Can Catchers Control the Running Game?

It might sound like a silly question at first, but when you think about it intuitively, it makes sense. Leg times--the time elapsed between a pitcher lifting their leg during a windup and the ball crossing home plate--vary from pitcher to pitcher. They also vary from pitch to pitch on each individual pitcher. Sure, JT Realmuto (above) is a great catcher, but no matter how strong his arm is or how quick his pop time is, if his pitcher has a 1.5 second leg time, he's got a much lower chance to throw a runner out. And I haven't even mentioned variations in runner speed, leads, or jumps. The pitcher and catcher can do everything perfectly, but maybe the player stealing is too fast. Maybe the shortstop drops the ball. Maybe the throw was off line, but the runner over-slides the bag.

To me, there's too many moving parts in a steal attempt to be completely confident in catcher-only metrics like pop time. But don't just take my word for it. Something incredibly telling is that when looking at a total of 37 backstops who played in both 2021 and 2022, the r-squared value between 2021 caught stealing rate and that of 2022 is a measly 0.03. This tells us that in these two years, it was seemingly impossible to judge how many runners a catcher would throw out in 2022 just by looking at their caught stealing rate from the previous year. Clearly, a ton of variation exists in caught stealing rates between these two seasons. Thus, the stats may back up the fact that catchers have minimal control over the runners they throw out.

My plot (below) contains catcher data from Baseball Savant from the 2021 and 2022 seasons, and compares the pop time for each catcher measured against the rate at which that catcher threw out runners.

The blue line attempts to model a linear regression between the two stats, but there's really little point of doing so, with the data being spread significantly above and below the regression line. Now, it isn't completely a wash. The r-squared value informs us that a catcher's caught stealing rate between 2021 and 22 is 23% explained by pop time alone. Meanwhile, when running a multivariate regression using a combination of pop time, exchange time, and arm strength, all within the same time frame, the r-squared value only increases to 0.24, leaving much to be desired.

I added one more variable to that same regression model--a runner's lead. This was measured in how many feet away from second base the runner was, before the runner took off. This increased the model to a 0.44 r-squared value. Additionally, as seen in the P column (below), the variance in leads are more statistically significant than pop time, and much more significant than exchange time and arm strength.

This led me to think that maybe it's the pitcher, not the catcher, who has more control over caught stealing rates. The pitcher, by being quick to the plate and attempting pickoffs, has a much greater ability of decreasing a runner's lead. I gathered pitcher data between 2021 and 2022, to conduct the same correlational test I conducted initially on year to year caught stealing rates for catchers. Over these two seasons, I found that among the 153 pitchers who allowed at least 1 successful and 1 unsuccessful stolen base attempt in each year, 40% of these pitchers' caught stealing rates in 2022 can be explained by their rates from 2021. A staggering increase in predictiveness compared to catchers. I then filtered the data down to pitchers with at least 3 successful and 3 unsuccessful attempts in each of the two seasons, increasing the r-squared value from 0.4 to 0.66. 66% of caught stealing rates for a 2022 pitcher can be explained by that same rate from the previous year!

Now, granted, it's hard to draw meaningful conclusions from these tests. Unlike the 153 pitcher sample size in the first pitcher dataset (or the 37 catcher sample size from their year to year data) the filtered-down data contained only 14 pitchers. Additionally, every dataset I studied for these tests only include the 2021 and 2022 seasons. Not to mention the new rules in 2023 (such as larger bases and limits on pickoff attempts) that have thus far increased stolen base attempt rate and success rate across the league. These data and statistical tests can inch me closer to my hypothesis of a catcher having little control over the runners they throw out. But they cannot prove anything. This is a good and thought provoking start, but as more data become available and more statistical testing knowledge is acquired on my part, I will continue to revisit this discussion and inch closer to the answer.