February 2, 2016

Baseball Therapy

The Dark Side of Pitch Framing?

by Russell A. Carleton

Okay, okay, we get it! Catcher framing is really superduper important. Getting an extra called strike can turn an at-bat, which can turn an inning, which can turn a game, which can turn a season, which can turn someone’s summer from one that they’d rather forget to one that they’ll write a bad novel about 20 years later.

I don’t want to pooh-pooh how important catcher framing is. The good framers really are worth 20 runs or so more per year than the average guys at this point, so there’s plenty of hay to be made here, and teams are clearly jumping on the pitch framing bandwagon. But here’s a question worth investigating. Other than stealing the occasional strike on a ball that was actually just outside (or bumbling in the other direction), does a good framing catcher have any other effects? Maybe if a pitcher knows that he’s got a good framer behind the plate, he knows that he can work the edges of the zone a little more. Maybe in his preparation for the game, he plots out a few sequences in which he will try to exploit that, setting up a pitch where he is likely to get a strike if the batter takes it. Maybe since he knows that the strike zone is a little “wider” with this catcher, the pitcher knows that the batter will have to expand his zone a little (which Rob Arthur has shown actually happens), so he spends more time “baiting the hook” and trying to get a batter to swing at something a little too far away and poke at it. We understand very little about secondary effects in baseball.

But there’s a methodological issue here. It’s not enough to look at a team that has a good framing catcher and compare results for those pitching staffs. What if the catcher just happened to have the pleasure of catching a pitching staff that was just good, regardless of his amazing framing abilities? We need another way to try to tease this out of the data. Thankfully, baseball provides us with a nice natural experiment.

Warning! Gory Mathematical Details Ahead!

We’re going to take advantage of the fact that baseball teams usually carry two catchers and that the backup backstop is usually the member of the bench who is mostly likely to see game action. After all, catchers need more breaks than do most players. It’s bad form to let your catcher don the tools of ignorance in a day game after a night game or on both ends of a doubleheader. Some pitchers have personal catchers, usually the backup, so the regular guy has a scheduled beat that he skips.

Now, framing is one of those skills that has been talked about for years by teams, and there was at least some idea that it existed, but part of why it took so long to break into the public consciousness is that it’s something that’s so hard to miss unless you’re looking really really hard. Still, even before catcher framing became a big thing, there were catchers who were well known as “good to throw to” and a few backup catchers who were beloved by their pitchers and had curiously long careers given that they were awful hitters. While the guys on the extreme edges were apparently wellbknown, I wonder how many pitchers could have picked apart the shades of grey in the middle, to the point where it would have entered their minds to fashion a strategy that incorporated that idea.

Instead, to answer the question of whether a good framer can make a pitcher better (or at least, different), we’re going to look at the edges.

I looked for situations that met the following criteria:

· It was somewhere between 2006 and 2015.

· A pitcher who had faced at least 250 batters in the season in question faced off against a batter who came to bat at least 250 times that season.

· The pitcher had not switched teams during the course of the season. This allows us to rule out cases where a pitcher was traded from a good catcher to a bad catcher (or vice versa), but where he also went to a less (or more) favorable park or was put in front of seven other guys who were better (or worse) at fielding.

· The pitcher and catcher in question worked together for at least 100 batters that year.

· The pitcher worked with two different catchers for at least 100 PA that year. One of them was in the top fifth of the league in BP’s called strikes above average (CSAA) metric. The other was in the bottom fifth.

All in all, it left me a data set of nearly 25,000 plate appearances to work with. As is my custom, to control for the quality of the batter and pitcher involved in the plate appearance, I controlled for their relative strengths using the log-odds ratio method. This creates a control variable based on the expectation that the plate appearance would have ended in a strikeout (or walk or whatever) anyway.

I set up a series of logistic regressions which used this control variable as a predictor and also an indicator of whether the pitcher was throwing to the good or the bad catcher. So we have the same set of pitchers in front of the same (basic) defense in the same home park and are controlling for the hitters they faced, with the only difference being whether the catcher is the good one or the bad one. If the pitcher is somehow getting better (or different) results than we might otherwise expect, we can at least make a good case that we should blame the pitcher.

I looked at a series of outcomes, including walks, strikeouts, singles, home runs, outs on balls in play, along with BABIP, OBP, and different types of batted balls. Did the catcher make a difference?

Not surprisingly, a good framing catcher did exactly what you might think for a pitcher when it came to strikeouts (they went up) and walks (they went down). Remember, this is compared to the pitcher’s own baseline, controlling for the hitters whom he is facing. But did any other outcomes change? Actually, no. The pitcher was just as likely to give up a single, double/triple, home run, or record an out on a ball in play. BABIP was also unaffected. OBP did go down, but that was largely the effect of the walk and strikeout effects. But something did emerge from the chaos.

When there was a good framing catcher behind the plate and the batter put the ball into play, it was more likely (again compared to the pitcher’s baseline rate and the batter’s whom he was facing) that the ball would be a grounder and less likely that it would be a fly. (Line drive rates appeared unchanged.) How big of an effect was it?

Let’s say that we have a situation where a pitcher with an overall 40 percent GB rate (roughly league average) is facing off against a batter who hits a worm burner 40 percent of the time. In front of a good framing catcher, we would expect that to jump to about 41.1 percent. In front of a poor framing catcher, we expect a grounder rate of 39.4 percent. There are similar effects (in the opposite direction) for flyballs.

Again, that 1.1 percent bump up in groundball rate might not seem like much, but let’s do a little math. In 2015, the average MLB team saw 4,349 balls hit into play. Let’s assume that the no. 1 catcher for a team handles 75 percent of the playing time, and so he sees 3,262 of those balls hit in front of him. One-point-one percent of that would be 36 balls in play that would be “transformed” from flyballs into groundballs. Not necessarily groundouts, mind you, but even hitting the ball on the ground, rather than in the air has value. Research by none other than Harry Pavlidis suggests that turning a flyball into a grounder is worth between .12 and .15 runs. (One main reason, flyballs have a chance of going over the wall…)

Even taking the low end of that range, 36 “extra” groundballs times .12 runs is 4.32 runs saved above the average catcher. A good framing catcher can add about 20 runs of value above average just by snagging a few extra strikes. It looks like he adds four or five more, likely by getting a hitter to swing at a bad pitch or two and getting a groundball, rather than a flyball, as a result.

Or does he?

There’s a problem. The finding on groundballs doesn’t directly square with the findings about rates of HR, extra-base hits, singles, and outs in play (as well as BABIP) not being significantly affected. Here we might have to dig a bit deeper to explain that. In general, groundballs are less likely to produce an out than a flyball, but perhaps the “extra” groundballs are from the batter reaching a bit more and making weaker contact, enough that the rough number of singles and outs would stay the same. Home runs and extra-base hits are relatively rare events to begin with. They can also be very park dependent, although my method tries to control for that. Maybe there’s just not enough signal and a lot of noise in there. I thought that might be the case until I noticed that while the effect of a good framer for home runs wasn’t significant, the trend line actually pointed up. A good framer, despite “generating” more groundballs than we might expect, actually also generated more home runs.

There might be a more sinister explanation that might be going on as well, one that might effectively wipe out what we’ve gained. The Peltzman Effect, which is the idea that when something comes along that reduces the risk of something, people respond by taking more risks in some other related area. For example, people wearing bicycle helmets are without a doubt safer if they have an accident on their bike and hit their heads. However, when people wear bike helmets, they tend to ride their bikes into riskier situations. And a helmet only protects your head.

In the same way, a pitcher throwing to a good framer might realize that he has a little extra margin to work with, and so he decides to get a little more swashbuckling in painting the edges. And sometimes you when you work around the edges, you miss and catch a bit too much of the plate. And someone hits a flyball that goes really far. Sure, the batters hit more grounders, but they also get a few more chances to unload when they do hit a flyball.

The Dark Underbelly of Pitch Framing

I have no doubt that pitch framing is a valuable skill, and value beyond just the extra strikes that a catcher can steal for a pitcher. Having a good framer behind the plate means that the pitcher is working with a slightly larger strike zone. In isolation, that’s a great thing, but there are few things in baseball (or life) that happen in isolation. When you push a button, it changes all of the other buttons.

Pitchers might become a little too enamored of the bigger strike zone and essentially give up a few extra home runs because there’s a good framer behind the plate! In theory, that shouldn’t be happening. Pitchers do induce more groundballs when they are pitching in front of a good framer. But they aren’t actually getting better results on their balls that go in play. So, we’re left to either explain that away as randomness or reckon with the fact that catcher framing actually has an unfortunate downside.

With that evidence in mind, may I ask a somewhat contrarian and unsettling question. Might we find evidence that even the “called strike” portion of catcher framing, while undoubtedly real insofar as how we have defined it, might have a kickback effect that we aren’t accounting for? For example, is it possible that Milwaukee pitchers looking into Jonathan Lucroy are tailoring their pitching plans to Lucroy’s framing strengths, rather than focusing on an optimal strategy for pitching? And while we can see the PITCHf/x plot of where the ball crossed the plate and see that Lucroy “made” it a strike, could it also be that a fastball on the edge should never have been called in the first place?

Russell A. Carleton is an author of Baseball Prospectus. Follow @pizzacutter4
Click here to see Russell's other articles. You can contact Russell by clicking here