Painting with numbers to create better pitchers.
If you clicked on this link expecting the bangin’ beats of experimental electro band, Art vs Science, then what follows in the post ahead is going to be a profound disappointment to you and you should probably go back to your search. If you’re a baseball fan however, this should be something fun and interesting and I would recommend you read on, for what appears below contains plenty of gifs, graphs and other goodies.
Traditionally, data and baseball have combined as a tool strictly to valuate and evaluate – to determine which players were good, which were not and how much each player was worth to a given team. More recently however, the tools and concepts designed and employed by baseball analysts have leaked over into the player development side and is now becoming an increasingly important part of maximizing player’s abilities. The early application of data naturally puts players, coaches and evaluators on edge as it threatens many of their ability to make a living. Data as a tool in player development however, should be something to be welcomed by the game as it presents a new frontier of growth for the aforementioned groups. What I hope to do here is take some of these tools, some of which have been rolling out throughout Baseball Prospectus’ annual Pitching Week and apply them in a way that is accessible for developing pitchers, especially amateurs.
Stay With Us A While:
- Sonny Gray’s Mid-Career Crisis
- Rebuilding the Baltimore Orioles
- How to Subtlely Become an Ace: Ross Stripling
- The 7th Visit Ep 5 | Special Guest
- Re-Imagining Pitch Classification
What appears below will likely seem fairly obvious to some, perhaps many of you and that’s fine. Firstly, it probably means you have had quality tuition throughout you career and secondly, in many cases, data in baseball is supposed to help us further understand or visualize information we already know. To present this information, I will be testing out some popular beliefs about pitching.
Belief: High Fastballs are the best pitch to ‘set up’ a breaking ball.
Short Response: Kind of, Not really
This adage taps into what is commonly referred to as pitch tunneling and pitch sequencing. Simply, Pitch Tunneling refers to the degree to which a pitcher is able to make two pitchers appear similar to a hitter and have them finish in different locations, making them harder for a hitter to handle. Pitch Sequencing generally refers to the way in which a pitcher combines his various pitch types. When someone talks about ‘setting up’ a pitch, it broadly means using a particular pitch to add greater deception to the pitch that will follow. Now of course there are two general types of breaking balls, curve balls and sliders and when answering this question it’s important to distinguish between the two … now to the fun stuff.
Let’s start with the curve balls though the answer here is a little blurry. In one sense, a high fastball is indeed the best pitch to ‘tunnel’ with a curveball as it most closely replicates the initial flight path of the curve. Visually, it looks like this:
The image above shows the flight of a fastball (black) and a curveball (yellow) thrown by Stephen Strasburg. As you can see the two pitches appear to come somewhat out of the same tunnel. In theory, this makes the curve ball more difficult for a hitter to recognize and therefore put a good swing on. Initial research by Josh Kalk at the Hardball Times suggests that this is true as Curveballs that proceeded a high fastball were significantly better than the average curveball. However, more recent work by Eno Sarris found some evidence that may be to the contrary:
|Situation||Called-Strike Rate||Swinging-Strike Rate|
|High Spin CB after High Spin FA||10.0%||12.5%|
|Low Spin CB after High Spin FA||12.7%||13.3%|
|High Spin CB after Low Spin FA||11.3%||16.1%|
|Low Spin CB after Low Spin FA||12.8%||14.0%|
The table above shows that curveballs thrown after a low-spin fastball generally induce the best results. While low-spin and high-spin fastballs don’t correlate perfectly with low in the zone and high in the zone, they should be fairly close if a pitcher is using them correctly. From this, we can tentatively say that a fastball lower in the zone leads to more swinging strikes for a curveball that follows. The theory behind this (more work needs to be done) that when a fastball is thrown lower in the zone, a curveball off the same plane will finish below the zone, rather than in the zone leading to more swings and misses. Visually, it looks something like this:
While that fastball isn’t quite low in the zone, it’s lower than the previous example and as a result, the CB, off a similar plane to the FB finishes below the strike zone. So what we know is this: when it comes to curve balls, a high fastball in the pitch prior generally leads to better outcomes than the average curve ball. However, when it comes to swinging strikes, it may be better for the fastball to be located somewhat lower in the zone.
However, all this says nothing of the relationship between high-fastballs and sliders. Perhaps that’s appropriate however, as in reality, there is none. Again, let’s start by attacking this visually:
Here is another chart from Strasburg, this one showing a fastball (black) followed by a slider (red). As you can see, at no point in their respective trajectories do these pitches bare any semblance to each other. From. the moment these pitches leave Strasburg’s hand, they are easily differentiated by a hitter. Of course these are both great pitches in isolation and are therefore capable of producing great results on their own but it’s tough to argue that the high fastball has any positive effect on the proceeding slider. Instead, a better sequence for a slider would involve a fastball lower in the zone:
The image above shows both a fastball and a slider thrown to former NL MVP Buster Posey. You can see that the slider stays within the band of the fastball for a reasonable amount of time. This is a great example of tunneling and it’s what makes Kershaw’s slider one of the best in the game. To further illustrate this, let’s take a look at a well tunneled FB-SL sequence from Carlos Martinez:
And from Michael Fulmer, cos why not….
— Rob Friedman (@PitchingNinja) August 1, 2017
Neither of these sliders are set up by high fastballs and that’s what makes them so nasty.
In sum, it is clear that a high fastball is not the best way to set up a breaking ball. When throwing a curveball, using a high fastball as the set up pitch may lead to generally positive outcomes however, pairing it with a low fastball may actually lead to more swings and misses. When throwing a slider, there is no relationship at all.
Belief: Pitch Tunneling only works in relation to the pitch that came before it.
Short Response: Wrong
To answer this question it is necessary to think about the effect of tunneling from the hitter’s perspective. As we know, it is effective because it makes one pitch look like one that is completely different. What is more important here however, is how this works. Let’s start with what a hitter (doesn’t) sees:
“When a pitch travels too fast for the eye to track, the eye skips ahead in the projected flight path, often to the expected point of contact. This means that not only is anything that happens in the last 150 ms too late for a batter to adjust their swing in any way, it’s likely that they aren’t even actively looking at the ball beyond that point.”
The thing that I want to extrapolate here is that a hitter determines the timing and the location of his swing based on an expectation of where the ball will be. As a result, rather than simply just the visual information gathered during the flight of a pitch, a hitter is using a number of cues including a pitcher’s tendencies, previous pitches, changes in delivery, release point, or pre-pitch routine, to develop an expectation of the pitch type and its final location. Tunneling then, might be better thought of as the ability of a pitcher to make a pitch look different than a hitters expectation.
As noted above, one cue which helps to form a hitter’s expectations are the previous pitches faced. While this most prominently includes the pitch immediately prior it can also encompass a pitch from earlier in the at-bat or one from a different at-bat entirely. The image below shows pitches 1, 4 and 5 between of a sequence from Chris Archer. As you can see, there is no tunnel relationship at all with pitches 4 and 5. Instead, it is the first pitch of the at-bat that creates the deception on the slider later in the at-bat:
Pitch tunneling most prominently involves the pitch that occurs immediately prior and is perhaps the most important cue in developing a hitter’s expectation. Nonetheless, a number of factors contribute to this expectation and this should be accounted for when trying to apply pitch tunneling in a game setting.
Belief: Throwing Harder makes pitchers inaccurate
Short Response: Wrong, Completely Wrong
This is one you will hear throughout all levels of baseball from your local little league field to an MLB clubhouse – that pitchers who throw hard or pitchers that try to throw hard have a hard time throwing strikes. It is often presented as a trade-off, either throw hard or throw strikes and nothing has ever been more false in the game of baseball. Let me prove it to you:
Here I have plotted each pitchers strike zone% against their average fastball velocity for all pitchers who threw more than 30 innings in 2017. The R² value for this plot is 0.0103 – statistical speak for ‘there is absolutely no relationship between these two things’. Of course there is something of a selection bias here as to have reached the major leagues you have to be able to throw some minimum rate of strikes but nonetheless for the above claim to be true there would be some significant relationship shown here.
In fact the idea that hard throwers have a tough time throwing strikes doesn’t even hold any logic. Those pitchers who are able to exhibit elite velocity tend to be, by definition, superior athletes than their soft-tossing counterparts and therefore should be better able to throw strikes on a consistent basis. But what about the claim that it’s the pitchers that try to throw hard that have trouble throwing strikes?
Again, there is no evidence to support this. I’ll defer to a 2011 article at Baseball Prospectus:
“So in the above chart, we see that from the fastest 25% fastballs to the next fastest 25%, we see no change in command for the 1 mph loss in velocity. We see almost the exact same thing over the next fastest pitches, no change in command for about another decrease of 1 mph. And then for the slowest pitches, pitchers actually lose almost a third of an inch of command for about a 1.5 mph decrease in velocity.”
The math here is fairly involved so I won’t include it here – if you’re interested just click the link above. Instead, I’m happy to simply say that the study found no evidence that trying to throw harder has any effect on a pitcher’s command. While the final line in that quote suggests that trying to throw softer actually decreases command, the author admits this is likely due to fatigue later in games, so rather than arguing for a negative relationship between effort and command, I’m happy to simply say there is no meaningful relationship between the two.
If you want to pitch in professional or even collegiate baseball you’re going to need both velocity AND command – to players and coaches, quit being lazy and train both. There is no evidence to suggest that there is any kind of trade off between the two. To conclude this point let me draw from the guys at Driveline Baseball:
Do the work – get yourself in the blue.
Belief: Strikeouts are bad/selfish/inefficient, you should pitch to contact/groundballs.
Response: Yeeeeaaah, that’s a no.
This is another one we’ve all heard before and before we get to debunking it I will allow that chasing strikeouts at the expense of deep counts and walks by consistently working out of the zone is not a great thing. However, if this is the problem then it’s the pitcher’s approach to getting strikeouts that is at fault rather than the strikeout itself. Let me first address the belief that strikeouts are inefficient:
I limited this chart to only include starters as efficiency really isn’t a thing worth considering for relievers throwing one inning at a time. If you squint hard enough you an probably see a slight upward trend here but it’s negligible and when I show you why the strikeout is the best mode of out for a pitcher the extra quarter of a pitch per batter faced is not going to seem to notable at all. This is especially true when we view this within the context of the trend towards ‘super bullpens’ and shorter starts for starting pitchers, being able to work deep into games really isn’t the virtue it once was.
Now onto why strikeouts are intrinsically good. Let’s start with a quick table:
Ground balls turn into outs about 75%, fly balls about 80% and strikeouts? Those are always outs. Literally always. You could include the rare dropped 3rd strike if you want but whatever. So given that it is probably of no surprise that pitchers who generate strikeouts tend to be better than those who don’t:
The plots above, in order, show the relationship between ground balls, strike outs and ground balls + strikeouts with ERA. I have listed them in order from the weakest relationship to the strongest.
The results are fairly clear, if you can generate both high rates of ground balls AND strikeouts then that’s good, do that. Ground balls however, a fairly useless without the presence of strikeouts. If you want to pitch in high level baseball, you need to strike people out.
The proper use of data is one of the biggest weapons coaches and players have available to them in player development. Rather being met with resistance from the baseball community, data should be embraced players and coaches. Some of the data we have confirms things we already know to be true. In other cases, like some of those above, we can use it to debunk inefficient and incorrect beliefs prevalent in the game. Learning to use it to maximize your performance might just be the difference between a career in baseball or an early exit from a great game.
If there are any Art Vs Science fans who made it this far anyway, here’s a little gift: