command strikes me as a quality one that demonstrates a mature approach to getting batters out. Bettis’s CS Prob was 44.3 percent in 2016 and his CSAA 1.14 percent. His CS Prob ranked 121 out of 144 pitchers with at least 100 innings pitched, but his CSAA ranked 26th. This profile—CS Prob closer to the bottom and CSAA closer to the top—represents what Anderson might advance toward, though Bettis probably has less potential overall than Anderson does.
It’s tough to identify a comparison for Bettis. For CS Prob, he sits just below Wei-yin Chen and just above Madison Bumgarner; for CSAA, he’s just below Michael Pineda and above Patrick Corbin. The closest overall, however, is probably Ubaldo Jiménez, who had a 1.2 percent CSAA and a 45 percent CS Prob in 2016.
Tyler Chatwood
Chatwood is different from all of the others because he showed low control, 43.3 percent of his pitches were probable strikes, without a lot of command in the gray parts of the zone, evident in his 0.58 percent CSAA. The closest comparisons to Chatwood are not terribly enticing either: James Shields posted a similar 42.3 CS Prob with a 0.58 CSAA, and Yovani Gallardo had a 43.8 percent CS Prob and a 0.69 percent CSAA. Both Shields and Gallardo also posted below replacement level seasons in 2016.
But Chatwood didn’t post a below replacement level season. In fact, he put up 2.2 WARP, according to Baseball Prospectus. Chatwood has long been a tough nut to crack because he’s had some success while walking a lot of batters, giving up a ton of contact, and not striking very many guys out. He should turn into a pumpkin, but he hasn’t. CS Prob and CSAA may not get us closer to understanding Chatwood because the answer will probably be found in pitch movement instead of strike probability. These statistics won’t help, but Baseball Prospectus’s other new analytical toys based in tunneling just might (but that’s a topic for another article).
What now
We undeniably know more now about these four pitchers than we did before, even if, following sabermetric tradition, it’s primarily a matter of hanging concrete evidence on the peg of presumptive knowledge. Gray doesn’t piddle with the edges of the strike zone because he doesn’t really need to, but Anderson does. Bettis has learned a formula that works well for him. Chatwood is still mysterious.
The next step is for enterprising minds to use these figures and identify metrics that capture command, control, or both. As the four pitchers analyzed here who only have a team in common demonstrate, it’s not a matter of looking at the top and bottom of a ranking. A single metric would have to account for potential value pulled from high marks as well as low ones (and probably in between). For now, however, it’s enough to keep an eye on probable strikes and commanding the strike zone’s de facto borders, even if what we’re looking for differs for each of these four players.