“Catcher Framing” on SoSH.com

My article on catcher framing at sonsofsamhorn.com.

Catcher framing: Brian McCann’s differences from the average strike zone since 2008

Posted in Uncategorized | Leave a comment

“Poor umpire performances” on SoSH.com

My article identifying the best and worst umpires at calling strikes is up at sonsofsamhorn.com.

Chris Segal, 2014: Fewest missed calls

Doug Eddings, 2008: Most missed calls


Posted in Strike zone, tools, Umpires | Leave a comment

Baseball on SoSH

Many of my baseball posts are now on Sons of Sam Horn Baseball Central site:

Posted in tools | Leave a comment

The Always Changing Strike Zone

Published on sonsofsamhorn.com April 9, 2015; scraped from archive.org in 2018

Balls and strikes are the building blocks of baseball. The pitcher’s stuff, the way the catcher receives the ball and how the umpire sees itare just a few of the factors that comprise the strike zone. Ian Yorkexamines the always changing strike zone and the influence of ball-strike counts on subsequent calls.

Baseball fans know that there is really no such thing as the strike zone; there are many different strike zones, with different shapes and sizes. There is the rulebook strike zone, which is what many people have in mind when they talk about “the strike zone”; but not only has the official zone been changed a dozen times since baseball started, no umpire has ever actually called the official zone anyway. There is a de facto strike zone, which is what umps actually call. Of course, that “strike zone” is really two strike zones, one for right-handers and another for left-handed hitters. The strike zone is different shapes and sizes: Right-handed batters have a slightly larger zone than left-handed batters.

But the de facto strike zone has changed significantly since the advent of PITCHf/x in 2008, when umpires started being evaluated on how well their call matched the actual location of the pitch. Since 2010 the zone has gradually expanded, year by year, so that the zone in 2014 is closer to the official zone’s shape and size than the 2008 zone was.

Batter-handedness and PITCHf/x are not the only things that affect the strike zone. Catchers have varying ability to affect the size and shape of the zone. Catchers who are good at framing pitches can expand the zone – sometimes all around the zone, sometimes in one particular region – and catchers who are bad at framing can cause the zone to shrink. A team might see one strike zone for most of a series, then abruptly see a different zone when the backup catcher starts the series’ finale.

On top of that, even though most umpires call a strike a strike and a ball a ball, there are several umpires whose personal strike zone is quite a bit larger or smaller than the average umpire. A team may see a different zone from one day to the next as new home-plate umpires rotate in.

Changes to the strike zone take place at an even more discrete level than by umpire or catcher. In addition to the slow, year-to-year expansion of the de facto strike zone, the intra- and inter-series changes in catcher framing, and the daily changes in the umpire’s strike zone, the strike zone changes with almost every pitch, because umpires call a different zone for different ball/strike counts.

Here is what the strike zones looked like in 2014 for all batters, umpires, and strike counts:

Strike Zone Changes in Counts IMG 1

Note: These strike zones are shown from the umpires’ viewpoint.

The PITCHf/x database was used to select called balls and strikes from 2014. These plots show the probability of a pitch being called a strike in sub-regions in and around the plate for left- and right-handed batters. In blue areas, the pitch will almost always be called a ball; in green areas, it is almost always a strike. The red areas, where a pitch might be called either way, delimit the edges of the de facto strike zone. The white polygons show the 50% zone – outside the lines, a pitch is more likely to be a ball and inside it is more likely to be a strike.

Next, the zone was subdivided further by each of the 12 different possible ball/strike counts. Here is the frequency:

Strike Zone Changes in Counts IMG 2

The 0–0 count, of course, is the most common (27.3% of all pitches), and the 3–0 count is the least common (just 1.1%).

Here is how the probability maps look for each count. The white polygon on each plot shows the zone for all counts:

Strike Zone Changes in Counts IMG 3

You can easily see that the strike zone changes size in some of the counts – the zone for 0–2 counts, for example, is clearly quite a bit smaller than the zone for 0–0 counts.

We can make these changes easier to see by plotting differences from the average probability. That is, if a pitch region has a 0.9 chance of being called a strike in an 0–0 count, but only a 0.1 chance in an 0–2 count, it would have a difference of –0.8. Regions where the probability is lower are colored blue with the intensity corresponding to the probability difference; those with a higher probability are red; neutral regions, where the probability is the same, are white or faintly colored:

Strike Zone Changes in Counts IMG 4

The 0–0 counts are essentially neutral ‒ virtually identical to the overall average. At 0–2, the strike zone is much smaller; the edge of the normal strike zone is less likely to be called a strike. The umpires, consciously or unconsciously are helping out the batter and making him less likely to strike out. The same trend, though less blatant, is also there in other pitchers’ counts: 0–1, 1–2, and 2–2. Even at 1–1 and 3–2, the zone becomes slightly smaller than in the neutral 0–0 count.

At 3–0, umpires do the opposite; they expand the zone significantly, making it easier to throw a strike and avoid walking the batter. Again, the same trend is there for other hitters’ counts; 2–0 and even 1–0 show a bit of zone expansion.

None of this is new or surprising to anyone who follows the game; umpires have been doing this since they first started calling balls and strikes. It is something to consider whenever the annual plea for robots to call balls and strikes comes up. Whether it is objectively better to have exactly the same strike zone for every umpire, count, and catcher, it would be a radical departure from how the game has been played. Just as the PITCHf/x-driven expansion of the strike zone could be one of the factors in the profound reduction in offense baseball-wide since 2010, having a constant strike zone would certainly change baseball in drastic and unpredictable ways.

Posted in Uncategorized | Leave a comment

Catcher Framing or: How I Learned To Stop Worrying And Love The Umps

Originally published on sonsofsamhorn.com April 6, 2015; scraped from archive.org in 2018, possibly with some missing images.  

Nearly twenty-five thousand batter-pitcher confrontations take place each season, about eighteen inches from the catcher. The role of the receiver has evolved in the modern game, with catcher framing the newest component to deeper understanding and analysis of the game.

Batters have four or five plate appearances per game. Prolific fielders will average 4 to 5  plays per game ‒ Andrelton Simmons had 4.3 assists, putouts, and errors per game in 2014. Successful starting pitchers will throw around 100 to 120 pitches per game; Relievers can throw another 30 or 50 pitches in those games. Meanwhile, catchers handle the ball 150 times or more each game. It should follow that a catcher has the opportunity to affect the game’s outcome more than any other player except the pitcher.

One of the ways he can do that is by stealing strikes ‒ converting marginal pitches that the umpire would normally call a ball, into a strike. Or, converting a potential strike into a ball. That is catcher framing.

Framing and PITCHf/x

The advent of PITCHf/x has helped quantify the effect of catcher framing, and it is much larger than anyone had suspected. (Unless some team analysts had already identified it and kept it a secret, of course.) Even conservative estimates of the impact of catcher framing conclude that a catcher with strong framing skills contributes the equivalent of dozens of runs ‒ and therefore multiple wins to the team over the course of a season ‒ through framing alone.

Catcher framing has been measured in many ways, with both complicated and subtle approaches being used to get a detailed understanding of the exact effect each catcher has on the strike zone. Still, even a fairly simple approach can help show trends, strengths, and weaknesses. Here, I am going to look at catcher framing as a difference in the probability that a pitch is called a strike.

On average, in any particular region around the plate, a pitch has a certain chance of being called a strike. In the center of the strike zone, that’s nearly 100%; way outside, it’s nearly 0%. There’s not much a catcher can do about those pitches.

At the edges of the de facto strike zone a catcher can influence the probability. For some catchers, a pitch just outside the top of the strike zone might be called a strike 90% of the time, even though baseball-wide that pitch might only have a 10% chance of being a strike. Other catchers might hurt their pitcher’s chances, causing the umpire to call a ball in a region that has a 90% chance of being called a strike.

One Of The Best

Let us start with the catcher acknowledged as one of the best, if not the best, at framing: here’s how Jose Molina influenced the probability of having a pitch called a strike in 2008 (the earliest year for which we have PITCHf/x data). As with all PITCHf/x graphs, this is from the umpire’s point of view.  Across this plot, areas where Molina was more likely than the rest of baseball to have a strike called are shown in red. Areas where he was less likely to have a strike called are blue. The gray polygon shows the de facto strike zone in 2008 — the region in which a pitch had a 50-50 chance of being called a strike.

Molina Zone 2008

There’s not a lot of blue there. In 2008, Molina had more strikes called at the edges of the strike zone than any other catcher. For left-handed batters, he pushed the strike zone down and outward; for right-handed batters, up and in. Virtually nowhere was Molina less likely to get a strike than the league average.

Framing is a repeatable skill. Year after year, the same catchers push the strike zone outward for their (lucky) pitchers, and the same catchers  shrink the zone for their (unfortunate) pitchers. Here are Molina’s differential probability maps from 2008 through 2014:

Molina Zone 2008-2014

Every year has a lot of red, though by 2014 the intensity of the red is dropping off — the 39-year-old Molina in 2014 was still better than the rest of the league, but not by as much.

It’s worth pointing out that the strike zone (the grey polygon in each frame) changed significantly over that period, too. The strike zone that umpires called in 2014 was approximately 10% larger than the one they called in 2008, with the expansion happening almost entirely at the bottom of the zone. The extra strikes below the 2008 strike zone that Molina grabbed would be clearly in the strike zone in 2014, and would be unambiguous strikes for most catchers. Nevertheless, while the strike zone moved outward, Molina’s expanded strike zone moved outward with it.

One Of The Worst

What does a bad framing catcher look like? Ryan Doumit has consistently been a terrible framer. Here’s what his probability maps look like from 2008 through 2013. (In 2014, understandably, he only caught 2 games for Atlanta.)


Since at least 2008, Doumit has lost strikes for his pitchers everywhere around the zone. In 2013 he wasn’t completely hopeless at the top of the zone for right-handed batters, but there’s not enough there to compensate for the losses everywhere else.

Average Catcher

Here is league-average catcher Alex Avila in 2014:

Avila Zone 2014

There is a little misty color there, but in general Avila has no particular strengths, and no particular weaknesses.  His pitchers get the strike zone: No more, no less.

But, interestingly, there are other ways to be average. Evan Gattis in 2014 was just about exactly as average as Avila (I’ll explain what I mean by that in a moment). Here’s Gattis’s 2014 map:

Gattis Zone 2014

Gattis achieved averageness by being pretty good at expanding the inside part of the plate to righties while losing the outside.

Quantifying The Skill

There are many ways of trying to put an actual number on the extent to which framing increases strikes, but they mostly come up with very similar answers. Here’s how I identify great, terrible, average, and good framers: combine the probability of a particular catcher getting a called strike in each region, the actual number of pitches he received in that region, and the league probability of a called strike in that region, to estimate the amount of called strikes that the catcher influenced per game.

For example, a pitch just outside the bottom corner of the zone might be called a strike 10% of the time across the league. If, in four games, our catcher had 10 pitches in that region and got strikes on 9 of them instead of the expected 1, he is awarded 8 extra strikes; that means 2 extra strikes per game.

Obviously, this approach misses out on many, many subtleties. It doesn’t include the pitcher’s contribution; it doesn’t consider the umpire or the game situation ‒ the strike zone changes with different counts, something I will explore this season. It doesn’t take into account the fact different pitcher/catcher combinations throw pitches to different locations. But, reassuringly, this approach ends up with numbers very similar to other approaches, so it’s probably measuring the same general thing.

Here are some of those numbers, showing that framing is a fairly consistent skill year to year:

Extra strikes per game
20083.19-4.55 –2.75
20093.00-2.90-1.04 –2.98
20102.85-1.20-0.10 –3.351.73
20112.06-2.500.02 –3.262.59
20122.10-3.120.14 –2.841.92

Doumit is consistently awful, Avila is consistently average, Molina, Lucroy, and McCann are consistently good to excellent (as are some other catchers, but that is a story for another day). All three of the elite framers in this set do show a decrease in extra strikes per game over time, although they all remained very good. In the case of Molina and McCann, that could just be age catching up with them, but another possibility is that as the overall strike zone has expanded over time, there are fewer opportunities for framing.

Turning pitches into runs

In 2008, Dan Turkenkopf estimated that each switched call is worth about 0.13 runs on average. If that’s about right, then having elite framers like Molina or Lucroy on staff, instead of average catchers, adds between a quarter and a third of a run per game over an average catcher, or the equivalent of about an extra 20-35 runs of value over an entire season from framing alone.

Returning to the average framers: In 2014, Avila lost about 0.37 extra strikes per game and Gattis lost about 0.14 extra strikes per game. As I said, they reached that in different ways; Avila is neutral all over the zone while Gattis shifts the zone slightly left.

This is also a common pattern among good framers: They pick up strikes from one part of the zone, but lose them elsewhere. Aside from Jose Molina, not many framers add strikes all around the zone. Even Molina lost his ability to expand the strike zone in all directions as he declined from his peak. By 2014, he was not getting many extra strikes at the bottom or the sides of the zone, though he was still able to get them at the top.

For example, Brian McCann has ranged from a good to an elite framer over his career, but he gets there in a very different way than Jose Molina:

McCann Zone 2008-2014

Rather than expanding the strike zone in all directions, as Molina did from 2008 through 2013, or expanding the zone at the top while being neutral at the bottom like 2014 Molina, McCann greatly expands the bottom of the strike zone and gets fewer than average calls at the top of the zone. And he has done this consistently, year after year, for the seven years that we have had PITCHf/x data.

Not only is catcher framing a reproducible skill in general, but particular catchers can focus on framing specific regions of the strike zone, and those regions, too, are generally reproducible year to year. This predictability might let some teams choose specific sets of pitchers and catchers, working to the strengths of each.

Posted in Uncategorized | Leave a comment