Getting and Not Getting the Calls: Hitters
Not very long ago, I published a post titled Getting and Not Getting the Calls: Final 2012 Results. The post examined the differences between actual strikes and expected strikes for individual pitchers and teams, based on the PITCHf/x plate-discipline data available right here. The results were interesting, to me, and hopefully some of you. It makes sense that some pitchers might be more able to get bigger strike zones. It also makes sense that some catchers might be more and less able to get bigger strike zones. It wasn’t a huge surprise that the Brewers came out looking good, and that the Pirates and Mariners came out looking bad.
Well, as it happens, that same methodology can be applied to both pitchers and hitters, so we might as well check to see how the data looks for individual batsmen and groups of batsmen. It’s less obvious how a batter might end up with a bigger or smaller strike zone, relative to the expected strike zone, but that doesn’t mean there might not be anything there, and it only takes a few minutes to make all the calculations so why not just proceed, that’s what I say. Below there are tables of names and numbers.
As a quick refresher: we have raw strike counts and pitch counts. We also have zone rate, and out-of-zone swing rate, as determined by PITCHf/x. By putting the first numbers in one hand and the second numbers in the other hand and then clapping a bunch of times, we can figure out an “expected strike” count by adding zone pitches and out-of-zone swings. Then that can be compared to the actual strike count and, presto, desired results.
I’ll repeat that it’s far less clear how a batter could have an effect relative to pitchers and catchers. I’ll also note that this data might be fraught with complications since pitchers presumably work against a similar collection of strike zones over the course of a season, while batters have one PITCHf/x strike zone that might not be a whole lot like the given umpire’s strike zone. For these reasons and others, this is the less interesting of the two posts, but if we can look at the hitters then we might as well look at the hitters, and here we look at the hitters.
Note also, again, that the league average is not zero. It’s roughly five fewer actual strikes than expected strikes per 1,000 pitches. The key stat is listed as “Diff/1000″, and it refers to actual strikes minus expected strikes per 1,000 pitches. A positive number means a player or team saw more called strikes than expected, and a negative number means a player or team saw fewer called strikes than expected. Away we go now, beginning with the teams.
Table 1: Team Data
| Team | Diff/1000 |
|---|---|
| Nationals | 4 |
| Reds | 4 |
| Mets | 3 |
| Padres | 0 |
| Phillies | 0 |
| Cubs | 0 |
| Pirates | -1 |
| Braves | -2 |
| Brewers | -3 |
| Marlins | -4 |
| Diamondbacks | -4 |
| Cardinals | -5 |
| Giants | -5 |
| Blue Jays | -5 |
| Astros | -5 |
| Yankees | -6 |
| Red Sox | -6 |
| Rays | -6 |
| Angels | -7 |
| Dodgers | -7 |
| Mariners | -7 |
| Orioles | -7 |
| Royals | -7 |
| Athletics | -7 |
| Rangers | -8 |
| Rockies | -8 |
| Indians | -9 |
| Twins | -10 |
| White Sox | -11 |
| Tigers | -12 |
When looking at pitchers/catchers, the spread between the top and the bottom was about 30 strikes per 1,000 pitches. Here the spread is roughly half of that, which makes sense, because a larger spread implies more of an actual skill. I don’t know how to explain that the top 13 teams in this table are all in the National League. I also don’t know how to explain that the bottom four teams are all in the American League Central. These are just facts, presently without obvious reasons for being. Remember, a positive number here is worse for the hitters, so kudos to the Tigers for doing whatever they might have been doing. There is…there does not seem to be much here although you are free to interpret to your heart’s content.
Let’s look now at some individual players. I looked at all 459 players who batted at least 100 times this season. Our first table shows the ten top hitters who saw more strikes than expected.
Table 2: Top 10, More Strikes Than Expected
| Name | Diff/1000 |
|---|---|
| Starling Marte | 34 |
| Will Middlebrooks | 26 |
| Chris Gimenez | 26 |
| Justin Turner | 21 |
| Travis Hafner | 19 |
| Eric Sogard | 18 |
| Lucas Duda | 17 |
| Adam Rosales | 17 |
| Mike Baxter | 17 |
| Cameron Maybin | 17 |
“Whoa!” you say, “poor Starling Marte!” Indeed, the numbers suggest the Pirates’ rookie outfielder kind of got the royal screwjob. But then, (A) we don’t know if this data is truly meaningful, and (B) Marte batted just 182 times and swung often, so the sample size is limited. Over more time, probably, Marte’s numbers would’ve looked a lot more normal. But this is enough to make you wonder at least just a little bit.
Now for the other end of the spectrum.
Table 3: Top 10, Fewer Strikes Than Expected
| Name | Diff/1000 |
|---|---|
| Nick Punto | -40 |
| Lou Marson | -33 |
| Chris Coghlan | -31 |
| Chone Figgins | -29 |
| Carlos Santana | -29 |
| Ryan Roberts | -28 |
| Don Kelly | -28 |
| Donnie Murphy | -27 |
| Brooks Conrad | -27 |
| Mark DeRosa | -26 |
We’ve got a little dude ahead of the pack, which makes some intuitive sense, although Punto batted just 191 times. Marson batted just 235 times. Coghlan came in below 200 plate appearances, and Figgins also came in below 200 plate appearances. In this table, only Santana and Roberts were regulars or pseudo-regulars, meaning the other guys might well just be showing sample-size noise. There’ll be noise in the Santana and Roberts numbers, too, but just probably a little less of it. I don’t have explanations, and Chone Figgins is terrible.
That’s all I’ve got for now, although I’m curious to see if you guys find anything really meaningful. I’m a lot more interested in the pitcher/catcher numbers, myself. Click here for an Excel sheet of the individual hitter data. Or don’t, and make your own. You are your own person.
Those are all fabulous points.
And maybe a way that Billy Hamilton might actually be able to get on base more than we would normally expect. I still dont think he has the sufficient talent to make it, but at least food for thought.
Did Billy Hamilton take your girlfriend or something?
Oh, I could never get a girlfriend.
you’re really still doing this? this is lamer than fake everdiso
Dont playa hate… participate.
Can Fangraphs just block this IP already?
Where was Matt Holliday? Through the first month of the season when his numbers were awful, it seemed he’d get 2-3 bad strikes called against him per game.
The stats know better than your eyes.
Yeah, which is why I’m asking what the stats showed…
The Excel file is linked in the post. You could just look Matt Holiday up.
But here, I’ll do it for you. Holiday’s Diff/100 was -8 on 689 PA.
My eyes tell me the same thing. I actually clicked on this article specifically hoping to see where Matt Holliday ended up. I’ve never seen someone get so many strikes called against them that were off the plate, both inside and out. I think he’s generally respected around the game for being a nice guy, but he gets some awful, awful strikes called on him so he must have wronged an umpire at some point.
How do pitchers do in this statistic when they appear as hitters? I could see umpires being less likely to give the pitcher who is hitting the benefit of the doubt on a close pitch. It would also help explain the NL teams appearing at the top of the list. Because of their 4-man rotation, low pitch allotments and high offense park the Rockies probably saw fewer at bats by pitchers than other NL teams, so if this were part of the explanation their location on the lists would make sense as well.
Interesting possibility.
Off the top of my head… my guesses would be:
which batters get screwed?
-less established players
-players that seem to be hackers anyway
-players with a “bad attitude”
-”lackadasical” appearing players
-taller players
-minorities
which batters get the benefit of the doubt
-established players
-players that don’t seem to be hackers
-players with “good attitudes”
-”gritty” players
-shorter players
-white guys
someone smarter than me should be able to define some parameters for my hunches, and risk adjust it.
Just saying
OK… risk stratification categories:
“established” player… use an AB cutoff (IP for pitchers), or age. age seems less useful to me though. sometimes you have older rookies, or younger veterans. or simply if player has ever played in MLB previous to the observed season.
“hacker” use a plate discipline stat for batter. Use a “control” stat for pitchers.
“attitude” and “gritty”… not sure how you’d qualify those.
“height” seems straightforward. Use average MLB height for your cutoff.
“race” seems straightforward also.
I think “pitcher batting” would be a good category too. Interesting at any rate.
You could do all sorts of other things too… location (ballpark)… ambient temp at start of game… other weather conditions…. local start time of game… time of each individual pitch? (do they have that data?)… just about anything you care to define and have data to support.
Create a risk adjusted model, and start finding some outliers!
The data set is way to small to parse so fine.
If your response is a somewhat continuous variable, like strike percentage, then you don’t actually have to stratify over all those explanatory variables. Just include them all in a regression.
Mike Baxter made the list, he is undersized, white, disciplined, and most likely regarded as gritty, ditto for Turner. So much for your theory. Duda is the third met on the list, but he is a big man who hacks.
its great to have the data for pitchers and batters. but to really close the loop, I think you’d need the data for umpires… that’s the key.
in fact, that’s where we really need to nail down the data. because once you can start doing that… you’ll begin to find the outlier umps. and you keep the good ones, and get rid of the bad ones. at that point, I bet the umps become MUCH more willing to accept video/computer identification of all ball and strike calls.
then, we get some real democratization of our national pastime.
long past time if you ask me.
Excellent point!
Are umpires attached to each league? That could explain the difference.
not anymore, they changed to MLB umpires from specific league umpires sometime in the 90s.
In fact the last documented case of specific league umpires came in the Oscar nominated Angels in the Outfield.
Thanks for the info. That fact completely escaped me.
Do infielders get more calls than outfielders?
Do middle infielders get more calls than corner infielders?
Do CFs get more calls than corner OFs?
Do catchers get more calls than anyone?
How does batting order play into getting calls? Do leadoff batters get more or less calls?
Does the #8 batter in NL games get more or less calls?
Do guys with batting streaks get more or less calls?
Do guys whose name ends in a vowel get more or less calls?
There are SOOOOO many ways to look at the data.
How do I look at the data?
The umps are F*CKING UP calls!!!!
Its a f*cking ball, or its a f*cking strike.
It should have NOTHING to do with who threw it, or who caught it, or who was batting, or who was calling it, or if it’s a meaningless game in the dog days of August, or if it’s a key pitch in the deciding game of the World Series.
Its a f*cking ball, or its a f*cking strike.
EVERY PITCH… is war.
It is the living, beating, heart of the game.
How about we get it right?
EVERY time. For EVERY player.
It’s inevitable anyway. So what are we waiting for????
I’m no statistician, but strict randomness might be enough to explain most or all of this. The team differences, in particular, which should have less variation than that of individual players due to having much more data, strike me as very small.
Kudos to Jeff for presenting this data here and in the prior post (and not offering explanations without proof). This is interesting and thought-provoking.
why are you censoring fuck?
/end rant
again…
Games are also called differently in the final month or two of the season, when a contending team is playing. The strike zone gets even weirder when two contending teams are playing each other head-to-head, with the playoffs on the line. Games that “count” late in the year have a zone much closer to what is called in the playoffs than normal.
Now watch a meaningless game in September, say between the Royals and Indians. The strike zone gets as big as it will ever get and the umpires do everything possible to speed the games up.
Do you have any evidence for this?
just wondering, will MLB ever get that Gameday or Pitch FX thing to be able to adjust for a batter’s actual height and stance?
That would be nice to have but very difficult to produce.
It does right now, depending on your source of data. However, it’s complicated because — surprise, surprise — batters aren’t 100% consistent in their stance, and adjustments for them aren’t 100% consistent from park to park and PitchFX operator to pitchFX operator. Humans! Clearly, in addition to robot umpires and robot pitchFX operators, we need robot batters.
Anybody that has played 2020 Super Baseball knows that that is a great idea.
I think it has already been documented that the strike zone is noticeably bigger in 3-0 counts than other counts and noticeably smaller in 0-2 counts. I think it is also bigger in 3-1 counts. This is an endemic problem to human umpires, it seems – too much empathy for the guy who is down.
As for how to explain the NL/AL discrepancy, I think this is obviously due to pitchers hitting. The zone gets expanded big time against pitchers, IMO, and maybe it also gets expanded against pinch hitters.
Average height of MLB players is 6 feet, 1 inches. Most umpires generally the “high strike” at about the same rate as each other, but there seems to be more variation between when an umpire calls a low strike on a batter.
Umpires can “cheat” on balls just off the corners, because the dugouts do not have a definitive angle on inside or outside pitches. However, when an umpire calls a low or a high strike, he is opening himself to infinitely more criticism from both dugouts (as a high school umpire who enforces a vertical strike zone rather than the typical horizontal one, I certainly know the how the two strike zones generate different reactions from the teams). Umpires are more likely to call a pitch just at the knee (or just below), even when the catcher’s mitt receives the ball lower to the ground than when it crossed the strike zone, if the player is a tall player such as Jayson Werth.
The reason for this is because the catcher’s mitt is farther away from the ground on taller players when the ball is caught than when a shorter player (such as Nick Punto) is at the plate.
Looking at the list, Nick Punto is 69 inches, Lou Marson 6-1 (73 inches), Coghlan 72, Figgins 68, Santana 71, Roberts 70, Kelly is 76 (the only outlier above avg height), Murphy is 70, DeRosa 73, and Conrad is 70, for an average of 71.2 inches per player.
Of the players seeing the most strikes, Marte is 6-0, Middlebrooks is 6-4, Gimenez is 6-2, Turner is 6-0, Hafner is 6-3, Sogard is 5-10, the outlier, Duda is 6-4, Rosales is 6-1, Baxter 6-0, and Maybin 6-3, for an average of 73.5 inches per player, and average of 6 feet, 1.5 inches tall, well above the average height of a hitter (because the MLB avg of 6.1 includes pitchers, who are the tallest position by far.
There is certainly an umpire effect on tall/short players, and my guess is the low strike.
Pardon the typos; I have been drinking for too long.
Nice effort. I want some of you’re drinking.
Outstanding post!
Am I the only one left who wants baseball to keep looking like baseball? If we go to an electronic strike zone, having umpires is really not even needed. Use cameras to make the calls at the bases and we will just watch the scoreboard for all of the calls like a video game. I can’t imagine the game without umpires, even ones who blow calls.
I understand that everyone should play by the same rules and we want the calls to be accurate but the quality of umpiring in the game is very good. There will always be mistakes and the ones that are made are very highly visible and publicized because of all of the replay discussion, but overall I think the umpires are doing a fantastic job.
The umpires have to be there to call plays at the plate, tags on the bases, etc. We don’t replace the umpires with technology. We use technology to help the umpires. The ump behind the plate can have a little cruciform of LEDs in his mask that indicate high/low/outside/inside or strike, getting its information instantaneously from the PitchFX system. He can make whatever call he wants, but he’s on notice as to what PitchFX says about the pitch. Now, sometimes PitchFX may be wonky or downright broken (it happens) but most of the time it’s telling him what he should be calling. And over time, the umpires will generally conform to that. The result is more consistency across umpires and fewer completely blown strike/ball calls.
Combine that with a fifth (or in the postseason, seventh) umpire in the booth with access to all video and a closed audio link to the crew chief on the field to whisper in his ear whenever the umps need to have a little “conference” to get a call right, and you pretty much have my ideal system. No humans eliminated (in fact, one is added) but far fewer umpiring mistakes to take the result of the game out of the hands of the players.
The only instant replay system that I have heard that I did not hate (and actually agree with) is the one you mentioned with adding another umpire to review the calls in the booth with some sort of direct communication to the crew chief on the field. This is a way that we can improve the quality of the calls without changing the look or feel of the game on the field so I agree with you on this point.
I would not like to see a system like the NFL put into play with challenges. This would slow down the game a ton (and would probably lead to people pushing harder for pitch timers etc which would just be terrible, but that is a different argument). I hate even now to see umpires go into the tunnels to look at fair/foul home runs. It takes tension away from a great moment and its just crazy. I refer you to the Michael Morse grand slam complete with re-routing to bases and a phantom re-swing. Sorry, I don’t have a link… YouTube it.
Finally, for the balls and strikes, adding technology like Pitch f/x goggles into the umpires mask is a bit much. I have never been much of a pitcher. In fact, I have been more of a Vlad Guerrero, swing at everything I can reach, kind of hitter. I do know, however, that there are pitches that are thrown that are consistently called strikes which Pitch f/x would call balls. I do not know the details, but I think this would take some weapons and deception away from the pitcher. I am sure that most people on here would agree that this is a good thing, however if we are going to whine and complain about called 3rd strikes, batters just need to go back to the little league “too close to take” philosophy.
I understand that I am not advocating for the correct calls all of the time philosophy because I am a huge purist and I hate the thought of technology coming into a game like this. I am probably wrong, but these are my thoughts and I was just curious to know if I was the only one!
To address how the eye-in-the-sky umpire would address Pitch F/x, have him monitor the calls and then between innings just relay down “Hey ________, you really have to start giving him that low strike, the pitch has been there.” This would correct issues and improve the zone as we go without creating bionic umpires.
Closer monitoring of the umpires with actual warnings and even suspensions for consistently poor calling would also help. Some umpires’ strike zones look more like Pam Anderson than the strike zone in the rulebook.
A lot of rules also need re-writing with clearer language and without contradictions, e.g. balk, infield fly.
The team data is probably skewed by the opponent catchers. This would explain why divisions (except NL West) seem to group in the data.
Interesting hypothesis. Some genius (not I0 should prove or disprove this.
*(not I)
What is the AL Central doing?
Really like this stuff. One thing I noticed, for the 459 hitters in the spreadsheet there are fewer strikes than expected strikes. Why is this? If we looked at pitchers batting would there be far more strikes than expected?
Interesting analysis!
One possible mitigating factor might be a larger strike zone on getaway games (usually midweek day games, where both teams and presumably the umpires have planes to catch). I don’t have any empirical evidence that an umpire’s strike zone becomes larger when everyone has a flight to catch, but consider this getaway game (a night game that was further delayed 3.5 hours by rain):
http://sports.yahoo.com/mlb/boxscore?gid=320917116
There were a total of 23 strikeouts. Yes, it was the Cubs and Pirates, but neither Kevin Correia nor Travis Wood have ever been mistaken for Nolan Ryan.
Surprised no David Murphy. He constantly gets pitches 6 inches off the outside corner called against him.
Interested in regression or decision tree to see how we can segment these… Does a batter’s swinging percentage affect the borderline calls?