Here it goes, my first project blog post.
One of the main focal points of pitching today is spin, the revolutions per minute (rpm) of the baseball. Pitchers with high spin on their fastball range from Justin Verlander and Gerrit Cole to Luke Bard and Lucas Sims. So what separates the league-leading, 2746 average rpm of Luke Bard's heater to Gerrit Cole's seemingly impossible to hit fastball? While there are quite a few factors in play, such as Cole's 3 mph average velocity advantage, active spin plays a huge role in how these two fastballs play completely differently.
Active Spin, also known as spin efficiency, is how much the spin contributes to the movement of the baseball. For independent projects, it was a highly unused metric before this year. With the breakthroughs of Driveline Edge and Baseball Savant, active spin statistics have become more popular and more accessible to view data for specific players.
What exactly will more active spin on a fastball do? The simple answer is that it will have more vertical movement and thus more rise. In this tweet from Driveline Baseball, their visualization demonstrates how a fastball with 100% spin efficiency will move compared to a fastball with the same tilt, spin, and velocity but with 40% spin efficiency. Furthermore, I compared every MLB pitcher's average active spin on their fastball with their average rise. Below is a ggplot generated through R Studio. With an R value of nearly 0.53, there is a moderately-positive correlation between active spin and rise on a fastball.
With the understanding that more active spin tends to indicate more rise on a fastball, we can dig further into how it affects a hitter's launch angle. This is where we see the polarizing difference between high and low active spin fastballs. Before depicting a large sample size of pitchers in another ggplot, let's take a closer look at some very specific examples.
After scraping data from Baseball Savant and running a SQL query, I wanted to view a snippet of information between pitchers who are in the top percentile of fastball spin high. Below is the terminal output of a CSV file. Between these four pitchers, the average launch angle given up on fastballs is higher with more active spin. There tends to be a lot of variance between the expected stats in relation to fastball active spin. I also find it interesting that the lowest Put Away % is the pitcher with the most spin and active spin—Luke Bard. The other variable that most likely plays a major factor with Put Away % and Whiff Rate is velocity. The leader in Whiff % out of these five pitchers, Jose LeClerc, also has the highest average fastball velocity. While these players have the highest fastball spin rates, it is clear that there is more than meets the eye.
Moving onward to reviewing the sample size of the population, there are a few interesting points. Firstly, it's worthy to note that when reviewing league-wide data, the R value (0.37) indicated a moderately-weak correlation. I think one possible statistical error that was made when charting this data is the sample size of number of pitches thrown by a particular pitcher. For example, Trevor Bauer threw 1400+ four seam fastballs in 2019. It is difficult to compare that large of a sample size against a pitcher such as Jared Hughes who only threw 65 four seam fastball last year. Another issue with running regressions on launch angle in a large dataset is the limitations. The average launch angle on a pitcher is going to mostly be within 0-30 with some outliers. However, I will say I am not surprised to only see a moderately-weak correlation of 0.37 due to the variance in number of pitches thrown and the limitations on average launch angle. Below is the ggplot comparing fastball active spin and launch angle. This data was collected through a web scraper built in Python and then exported into SQL.
I find it really intriguing to see which pitchers have improved in certain aspects over the course of one year through their metrics. Whether it was pitch design bullpens or just a change in feel, there are several noticeable improvements when comparing 2018 to 2019 data. The first of which is Dillon Peters, a University of Texas product (perhaps in-coincidentally...) and Angels left-handed pitcher. In 2018, Peters averaged 61.4% spin efficiency on his fastball, which placed him at the bottom. Dillon would come into the UT Baseball facility during the offseason to throw bullpens. Could spin efficiency have been something he was looking to improve? It sure looks like it as his average jumped 17.7% over the course of that one offseason to 79.1% in 2019. Moreover, there is a clear change in vertical movement on Peters' fastball from 2018 to 2019 as he decreased the amount of drop. Another note to consider is his move from the Marlins to the Angels. Different teams have different philosophies and encourage a variety of pitching styles.
The most interesting active spin change comes when looking at Lucas Sims, a right-handed pitcher for the Cincinnati Reds. In 2018, Sims averaged around 70% spin efficiency on his fastball, which put in close to the bottom. His fastball spin sat in the upper 2300s — fairly high among MLB pitchers that year. Then something clicks. In just a year, Lucas Sims' fastball spin efficiency shoots up 15% to an average of 84.8%. He starts throwing more fastballs up in the zone, which is a result higher active spin. Sims was in the top percentile for fastball spin in 2019 averaging well over 2600 rpm. While Kyle Boddy and the rest of Driveline hadn't yet arrived in Cincinnati, it's clear that they have been actively making developments on their pitching side.
When it comes to fastball active spin, the final aspect I want to analyze is whiff rate. I think this is a crucial question to look at as the game's hitters have moved towards the three true outcomes of home runs, strikeouts, and walks. After using the same Python scraper to obtain data from Baseball Savant, I then created another ggplot in R. I also decided to limit the sample size to pitchers who had thrown more than 200 four seam fastballs in 2019. Surprisingly, fastball active spin has extremely weak correlation with whiff rate on a fastball. Because of this, I went back to see if fastball velocity has a stronger correlation. While it isn't necessarily strong or even moderately correlated, we do see more of a relationship between velocity and whiff rate than active spin and whiff rate. The two ggplots are below.
In this article, I discussed what active spin is and how it affects the movement on the baseball. After graphing the relationship between fastball active spin and fastball rise, we learned that there is a moderate-positive correlation between the two variables. I then dove into some specific examples on relating active spin and launch angle. Discussing these pitchers gave us a better idea of the specific outcomes that more active spin could lead to. However, after plotting a population sample size, there is only a moderately-weak correlation between fastball active spin and average launch angle given up against fastballs. I then looked at changes in active spin over the past two years. The two pitchers highlighted, Dillon Peters and Lucas Sims, had some really interesting developments to their fastballs as their active spin increased. Finally, I was intrigued to discuss how active spin might affect whiff rate. I found that it almost has no relationship. After going back into the data, I learned that comparing active spin to average fastball velocity produces a higher R-squared value, although by no means is this a strong correlation.
To wrap up, all of my code for the web scraper, sql queries, and ggplots will be posted to my Github page. I am always open to suggestions on how to improve my code, so please don't be afraid to reach out on Twitter or LinkedIn with any advice. As this is my first project post, let me know how you liked it and what I could do better. My idea here was to discuss how active spin can affect different aspects in baseball, whether it be results-based or pitcher-controlled.
Be on the lookout next week for my second project post. The topic for next week will pertain to defensive metrics such as OAA, reaction, and UZR. I believe defense is far behind analytically than it should be. We will also cover shifts and maybe some creative ideas for the future of shifting. I will post any updates for this post on Twitter @igabriel826.
Comments