Perhaps the Hitter Knows Best…
Baseball has been classifying pitch types for as long as… well, as long as we’ve had pitch types and we’ve had a fairly uniform method for doing so. While the steady growth of technology in the game has certainly digitised the process, the classification of pitches still relies on the same two inputs – movement and velocity. This of course makes sense. A pitcher throws these different pitches with the express intent of making them move differently and at different velocities. While we may argue over which movements and velocities constitute which pitches, the way we come to define these pitches is fundamentally the same.
A few weeks ago, I wrote that perhaps we should think of Shohei Ohtani’s slider as a curveball. To argue the point I suggested that while Ohtani’s breaking ball might be thrown as a slider, it (so far in his brief career) has been treated like a curveball by hitters. As each pitch type has a distinct movement and velocity profile from a pitchers perspective, the way hitters react to these pitches have differing characteristics too. Today I want to sketch out the argument for including the hitter’s reaction in pitch classification.
If it ain’t broke, don’t fix it
I’m sure you’re probably wondering why this kid is trying to reinvent the wheel here and I should probably try to answer that. My response can be summarised fairly succinctly: Hitters are entirely apathetic to what the pitcher labels his pitches. Alex Cobb might call his change-up ‘The Thing’ but to a hitter it’s just a change-up and he will react accordingly. If the goal is to get the hitter out, then the hitter’s reactions should shape our approach.
The goal of classifying different pitches is two-fold:
- Preventing the catcher from being hit in the face repeatedly
- To allow a pitcher to sequence them effectively
And anyone who has spent time around catchers knows it’s in precisely that order.
If a hitter’s reaction already shapes a pitcher’s approach and we know that pitch classification is how a pitcher executes that approach, then shouldn’t the hitter’s reaction help shape pitch classification?
And let me clarify what I mean when I say a hitter’s reaction. When a pitcher throws a pitch, there are a number of possible outcomes for the hitter: swing, take, ground ball, fly ball and so on. These outcomes all occur at different rates based on the pitch type thrown. We know that sinkers tend to get more ground balls than four-seam fastballs. Same for curve balls and sliders. Sliders generate more swinging strikes than fastballs. Curveballs generate more called strikes than sliders.
But what then if a pitcher’s slider generates ground balls at a rate comparable to the average curveball? What should we label this pitch? Sure, the pitcher is telling us that the particular pitch is a slider but the hitter is telling an entirely different story. To the hitter this pitch is a curveball.
Let’s now turn to some examples.
Trying to re-classify every pitch type thrown for even a day would require a huge amount of work. Further it would require a systematic approach to doing this and that is not my intent here. Instead, I simply want to open the dialogue.
To do this I picked 4 pitches that tend to be problematic to define. Corey Kluber’s ‘Slider’, Zack Greinke’s ‘Change-Up’, Noah Syndergaard’s ‘Slider’ and Lance McCuller’s ‘Curveball’. These are some of the best pitches in baseball and all sit on the fringes of their conventional pitch classification.
What follows below is a series of tables. Each table compares a particular pitch against the league average for similar pitch types, over a range of statistics. I have highlighted the characteristics most similar to the pitch in question. The league averages were taken from this article by Harry Pavlidis (which is admittedly a little out of date) while the pitch type characteristics were taken from Fangraphs,Baseball Prospectus and Baseball Savant.
Let’s start by looking at Corey Kluber’s Slider. This is the pitch that probably defies classification more than any other in baseball and even the pitcher himself tends to avoid it. It is however, most commonly referred to as a slider and compared to other breaking balls it behaves like one against MLB hitters. While the groundball rate on the pitch most closely reflects that of a cutter, the chase rate (swings out of zone), whiff rate and called strike rate all resemble those posted by the average slider.
Let’s move away from breaking balls for a second and take a look at Zack Greinke’s Change-Up. The pitch features much less velocity separation relative to his fastball than the average Change-Up and is therefore one of the more interesting pitches in the league. However, by results, this pitch more closely resembles a splitter. This of course is only a minor classification change but hitters have more consistently chased this pitch out of the zone and hit it into the ground than the average Change-Up. It’s not much of a shift but based on hitters reactions this pitch most closely resembles a splitter.
One more before we look at how we can put these different classifications to use. Syndergaard throws most pitches harder than the average pitcher and his Slider is no exception. Syndergaard averages in excess of 90 mph on his slider and it features less vertical break than most. As such, many people argue that this pitch should be classified as a cutter. Relative to the average cutter however, Syndergaard’s Slider gets a greater number of groundballs, whiffs, and chases. By results then, this pitch is closer to a slider than its younger sibling.
Putting it to use
You’ll notice I saved one particular pitch, Lance McCuller’s Curveball. The reason being, I want to use it to compare against another breaking ball, Johnny Cueto’s Slider, to demonstrate how we might put this information to use.
I’ll let you work through the table yourself – you’ve got the hang of it by now. As you can see, it’s evenly split between the average slider and curveball though I would argue that based on these results, it’s a slider that happens to also get high rates of grounders and called strikes – those chase and whiff rates are just so extreme. I’ll let you make up your own mind. Compare that now to Johnny Cueto’s slider:
As you can see, this pitch was almost a perfect example of a curveball in 2017. Hitters produced high groundball rates against it as well as low whiff rates and plenty of called strikes. How then should we use this information?
The most obvious answer to this question is sequencing. For McCuller’s, his breaking ball is elite by all four metrics we have discussed here. The ability to create groundballs and called strikes makes it an excellent option both early in the count and behind in the count. The pitch can help McCuller’s get quick outs on the ground and get the right-hander back into counts. However, it can also generate high chase and whiff rates making it a great pitch late in the count as well, helping to drive high strike out rates. As a result, McCullers should throw this pitch in any count – something he is famous for doing:
McCuller’s breaking ball is an excellent pitch, one he should throw in any count and clearly he does. This is an example of optimum pitch usage. Lets compare that now to Johnny Cueto’s Slider.
Compared to the average slider, Cueto’s breaking ball elicits a greater number of groundballs and called strikes. As with McCuller’s breaking ball, this makes for a good pitch early in the count. On the other hand, while Cueto’s breaking ball posted a strong chase rate in 2017, the pitch has been underwhelming in terms of generating swings and misses. As such, Cueto should limit his usage of the pitch later in counts.
Unlike, McCuller’s you can see that Cueto’s pitch slider usage hasn’t really been optimised. While you can see a number of early count sliders (good), Cueto is still using the pitch heavily later in counts as well (less good). Using this pitch in two-strike counts unnecessarily invites contact from the hitter, where a strike out is a superior outcome. Where a swing and miss in this count all but guarantees an out, a ball in play can lead to a number of undesirable outcomes.
Johnny Cueto might think of his breaking ball as a slider and that’s fair enough, it’s thrown like one after all, but looking at the pitch characteristics from a hitters point of view tells a very different story. That has tangible consequences on the way a pitch should be used.
By now I’m sure you’ve read 1500 more words on pitch classification than you intended to today so I will wrap it up here. As I said earlier, my goal here wasn’t to present a systematic method for classifying pitches but rather to open up the question of whether we should include hitter tendencies in our measures. I think I also showed, although briefly that there is some real world benefit for doing this.
Of course, I’m not arguing to do away with traditional concepts of pitch classification. I haven’t even done so here. Instead, I hope to raise the question of whether hitter tendencies can help improve our understanding of pitch classification.
Let me know what you think.