Don’t run on Realmuto?

If you walk into the team store at Citizens Bank Park and head to the display of pins, you’ll see a pin with a picture of J.T. Realmuto and a slogan: “Don’t Run on Me”. This is more than a winking reference to Philly’s Revolutionary War history and the Gadsden flag. Since 2016 (the first season for which Statcast records detailed catcher throwing statistics), Realmuto has thrown runners out on 36% of caught stealing attempts, 7^th best over that time period. And while counting statistics have their well-known limitations, it’s worth noting that no catcher has thwarted attempted larcenists over that time period than Realmuto’s 123. Don’t Run on Realmuto is not a taunt. Historically, it’s just good advice.

But you might not know from looking at his stats from this year. In 2023 Realmuto has only thrown out 12 of 55 attempted stealers for a rate of 22%, ranking 24^th among qualified catchers. The season isn’t over, and the numbers may fluctuate. But Realmuto is decidedly average this year in a category where he is typically the cream of the crop.

What’s gone wrong? The environment of 2023 is different from those of past years, namely due to the expanded bases and the rules encouraging more runners to steal; it would be no surprise for a catcher’s caught stealing percentage to drop after rules explicitly designed to improve the success rate of stolen base attempts are implemented. That might explain Realmuto’s disappointing performance in comparison to his past seasons, but it doesn’t explain why Realmuto is looking up at 23 other catchers in terms of caught stealing percentage. His competitors are playing with the same bases and the same emboldened baserunners. Something else must be behind this.

In recent months, Statcast has introduced a statistic they call Caught Stealing Above Average. Statcast data (the runner’s speed and starting distance from second base, the location of the pitch, etc.) is used to estimate the probability that each attempt to steal will be successful. This is used to estimate how many of a catcher’s attempts would be made into outs by an average catcher, and this is in turn is compared to how many actually were made.

CS Above Average is new for this season, but data exists to calculate it back to the 2016 season. And since that season, no catcher has been more superlative than the hypothetical average backstop than Realmuto. Realmuto has seen 342 attempts to steal second since 2016. An average catcher, presented with those exact same attempts, would be expected to send the runner back to the dugout 24% of the time. Realmuto has instead turned 36% of those attempts into outs, for a cumulative total of 40 outs above the average. Second place, shared by Martín Maldonado and Salvador Perez, is 26. Over the course of his career, Realmuto has unquestionably been one of the best, and probably the best in baseball at punishing base stealers.

So what does Statcast say about Realmuto’s 2023? That he has a mere 1 CS Above Average, just the tiniest bit better than what you would expect an ACIB (Average Catcher in Baseball) to put up. He ranks 19^th among qualified catchers for this statistic. How is this happening? In addition to calculating the likelihood of an individual attempt to throw out a runner, Statcast assigns credit to the individual aspects of the attempt to explain how the catcher turned an expected steal into an out, or an expected out into a successful steal. The four categories are throw speed, exchange time (the time it takes the catcher to pick the ball from their glove and throw it), and teamwork (in other words, situations in which it’s the fielder’s effort making the difference rather than the catcher’s.)

Realmuto’s pop time in 2023, a combination of throw speed and exchange time, is unmatched at 1.84 seconds; the next-speediest catcher is three-tenths of a second slower. Add up all of the differences made by Realmuto’s throw speed, and you get the equivalent of 3 caught stealings that an average catcher wouldn’t have. Add up the differences made by his exchange time, and you get 2 extra caught stealings. So clearly, Realmuto isn’t getting weaker or slower. The teamwork is a wash; Realmuto has 1 extra caught stealing thanks to good work on bounced throws by his teammates, and has 1 fewer than you’d expect thanks to some sloppy work from them on catching the ball on the fly.

So it’s the last category, accuracy, that’s been dragging him and his numbers down. Add up all of the differences—both good and bad—made by Realmuto’s accuracy on caught stealing attempts, and you get 4 fewer steals turned into outs than you’d expect. This isn’t new; looking at Realmuto’s 40 caught stealings above average since 2016, negative 15 of them can be explained by accuracy. Which is to say that he has 40 CS above average in spite of his accuracy. In fact, his accuracy entirely wipes out the impact of his excellent exchange time (15 CS AA); it’s his blazing throw speed that explains the vast bulk (30) of his league-leading CS AA since 2016. That extraordinary speed does seem to come at the price of precision, and the price has been higher in 2023.

That said, there isn’t necessarily reason to believe he’s declining. While the accuracy issue is real, the seeming worsening of the problem this season may be the result of small sample size and chance (though the concern of something “real” behind the numbers might be elevated given that the stealing environment is vastly different this season). At any rate, while Realmuto’s accuracy has left something to be desired, his cannon of an arm and greased-lightning reaction time are still the envy of backstops everywhere. Go ahead, runners dreaming of basepath burglary— challenge him.