The title of this piece is one of my favorite Yogi Berra quotes ever. It is so direct, makes sense in some ways, but then again not really. Lately, I’ve been trying my best to master R. The process has included learning how to scrape data, manipulate data, and create ways to visualize data. I decided that I wanted to start publishing some of the things I have been doing in R using a Shiny App, and I just created my first one this week. It uses the pitch-by-pitch data for the 2020 season (I was planning on using multiple seasons, but that was too much data for Shiny to publish) and is focused on individual pitchers. The rest of this post will be about how to use the app, what everything means, and some ways to make use of it. One of the biggest parts of analyzing pitching is looking at the best pitchers, learning what they do, how that makes them successful, and comparing them to pitchers that aren’t as good. That is really what this app is designed to do, help users learn as much as possible about whatever pitcher they are interested in.
The first thing you will see is a dropdown box at the top of the screen (Gerrit Cole is the default selection). Using that you can select any player that pitched in 2020, which includes position players. Now that I have satisfied the Erik Kratz Knuckleball fans, I’ll explain a little bit about what you can do with each pitcher.
You’ll see there are some categories with boxes you can select, which are all filters. The filters are then applied to the first three visualizations, a pitch chart (pitch type and pitch outcome are labeled), a heat map, and a pie chart. The pitch chart gives precise locations of each pitch the pitcher threw during 2020 with colors that tell you pitch type and shapes that tell you pitch outcome. Next, the heat map shows you trends and where the most pitches were thrown. Some pitchers, especially starters have a lot of pitches on the pitch chart and it can be difficult to see which locations they targeted the most, the heat map can solve that problem. Lastly there is basic pie chart, which shows how often each pitch was thrown.
Using the filters, you can answer even more questions. How many balls did Shane Bieber throw that should have been strikes? And what kind of pitches were they? Where did Jacob DeGrom exclusively throw his breaking balls? Were they in similar locations to his offspeed pitches? Are more of Masahiro Tanaka’s fastballs to lefties 4-seamers when compared to righties? These are some of the things that you can begin to answer or understand. Or you can just allow me to answer them for you right now.
Shane Bieber threw around twenty pitches that were called balls but should have been strikes, ten of which were fastballs.
Jacob Degrom really only threw breaking balls low and away from righties, his offspeed pitches were almost the opposite and were thrown almost only low and away from lefties.
When Tanaka threw a fastball to a lefty it was close to 25% more likely to be a 4-Seamer than when he threw fastballs to righties.
On to the next part of the app…HISTOGRAMS! Everybody’s favorite graph is very useful in helping us learn the next, maybe the first, thing about pitchers, what pitches they throw. The remaining graphs try to illustrate and describe this. First and foremost, the following histograms all have a couple things in common. They separate and color by pitch type, and they all have a sliding bar that allows you to change the number of bins. More bins mean more is grouped together and less bins means less pitches will be grouped together. The default is 60, which I think is the best.
First up is effective speed. Effective speed is how fast the ball appears to be going based on how fast the ball comes out of the pitcher’s hand (how we normally evaluate speed), and how close they release the ball to the plate. I chose to use this because how fast the ball appears to the hitter seems more important when looking at how pitches are used. Some guys like Trevor Bauer have many pitches that are around the same speed, while on the other hand all you would need to know what pitch James Karinchak threw is a radar gun.
There are guys who seem like their fastball’s “jump” out of their hand, and this is a way to account for that. One other trait seen in a lot of guys who have fastballs that appear to have “life” without top release velocity is a high spin rate. The next histogram looks right at that and the spin rates tracked by Statcast. One downfall with just looking at spin rate is that you need to understand if the spin of the baseball is efficient. The true spin out of the ball out of its axis is what changes the movement of the pitch, which bring us to out last two histograms, vertical and horizontal movement of each pitch. You don’t need me to explain to you why movement is important, but there are other ways to use it. For example, you can tell by looking at the vertical movement that Yu Darvish almost has two different cutters. He has one set of cutters normally distributed around zero feet of vertical movement, and another set distributed around one foot of vertical movement. You can also tell how much more he uses the first type of cutter.
You could put it all together and find that Devin Williams’ screwball/changeup pitch is very similar to Adam Ottavino’s slider. Both have spin rates that are normally distributed around 2800 revolutions per minute, and movement measures that are very similar. Besides the fact that Ottavino’s slider is three mph slower, the other obvious, and big, difference is that there are no pitchers that throw a similar pitch to Devin Williams, and while Ottavino’s slider does move more than most, it is more easily replicated by other right handed pitchers, which is probably a big reason why Williams had much more success.
So a pitcher throws hard, spins the ball, and gets a ton of movement. How could he not be good? One reason could be because he is tipping his pitches, by altering his release point for different pitch types. That is the purpose of the last two graphs. The first one gives the is a graph of the zoomed-out version and includes the average release point for any righty of lefty. From this you can get an idea of their arm slot and might be able to find some release point differences. The next graph is a close up of the release points and is where you can get a better idea if there are any worrisome trends or not. The best pitchers like Gerrit Cole, make it very hard to find anything. While some pitchers like Adam Kolarek intentionally and effectively change up their arm slot sometimes.
This is the first app that I have ever made, I definitely understand that it isn’t perfect. I am going to continue to keep updating it so that it can look nicer and do more. My goals while doing this project was to further develop my coding skills, and actually help myself and others answer questions and learn more about MLB pitchers, so I hope you are able to use it for your advantage. Hopefully you can observe at least a little bit by watching. If something doesn’t work or you have any suggestions to on how I can improve it, feel free to leave a comment or reach out to @chaseseibold on twitter.