Catching Runs Saved/Lost Comparison

So a few months ago, I finished a project I'd been working on during the summer that had been an idea for a couple years. I didn't have the resources to start until this summer. The project (which I'll leave a link right here: bit.ly/2Li1UL2) had to do with weighted pitching values by type. I was able to find the weighted values of certain pitch types by pitcher, which would be great to find if a certain pitch is helping or hurting a pitcher. But, I had an idea recently.

Idea and Concepts

The idea was to use the weighted pitch values to help find how many runs a catcher was giving up or gaining on pitches outside and inside the actual strike zone due to framing or other issues. I say actual strike zone because there are two strike zones, as you can see in the graph to the left. The first one is a two foot by one foot box that extends a half inches off each side of the plate and starts about one foot seven inches off the ground. This is represented as the box zone that you see. The more oval-shaped zones are based on the actual calls of umpires during the 2019 season. Using the loess function and R, I was able to create a model to predict the probability of a pitch being called a strike. Now, this is just for CALLED strikes, not swinging strikes. This is a lot like what I did in the last post I did, about the two Gold Glove catchers, Salvador Perez and Yadi Molina (https://bit.ly/341IWi3).

Looking at the graph, we can see that a 50-50 oval shows that a strike would be called on a pitch about a foot from the middle of the plate on either side, about three inches of off each side of the plate. This is the catcher's view of the pitch, so we can see that the ovals extend more on the outside for a right-handed hitter. The 75-25 oval are about an inch off each side of the plate and is dead-center of the perceived zone.

Data Collection

For this project I selected two 1st-Team catchers from two different conferences, and went through collected Trackman data of their games for the 2019 season. This including going through each game I had to make sure that the catcher wasn't subbed out or subbed into the game. For Catcher 1, he had 14 whole games and three partial games for the 2019 season on Trackman. For Catcher 2, he had 25 whole games and three partial games. After I found this, I created a CSV file for each catcher. Next, in R, I made databases out of each catcher containing only pitches where the outcome was decided by a umpire (ball called and strike called). Finally, I used the loess model I talked about earlier to find the strike probability of each individual pitch for each catcher. Using this, I could sort out pitches with the strike probability that I wanted to look at for each catcher.

Analysis

Catcher Totals

Using Python, I was able to create a program that contained a function that could calculate the runs gained/lost by each catcher based on pitch type. For runs gained, the pitch had to have a strike probability of less than 50% and be called a strike. For it to be a loss, it would have to have a strike probability of greater than or equal to 75% and have been called a ball. I made it higher for losses because it's only a loss if it should have been called a strike and wasn't.

Looking at Catcher 1's numbers, we can we that he gained runs for his team for each pitch type. The part I'm most surprised with is that he managed not to lose any runs on curveballs. He saved a run and a third on fastball and a third of a run on change-ups. Overall, he was able to saved 2.23 runs by his defensive catching alone. If I were a coach of Catcher 1, I'd suggest that he spend more time working on framing curveballs. If he were to continue to only gain on curveballs, that would be huge for the team, as it would allow him to gain more runs for his team, and it would also allow his pitchers to throw more of what they wanted, knowing that Catcher 1's ability will allow him to succeed.

Catcher 2 had games in the database, which explains his larger gained numbers. However, he also has larger loss numbers to, so everything evens out. Catcher 2 however, saved more runs on fastball than Catcher 1 saved all together. Comparing, we can see that Catcher 2 edges Catcher 1 on fastball and curveballs, but doesn't beat him on change-ups and sliders. Catcher 2's slider numbers are alarming, and if I was coaching him I'd mostly focus on this. Just like with Catcher 1, if Catcher 2 could gain more runs on sliders, this would greatly help his team and his pitchers win more games.

Conclusion and Final Thoughts

Here are some of my final thoughts and conclusions about this project:

If I had to pick a catcher based on defensive ability, I would probably take Catcher 1
- His ability to frame change-ups and sliders surpasses Catcher 2
- Catcher 1 didn't lose a large amount of runs on curveballs
I could see defensive runs saved helping coaches fine tune a catcher's framing ability by quantifying his ability, therefore allowing both coaches and catchers to understand what they mean
If I had to do this project more in-depth, I'd split the database into left-handed and right-handed hitters, then make models to predict the strikes zones so i had even more accuracy in my probabilities

So down below, if you guys have questions you'd like to ask me or something you'd like me to look into let me know! I'd love to start doing more baseball research and posting about it! Thanks again for reading!

Google Sites

Report abuse