May 24, 2016
Framing the At-Bat
Good-framing catchers, as best as we can define them, seem to have magical powers. They can “steal” extra strikes for their pitchers, and while it might not seem like much in the moment to get an extra borderline call, it adds up. The generally accepted consensus has been that the top framers can save their team 20 runs compared to a merely average framer. Compared to the bottom of the barrel, that swing is 40 runs. When the general public figured out how big that effect was, they rightly made a big deal about it. (When teams found out, they quietly made a big deal out of it. In fact, in Francisco Cervelli’s case, they just made more than 30 million big deals about it.)
A few weeks ago, I looked into whether a good framer might just have some extra super powers. For example, even when he’s not framing pitches, does he change the outcomes of the balls that go into play? It turns out that the answer is yes… sorta. Pitchers pitching in front of good framers tend to see an uptick in their groundball rates, although they also see a small bump up in their home run rates as well, and it turns out that it all basically cancels out.
BP reader Richard Bergstrom contacted me with a different question that I thought was worth a peek. Bergstrom covers the Colorado Rockies for the site Rockies Zingers and noted that the Rockies in the offseason brought in framing expert Tony Wolters and gave him a place on the Opening Day roster. Perhaps the reader is familiar with the fact that Colorado is currently in its 24th year of trying to find a pitching strategy that works a mile above sea level. Maybe a good framer could help pitchers by getting them into advantageous counts and perhaps cutting down a bit on his pitch count so that he can stay in the game for an extra batter or two?
The extra batter or two might be particularly important. Figure that a good framing catcher might handle 100 games per year. If those 100 extra batters put up just a league average OBP against the pitchers that he’s catching, that’s a bit shy of 70 extra outs or 20-something extra innings that the bullpen doesn’t have to cover. Every little bit helps.
Warning! Gory Mathematical Details Ahead!
A few months ago when the crew at Baseball Prospectus put out historical catcher framing numbers, based on Retrosheet data, it presented a nice chance to look at a natural experiment. The problem with studying framing is that a good framer like Francisco Cervelli generally only catches the pitchers of one team during a year. It might be that for whatever else I’d want to look at, he just happened to be pitching in front of a bad pitching staff. But thankfully for us, teams usually carry two catchers, both of whom will get non-trivial amounts of playing time during the year. I looked for teams that had a catcher who was in the top 20 percent of the league that year for framing and another who was in the bottom quintile. In this way, we can keep the same sample of pitchers, but vary the framing ability of the catchers.
I used data from 2011-2015 and selected for teams with a good and bad framer, and then looked at their pitchers and the individual batters that they faced in the season in question. First up, whether good framers help pitchers get into good counts. We know that the most important pitch in any plate appearance is strike one. Recording strike one on the first pitch sharply increases the likelihood of a plate appearance ending up in a strikeout and reduces the chances of it ending up in a walk.
So I looked into how often—on the first pitch of the at-bat—a pitcher got a called strike (only on pitches when the batter didn’t swing) and how often a batter took a called strike. I normalized for expectations based on those numbers, and then ran a binary logistic regression that looked at whether the framing abilities of the catcher predicted a higher likelihood of a called strike. And the answer was yes. I looked at whether a good framer was more likely to get a 0-1 count, by any means (not just called strikes) and that was also a yes. This probably isn’t shocking given that the definition of a good framing catcher is that he gets extra called strikes. I found similar results on 1-0 counts.
In my previous work, I’ve found that good framers seem to be associated with an increase in strikeout rate for the pitchers that they catch and a corresponding decrease in walks, and this all fits in nicely. Strangely enough, I didn’t find similar results on 3-0 and 3-1 counts. Controlling for overall tendencies to get a called strike in one of those situations, catcher framing quality didn’t seem to make the same difference. There were fewer of these situations… every plate appearance has a first pitch but many fewer have a 3-0 count, so this could be a (statistical) power problem.
Does that translate into a reduced pitch count? Actually no. I created an expected pitches control variable based on the batter and pitcher’s average pitches per plate appearances and then ran a regression with that as a predictor and the indicator of catcher framing quality. Catcher quality made no difference in the model. I also ran pitcher and batter as random effects, but that didn’t change the outcome either. It seems that good framers get their pitchers into advantageous counts, which perhaps entices pitchers to go for the (slightly) more pitch-intensive strikeout than to try and get an out in play. Maybe he uses his advantage to “waste” a set-up pitch.
Well, Nothing Else Has Worked
We’ve also learned is that catcher framing seems to have a bigger effect early in the count. Starting an at-bat 0-1 rather than 1-0 is a big deal in terms of the rest of the at-bat. It’s the place in the game where empires and dynasties are won and lost, but most fans dismiss it as “the boring stuff.” The chess match between pitcher and batter (and now, we realize, the catcher) is the massively under-studied and under-understood. These may seem like minor details, but they are the essentials of the battle in baseball.