Zack Greinke and Pitch Sequences

Pitching is a game of strategy. At any given time, a pitcher must consider the batter, the ballpark, the count, the strength of his pitches, his stamina and more. And there’s not always a correct answer — there might be multiple pitch combinations, locations and speeds, all with the same probability of success.

Even when a pitcher has one plus-plus, unhittable, grade 80 pitch, they still need to know how and when to use it. One such pitcher is Zack Greinke. The owner of a nasty slider, the Brewers’ ace racks up strikeouts and swings-and-misses, thanks primarily to his slider. The pitch is dynamic: it doesn’t appear to suffer from platoon issues nearly as much as other sliders, which allows him to throw it both to left-handed and to right-handed batters. Indeed, the whiff rate (whiff/pitches) against his slider is actually higher against left-handed batters than against right-handed batters, albeit by a very marginal amount.

The pitch is hard to hit, and Greinke knows it. When he’s not in two-strike counts, Greinke’s pitch selection looks like this:

CH   CU   FF   FT   SL

0.10 0.22 0.44 0.17 0.07

CH = changeup, CU = curveball, FF = four-seam, FT = two-seam, SL = slider

We see lots of fastballs and lots of curveballs, but very few sliders. It’s almost as if he is saving the pitch. But in two-strike counts, we see a different pitcher:

CH   CU   FF   FT   SL

0.04 0.08 0.35 0.08 0.45

It’s almost a flip of a coin. Either you face Greinke’s devastating slider, or you see one of his not-so-easy “easier” pitches — the changeup, the curveball, the four-seam or the two-seam.

Combining the two sets of information, here’s the pitch-usage difference between two-strike counts and non-two counts:

 

CH = changeup, CU = curveball, FF = four-seam, FT = two-seam, SL = slider

*These classifications are not from Gameday – they’re the result of the output from clustering analysis and manual reclassification.

And the strategy, although simplistic, is very effective. Greinke obviously has a plan in mind – use the fastball and the curveball to get ahead, then use the slider to finish off the batters.

What is he doing to set up his slider?

Here’s his distribution of pitches* thrown against righties, prior to a slider in a 0-2, 1-2 or 2-2 count:

CH   CU   FF   FT   SL

0.00 0.07 0.33 0.09 0.50

Here we see that, against righties, he seems throw slider after slider until he records a strikeout. Intuitively, this strategy seems to make sense. His slider is so good that he might as well keep throwing it until the batter gets himself out — a result Greinke frequently accomplishes.

If we break down the success of his slider based on the previous pitch thrown — as measured by whiff rate (whiff/pitches) — we find that his sliders in 0-2,1-2 and 2-2 against right-handed batters are most effective when preceded by a four-seam fastball (29% whiff rate) or a slider (24% whiff rate).

 *The one previous pitch, not all pitches from the at-bat. 

Here’s the distribution of pitches thrown against lefties, prior to a slider in a 0-2, 1-2 or 2-2 count:

CH   CU   FF   FT   SL

0.38 0.08 0.26 0.05 0.23

Against lefties, we see that he often sets up his slider with a changeup. Interestingly, Greinke often uses his change against lefties in 0-1, 1-0, and 1-1 counts — but not very often in two-strike counts. This seems to suggest that he doesn’t see his changeup as a strikeout pitch, but more as a tool to advance an at-bat.

If we look at the whiff rates for his slider based on the previous pitch thrown, we see that his sliders in favorable pitcher counts against left-handed batters are most effective when preceded by a curveball, which has an astonishing 50% whiff rate. When preceded by a four-seamer or a slider, the whiff rate on his slider is about 23%. And when the previous pitch is a changeup, his whiff rate on sliders is 22%.

Much like his sliders against right-handed batters, the pitch doesn’t lose effectiveness when he throws it consecutively. While the 50% whiff rate on his sliders thrown after a curveball is very impressive, it’s also from a sample of just 10 curveball-slider sequences, so it doesn’t seem noteworthy.

 

References and Resources

*PITCHf/x data from MLBAM via Darrel Zimmerman’s pbp2 database and scripts by Joseph Adler/Mike Fast/Darrel Zimmerman