In my last post I walked through an analyst’s perspective of crafting a visual, but never got to talking about how that particular visual could be used to derive insight. So as much as I would love to continue discussing nuances, let’s talk about what these beauties have to offer.
Flashback to over a year ago and I was looking at this graph. Off a quick glance, it looks pretty standard. But for whatever reason, something about the distribution shapes caught my eye. Both the slider and fastball have hints of bimodality – they seem to be two distributions grouped together.
Spotting bimodality can be difficult, especially when savant smooths the graphs as much as they do. They want the curves to look like normal, gaussian distributions. When the graphs stray from that shape, it is an indication of something interesting going on. In this instance, Aroldis’ “Fastball” grouping contains both his four-seam fastball and two-seam fastball in the distribution.
His two-seamer averages two extra ticks relative to his four-seamer. The graph above – less smoothed than savant’s version – shows how the sinker contributes to the weird shape of the combined fastball distribution. It does prompt an interesting question: Why is his sinker thrown (much) harder than his four-seam fastball? Of the two, four-seamers would probably receive a bigger benefit from increased velocity.
Aroldis somewhat exclusively uses his sinker in pitcher-favoring counts. Through the magic of two strike counts he finds a little extra juice resulting in the elevated average velocity. In all seriousness, Chapman does seem to be accessing a second gear. If you scroll back up to my velocity density plot, Aroldis’ four-seam distribution itself hints at bimodality. It could well be the result of randomness, but considering the pattern presented by his sinker’s velocity is worth further investigation.
As I suspected, Aroldis’ four-seam velocity increases significantly in pitcher-favoring counts. Whether to conserve energy or aid command, the count-to-count changes are by design. The speed gap between his fastballs are due to the count they are thrown in. Chapman probably reserves his sinkers for such favorable two strike counts because they are harder to control than his four-seamers.
The Yankees closer is not alone in experiencing such a velocity bump. In the statcast era, the average pitcher has gained just over 0.7 mph on their fastballs when the count shifted to two strikes. Chapman’s 1.8 mph gain is the highest of pitchers who threw 1000+ fastballs since 2015. Bringing that threshold down to 500 fastballs puts him second to Steven Wright. As fun as that is, I feel like that should not count.
As tempting as it is to dive down the rabbit hole on two strike velocity gain, I gotta stay on the tracks here. As demonstrated with this long-winded example, velocity density plots can be leveraged to find arsenal quirks. The chart made Chapman’s abnormality obvious (especially with less conservative smoothing), it was just up to me to figure out when he was shifting gears. After all, if he is really shifting gears, that is important to know as a coach or analyst.
Think of it this way. Unless his adrenaline is bouncing all over the place from pitch-to-pitch, he has the some ability to toggle speeds. For this example, let’s say he has three settings he can pick from: Command fastball (averaging 96 mph), Normal fastball (averaging 98 mph), and Reach-back fastball (averaging 100 mph). Each play slightly differently. The Command fastball may not miss as many bats, but he will be able to stay in the zone with it. On the other side of the coin, the Reach-back will get a few more whiffs, but may be prone to control lapses and accelerated fatigue.
As a veteran, it is not a terrible idea to let him pick his own spots with these fastball settings. Picking based off balls and strikes seems pretty simple, but perhaps there are other factors to be taken into account. Certain hitters can handle heat better than good placement. How much does that affect the optimal balance between Normal and Reach-back fastballs? If the lead is more comfortable, should Command fastballs become the default forcing hitters to earn their way on base? Suddenly the problem is a lot more complicated and might benefit from those darn analytics.
I dove down this rabbit hole in the YouTube video above, but in a way every subdivision is its own pitch. As an old coach of mine once said, “If you have three pitches each thrown at two speeds, you have six pitches.” (*Yu Darvish has entered the chat*) Whether you buy into that logic or not, this is where the value of velocity density plots lays. Along with general relative pitch speeds, they provide a window into these sub-pitch velocity variations. Wrapping your head around these subdivisions is challenging. Figuring out how to use them effectively even more so. For the pitchers who can effectively manage arsenals like this, think how easily they can customize their repertoire to torment anyone in the batters’ box.
With this in mind, I will leave you with Darvish’s 2019 chart to determine how large his arsenal really is.
|Jordan Jones on New Yakkertech Tagging Interfa…|
|Crafting a Gameplan… on The Jordan Hicks Dilemma: A Qu…|
|A Quantitative Evalu… on Avoid the Dead Zone: An Extens…|
|SB on Zach Davies: What Changed?|
|Can Detroit Tigers B… on Improving Pythagorean Winning…|
X Pitch Pitchers – BaseballCloud Blog 08 Mar 2021 at 2:05 pm
[…] even crazier about Darvish is that you could argue he has more than ten pitches. In a recent pitch velocity article I pointed out that some of his pitches have a bimodal distribution, meaning that he almost has two […]