Catching Up and Catching Down
Pitch-framing! Am I right? It’s still very much a fascinating subject, which is one of the reasons I write about it so often. But by this point we have a pretty good idea who’s good at it and who’s bad at it. That ground’s been covered. We know that Jose Molina is great. We know that Ryan Doumit was a problem. Yet we can break things down further still. Often, people don’t go beyond describing a guy as good, bad, or okay. But there are actually specific types of framers.
Which makes plenty of sense, doesn’t it? There are great hitters and there are bad hitters. Among them, there are guys with tremendous plate coverage, but there are also high-ball hitters and low-ball hitters. Every part of the zone area is different, and every player is different, so we should expect that different players respond differently to pitches in different parts of the zone. How this relates to framing is that some guys might be better with receiving high pitches, while other guys might be better with receiving low pitches. Intuitively, why not? And thanks to some awesome updates at Baseball Savant, this couldn’t be much easier to examine.
Of course, there are lots of ways you could go with this. Probably, different catchers are differently able to receive different pitch types. Catchers are differently able to receive inside pitches and outside pitches, and there are differing abilities inside and outside of the zone. What follows is just going to look at high and low pitches on the border of or within the PITCHf/x strike zone. This is one of many ways to slice up the information.
So, who looks best on should-be high strikes, and who looks best on should-be low strikes? I collected data from the entire reliable PITCHf/x era, spanning 2008-2013. I set a minimum of 10,000 pitches, which yielded a sample of 102 different catchers. For low pitches within the zone, I selected the lowest three zone boxes. For high pitches within the zone, I selected the highest three zone boxes. This study ignores the middle three boxes, and also the areas outside of the PITCHf/x strike zone. All that information would be great, when tackling other questions.
The sample gave more than 200,000 called pitches in the lower three zones, of which more than 77% were called strikes. It gave more than 130,000 called pitches in the higher three zones, of which more than 85% were called strikes. In the sortable table now, I’m going to show you everything. These are all 102 catchers, their performances, and the differences between low-strike rate and high-strike rate.
| Catcher | Low# | LowStr% | High# | HighStr% | Diff |
|---|---|---|---|---|---|
| A.J. Ellis | 2269 | 73.7% | 1655 | 88.6% | 15.0% |
| A.J. Pierzynski | 4462 | 68.6% | 3771 | 90.4% | 21.8% |
| Adam Moore | 564 | 70.4% | 392 | 87.2% | 16.9% |
| Alex Avila | 3415 | 79.0% | 2295 | 86.7% | 7.7% |
| Bengie Molina | 2336 | 69.6% | 1844 | 85.1% | 15.4% |
| Bobby Wilson | 1063 | 82.7% | 633 | 87.7% | 5.0% |
| Brad Ausmus | 749 | 76.8% | 562 | 83.5% | 6.7% |
| Brayan Pena | 1792 | 78.7% | 1175 | 89.5% | 10.8% |
| Brett Hayes | 658 | 78.3% | 491 | 84.7% | 6.5% |
| Brian McCann | 4780 | 89.3% | 2600 | 80.8% | -8.4% |
| Brian Schneider | 1767 | 72.5% | 1149 | 85.5% | 13.0% |
| Buster Posey | 2285 | 89.4% | 1721 | 88.1% | -1.3% |
| Carlos Corporan | 896 | 86.9% | 500 | 83.8% | -3.1% |
| Carlos Ruiz | 4856 | 78.4% | 3054 | 80.7% | 2.4% |
| Carlos Santana | 2161 | 71.6% | 1643 | 89.5% | 17.9% |
| Chris Coste | 795 | 79.5% | 570 | 77.9% | -1.6% |
| Chris Iannetta | 3577 | 68.7% | 2456 | 88.5% | 19.8% |
| Chris Snyder | 2483 | 81.4% | 1676 | 87.1% | 5.7% |
| Chris Stewart | 1520 | 87.1% | 810 | 86.4% | -0.7% |
| Corky Miller | 556 | 80.2% | 389 | 83.8% | 3.6% |
| Craig Tatum | 618 | 79.4% | 432 | 85.9% | 6.4% |
| David Ross | 1986 | 87.1% | 1205 | 82.3% | -4.8% |
| Derek Norris | 1133 | 86.8% | 611 | 85.8% | -1.0% |
| Devin Mesoraco | 1090 | 80.3% | 688 | 89.7% | 9.4% |
| Dioner Navarro | 2487 | 64.8% | 2139 | 88.6% | 23.8% |
| Drew Butera | 1183 | 81.0% | 778 | 86.4% | 5.4% |
| Eli Whiteside | 1243 | 86.1% | 825 | 86.4% | 0.3% |
| Erik Kratz | 773 | 88.4% | 494 | 88.1% | -0.3% |
| Francisco Cervelli | 1275 | 76.9% | 785 | 84.5% | 7.6% |
| George Kottaras | 1536 | 77.0% | 1030 | 85.6% | 8.7% |
| Geovany Soto | 3999 | 78.8% | 2979 | 88.0% | 9.2% |
| Gerald Laird | 3055 | 68.5% | 2028 | 85.1% | 16.5% |
| Gregg Zaun | 1245 | 81.8% | 692 | 80.8% | -1.0% |
| Guillermo Quiroz | 568 | 70.8% | 416 | 88.9% | 18.2% |
| Hank Conger | 996 | 88.9% | 580 | 85.7% | -3.2% |
| Hector Sanchez | 633 | 80.3% | 503 | 90.3% | 10.0% |
| Henry Blanco | 1650 | 80.4% | 1037 | 81.5% | 1.1% |
| Humberto Quintero | 2403 | 79.4% | 1538 | 82.5% | 3.2% |
| Ivan Rodriguez | 2485 | 76.1% | 1634 | 80.2% | 4.1% |
| J.P. Arencibia | 2526 | 83.1% | 1643 | 83.3% | 0.2% |
| J.R. Towles | 870 | 78.9% | 478 | 75.9% | -2.9% |
| Jarrod Saltalamacchia | 3071 | 69.7% | 2460 | 90.0% | 20.2% |
| Jason Castro | 1684 | 84.6% | 1095 | 84.5% | -0.1% |
| Jason Jaramillo | 775 | 73.9% | 475 | 81.1% | 7.1% |
| Jason Kendall | 3087 | 80.8% | 1761 | 76.7% | -4.1% |
| Jason LaRue | 773 | 78.5% | 493 | 75.1% | -3.5% |
| Jason Varitek | 2083 | 58.3% | 1839 | 90.4% | 32.0% |
| Jeff Mathis | 3349 | 83.3% | 1828 | 84.7% | 1.4% |
| Jesus Flores | 1396 | 78.2% | 905 | 84.1% | 5.9% |
| Jesus Montero | 532 | 73.7% | 438 | 85.2% | 11.5% |
| Joe Mauer | 3192 | 62.8% | 2881 | 92.7% | 29.9% |
| John Baker | 1919 | 78.7% | 1137 | 79.5% | 0.8% |
| John Buck | 4398 | 77.1% | 2851 | 86.7% | 9.6% |
| John Hester | 562 | 82.0% | 288 | 77.4% | -4.6% |
| John Jaso | 1614 | 78.4% | 1075 | 83.5% | 5.1% |
| Jonathan Lucroy | 3342 | 91.4% | 1638 | 83.9% | -7.6% |
| Jorge Posada | 1455 | 66.4% | 976 | 84.7% | 18.3% |
| Jose Lobaton | 1207 | 86.5% | 693 | 87.0% | 0.5% |
| Jose Molina | 2806 | 80.1% | 2013 | 90.7% | 10.6% |
| Josh Bard | 1302 | 75.4% | 772 | 82.8% | 7.3% |
| Josh Thole | 2036 | 84.2% | 1402 | 87.0% | 2.8% |
| Kelly Shoppach | 2856 | 71.0% | 1969 | 86.9% | 15.9% |
| Kenji Johjima | 1148 | 55.2% | 848 | 88.6% | 33.3% |
| Kevin Cash | 642 | 67.9% | 497 | 88.3% | 20.4% |
| Koyie Hill | 1602 | 75.3% | 1108 | 82.7% | 7.3% |
| Kurt Suzuki | 5283 | 72.9% | 3463 | 85.4% | 12.5% |
| Landon Powell | 757 | 71.7% | 588 | 87.9% | 16.2% |
| Lou Marson | 1906 | 77.2% | 1319 | 86.2% | 9.0% |
| Martin Maldonado | 960 | 89.9% | 424 | 88.2% | -1.7% |
| Matt Treanor | 1806 | 68.2% | 1315 | 85.6% | 17.4% |
| Matt Wieters | 4265 | 79.5% | 3057 | 86.6% | 7.2% |
| Michael McKenry | 1137 | 79.1% | 735 | 84.6% | 5.6% |
| Miguel Montero | 4518 | 88.7% | 2543 | 85.3% | -3.4% |
| Miguel Olivo | 3487 | 74.9% | 2177 | 86.1% | 11.1% |
| Mike Napoli | 2768 | 73.1% | 1886 | 86.8% | 13.7% |
| Mike Redmond | 588 | 64.1% | 440 | 89.1% | 25.0% |
| Nick Hundley | 3363 | 70.4% | 2219 | 85.5% | 15.2% |
| Omir Santos | 660 | 68.0% | 507 | 82.6% | 14.6% |
| Paul Bako | 941 | 79.0% | 569 | 82.2% | 3.3% |
| Ramon Castro | 824 | 68.0% | 689 | 87.8% | 19.8% |
| Ramon Hernandez | 2634 | 77.2% | 1678 | 81.0% | 3.8% |
| Raul Chavez | 532 | 63.9% | 424 | 85.1% | 21.2% |
| Rob Brantly | 743 | 78.1% | 442 | 88.9% | 10.9% |
| Rob Johnson | 1618 | 66.4% | 1138 | 88.3% | 21.9% |
| Rod Barajas | 3042 | 69.4% | 2309 | 89.8% | 20.4% |
| Ronny Paulino | 1878 | 75.6% | 1294 | 86.1% | 10.5% |
| Russell Martin | 5101 | 81.0% | 3465 | 86.3% | 5.2% |
| Ryan Doumit | 3053 | 66.5% | 1956 | 79.3% | 12.9% |
| Ryan Hanigan | 2854 | 82.8% | 2115 | 89.1% | 6.2% |
| Salvador Perez | 1922 | 78.1% | 1190 | 89.7% | 11.6% |
| Taylor Teagarden | 1134 | 73.3% | 744 | 84.8% | 11.5% |
| Tony Cruz | 610 | 88.2% | 382 | 84.8% | -3.4% |
| Tyler Flowers | 1170 | 83.3% | 883 | 88.1% | 4.8% |
| Victor Martinez | 1645 | 71.2% | 1184 | 84.6% | 13.4% |
| Welington Castillo | 1277 | 79.4% | 715 | 84.6% | 5.2% |
| Wil Nieves | 1857 | 81.3% | 1085 | 86.2% | 4.9% |
| Wilin Rosario | 1666 | 76.5% | 995 | 88.0% | 11.6% |
| Wilson Ramos | 1438 | 80.8% | 1023 | 91.2% | 10.4% |
| Yadier Molina | 5773 | 88.0% | 3417 | 83.5% | -4.5% |
| Yan Gomes | 629 | 87.0% | 454 | 89.2% | 2.2% |
| Yasmani Grandal | 565 | 92.4% | 284 | 83.8% | -8.6% |
| Yorvit Torrealba | 3159 | 78.7% | 1901 | 84.2% | 5.6% |
The king of low strikes has been Yasmani Grandal, although his sample is admittedly among the smallest. Following him are teammates Jonathan Lucroy and Martin Maldonado, and then Buster Posey’s a hair ahead of Brian McCann. At the other end, Kenji Johjima looks absolutely dreadful, although he hasn’t caught since 2009 and around then PITCHf/x had some more bugs. But Johjima was suspected to be a pretty lousy receiver. Jason Varitek and Joe Mauer are down there. Dioner Navarro’s still catching, and he’s had his issues with should-be low strikes.
And then you turn to high strikes, where suddenly Mauer reigns supreme. He’s got a big edge on Wilson Ramos, who has a smaller edge on Jose Molina, who has a smaller edge still on Varitek and A.J. Pierzynski. Turn this around, and Jason LaRue had some troubles. Carlos Ruiz looks the worst among prominent active types. Naturally, the bad end of the table features Ryan Doumit.
It’s interesting to see Mauer look so good up high and so bad down low. More generally, it’s interesting to observe that there doesn’t seem to be much of a relationship between low success and high success. The data isn’t entirely all over the map, but it’s pretty scattered about.
The general trend is that the better a catcher is at receiving low strikes, the worse he is at receiving high strikes, and vice versa. It isn’t the strongest trend in the world, but it also makes sense that something like that would exist, and though I haven’t looked at it yet there’s probably a relationship here with catcher height. And given the sensitivity of good receiving, it’s probably quite difficult to prepare to receive and stick low pitches and high pitches. The required movements are going to be different. Catchers, individually, might be better at one than the other.
Included in the table is each catcher’s high-strike rate minus his low-strike rate. The greatest difference belongs to Johjima, then Varitek, then Mauer. So while Mauer isn’t a catcher anymore, as recently as last season he had the greatest difference between his ability to receive high and his ability to receive low. The lowest difference belongs to Grandal, who’s been better with low strikes than high strikes, by almost nine percentage points. Within the sample, 21 catchers have higher low-strike rates than high-strike rates. 35 catchers are within five percentage points of being even.
Given that Mauer and Grandal have been very different catchers, then, might we be able to see anything in the video? The answer is: this isn’t enough video. We can’t learn anything from one-pitch samples. But just for the hell of it, here’s Mauer catching a high strike and Grandal catching a high ball, and Mauer catching a low ball and Grandal catching a low strike.
Mauer, we know, is the taller catcher, by a few inches. It also seems like he has a bit of vertical glove drift as the pitcher is in his delivery, and it might be easier for him to continue that upward than to suddenly reverse direction. Grandal seems adept at catching the low ball in his palm, which helps to minimize the required glove movement with a low pitch. Grandal, in general, is pretty quiet — simply averaging low-strike rate and high-strike rate, Grandal would be tied for fourth in the table, behind Maldonado, Posey, and Erik Kratz. Based on Grandal’s playing time, receiving is a somewhat surprising strength of his.
There’s more to be done with this kind of data. There are a lot of different things that we can try to isolate. For now, this is a beginning attempt, with thanks again to Baseball Savant for being such an outstanding and user-friendly resource. There are most assuredly good framers and bad framers. It’s time to pay a little more attention to the subgroups.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.
Jeff – don’t have a real feel for the sample size and distribution with each box here.
With the three higher boxes, are the samples large enough that the relative distribution of pitches within each box are similar from catcher to catcher? I assume the “framability” of a given pitch within each box will vary some, so it is important that the catchers see similar distributions to compare them against each other.
If it’s 130K and we just assume it’s evenly split over the 102 catchers (obviously it won’t be) that’s ~1300 per catcher. Then with 3 high zones that’s 433 per zone. Is that enough to get a pretty reasonable distribution (in terms of location, pitch type, pitcher handedness) within each box. I have no idea one way or the other, and would be interested to get your thoughts.
Good stuff though!
And somehow I just noticed that you had the individual pitch counts in the table (Yikes, I don’t know how I missed that) and didn’t need to do all the ‘fancy’ math.
Question still stands though – if you take these individual catcher samples and split them into 3 zones, is it large enough for each catcher to have similar distributions to the others within each zone.
I was just looking at this today. The thing that looked intriguing to me was the possible correlation between catcher height and framing high strikes.
Lucroy dominates on low pitches, but not on high ones, whereas even terrible framers like Jesus Montero are able to poach a few high strikes. The numbers don’t show this as much as the heat maps at baseballsavant do, and part of that may be because the out-of-zone, er, zones wrap around the edge, so that they’re not JUST “high” their “high and outside” or “low and inside.” This means part of what’s getting counted here as a “high strike” may just be a lefty strike that’s off the plate away, and just barely above the middle of the zone. Anyway, this is a fascinating area for someone to dig in on.
Jonathan Lucroy’s framing heatmap is basically a U shape. He gets calls off the plate on either side, and a ton of low pitches. Wieters/Mauer/Montero etc. have heat maps that look a bit more like a box – there’s a clear, identifiable grouping of pitches that are over the middle of the plate, but up and out of the zone.
I was thinking about catcher height when Mauer came up as a leader on high strikes. In addition to height though, notice how low Grandal crouches. This really helps the umpire get a clean view of low strikes.
I remember Tony Pena catching from basically a spread-eagle split on the ground, and how many low strike calls he would get. Roger Clemens had a great run from 1990-93 in part because of those extra low strikes.
I agree that Tony Pena was a fantastic pitch framer. But Clemens was even better from 1986 through 1989 with Rich Gedman and Rick Cerone. He had some great games when John Marzano caught. Gedman and Bill Haselman caught his 20-strike out games.
Don Zimmer in a rocking chair could have caught Roger Clemens in his prime.
True. I was thinking about Clemens’ falloff after Pena left, which could have been due to other reasons as well (such as the strike zone getting shorter and fatter, or Clemens himself for that matter).
Clemens fell off from mid-’91, when he stopped getting the strike call on his little 0-1/1-1 curve. He didn’t get his mojo back till he went to Toronto and picked up another secondary pitch.
It would definitely seem intuitively like taller catchers would tend to frame better in the top of the zone, there seems to be no correlation based on this data. Top 20 catchers by high-low% average 73.3 inches, bottom 20 are at 73 exactly.
I wonder if that would change at all if we just focused on torso height. I’m exactly 6’0″, but that’s with the torso of a 6’4″ guy and the legs of a 5’8″ guy, so chances are I’d be setting up my glove a bit higher than another 6’0″ guy with longer legs and a shorter torso.
Or maybe I’m the only person built this weirdly.
You’re not the only one. I’m 5’6″ with 5’10” torso and 5’2″ legs. Standing up, I look really short; sitting down, I look average.
As you said, this would affect setting up as a catcher.
My guess is that most catchers are more likely to be built that way than they are to have long legs and short torso. Squatting behind the plate would be a lot tougher and create more knee/leg problems with giraffe legs unless the players legs bend in three places.
Catchers have the lowest height differential (between shortest and tallest) of all positions – even after adjusting for being the shortest position overall – indicates a more similar body type at that position than at other positions. And C weight is also higher for any given height which would tend to indicate more torso less limbs.
Joe Mauer and Matt Wieters are among the tallest catchers in MLB history. It’s fluky (true multi-year outlier) that they are both playing now.
This is a very interesting question not only with regard to pitch framing, but also throwing out baserunners. Watching Mike Piazza for much of his career, it always occurred to me that his primary challenge was getting his long legs under him and in position to release the throw. Carlton Fisk, on the other hand, had a massive torso. On his knees throwing the ball back to the pitcher, he was almost as tall as many umpires behind him.
Chris Stewart also has to be as tall as Wieters. He towered over a lot of Giants when he played here.
Interesting article. Could this have anything to do with the rotations that the catchers are working with? If a catcher is mainly working with guys who live down in the zone then one might expect them to be more proficient with dealing with low pitches.
I was thinking the same thing David . It seems like the pitchers a catcher is catching and the types of pitches they throw would definitely have an effect on the high or low strikes called .
Jeff – I looked at the BaseballSavant web site and I don’t see where he explicitly states how he is defining his pitch location zones. From the data it appears that the 1+2+3 zones may be spanning 24 inches horizontally or about 1 inch greater than the rule book strike zone, but there is no clue whether the vertical zone is adjusted by batter height, or by Pitch Fx upper and lower limits, or is hard coded to some absolute pitch height values. In any case it appears from the 85% and 77% percentages that you gave that these are zones where most if not all balls should be called strikes. If so most of the pitches in those zones are just going to be noise for any study of pitcher framing because they are going to be called strikes no matter how badly they are framed. And you also are going to be totally missing out on data of pitches just outside the zone that should rightfully be called balls but end up being called strikes when caught by the good framers. The zones are just not defined in such a way as to make them suitable for pitch framing analysis.
I’m not absolutely certain, but my understanding is that the zones are defined by the PITCHf/x dimensions that you see on Gameday and find in the spreadsheets. The developer might confirm if he’s reading this.
I use the zones directly defined by MLBAM PITCHf/x gameday files.
Could pitch f/x calibration issues be affecting the data? If the cameras are misaligned so that balls just above the strike zone in reality are treated as in the strike zone in pitch f/x, then it stands to reason that at the same park balls at the bottom of the zone in reality might be treated as below the zone in pitch f/x. And vice versa. This would explain why catchers who are generally good at one are bad at the other.
It’s possible. Probably more possible the further back you go, as PITCHf/x used to have more bugs than it does now.
http://m.mlb.com/video/v25626423/lucroy-discusses-pitch-framing-with-mlb-network/?query=Jonathan+lucroy
This is a great video where Lucroy discusses what makes him so great at receiving in general and receiving low strikes in particular. Anyone who has watched Lucroy behind the plate knows he gets very low and very compact and he keeps his body exceptionally still. He doesn’t jab at the ball, he lets it come to him. He makes the point that he always keeps his target low because it’s better for him to come up on a ball then go down on it because gravity will naturally carry your glove towards the dirt. I also like his point about not liking the term ‘framing’ which implies moving the ball into the zone after you’ve caught it, because I think that’s a pretty negligible part of what’s happening with the guys who are grabbing extra strikes behind the plate.
I wonder if this is something that’s exploitable by pitching coaches? Focus more on pitches along the edges that your catcher is good at framing. I guess though that it would be balanced out by game theory issues where your opponent knows to look to a specific zone.
It’s probably impossible to quantify, but I wonder what the pitch framing numbers look like for catchers who shift over inside/outside pre windup to receive the ball vs sitting dead center and stabbing at them, last second moving over, etc.
You should try removing games played at Fenway Park, where all close calls go to the home team, since that could bias the sample…
How do you adjust the Baseball Savant search so that it shows catchers and not pitchers?
Special notice: Off to the Big Game. NEIGHHH!!!!!!!!!
Now, back to Jeff’s column.
“The general trend is that the better a catcher is at receiving low strikes, the worse he is at receiving high strikes, and vice versa. ”
The ‘trend’ looks like its entirely created by the 3 data points in the top left. It looks like there’s no slope what-so-ever otherwise.
Jeff, nice article.
With respect to framing vs. catcher height, Ben Lindbergh looked at this last year here:
He found a 0.35 correlation between height and high strikes, and a -0.12 correlation between height and low strikes.
Not that it wouldn’t be worth repeating with these numbers to see what you get, but that’s probably the general ballpark of the correlations it seems.