Tracking pitch framing

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
This started out as a post on the CV versus BS thread, but I want to break it out to start a more general thread looking at the influence of pitch framing in real world situations.

Here's the Brooks Baseball


The pitches that are most interesting are the pink squares that are outside the strike zone (gained strikes) and the green squares inside the strike zone (lost strikes). Yesterday, Porcello gained three strikes and lost one. Here are the game situations in which that happened:

Gained strikes
- 3-0 Red Sox, Second inning, Corey Dickerson leading off, count 3-0. Porcello's pitch missed a little bit outside on the upper end of the strike zone. Big game impact here as instead of a leadoff walk the count went to 3-1 and Dickerson would eventually fly out to left.
- 5-0 Red Sox, Fourth inning, runner on second with two outs, count 3-1. Desmond Jennings batting. Porcello's pitch looks to be wide by about two inches. Minimal game impact as Jennings walks on the next pitch anyway (the called strike required Porcello to throw an additional pitch).
- 7-3 Red Sox, 7th inning, runner on second with two outs, count 0-0 against Logan Forsythe. The opening pitch to Forsythe appears to be a few inches inside. Forsythe is a career .597 OPS hitter after starting the count 0-1, and an .818 OPS hitter with a 1-0 count. He flied out to right 3 pitches later.

Lost strikes
- 5-0 Red Sox, 5th inning, one out nobody on against Kevin Keirmeier. Count 2-2. Porcello's sinker was up and in on the lefty but clearly within the zone.Game impact was minimal as Keiermeier struck out on the next pitch, (although Porcello did have to throw 1 more pitch).

I didn't include relievers in this post because they neither gained nor lost by this analysis, but I will in future posts.
 

geoduck no quahog

not particularly consistent
Lifetime Member
SoSH Member
Nov 8, 2002
13,024
Seattle, WA
PW, thanks.

After reading SoSH Baseball articles (specifically IA York's) - the "typical" blobbish strike zone (LHH / RHH) seems to differ from the PitchFX rectangular one. Does that need to be factored into the assessment?
 

iayork

Member
SoSH Member
Apr 6, 2006
639
After reading SoSH Baseball articles (specifically IA York's) - the "typical" blobbish strike zone (LHH / RHH) seems to differ from the PitchFX rectangular one. Does that need to be factored into the assessment?
I think it does, because the relevant pitch framing question is how catchers compare to each other, not to a hypothetical computer-called strike zone that doesn't actually exist. In the image from Brooks from PW, the problem is doubled because it conflates right- and left-handed batters, who have different strike zones. Without specifically checking, I think it's likely that all three "extra" strikes that PW flags would be called strikes by every umpire, for almost every catcher, because they're in areas where the de facto strike zone bulges out and in compared to a neat rectangle.

(Note that Brooks does partially adjust for reality, in that the zone they use is not the official rulebook zone that some sites use, but rather is a reality-adjusted zone that is pretty close to accurate in the vertical dimensions; but it's still not quite accurate horizontally.)
 

phenweigh

Member
SoSH Member
Aug 8, 2005
1,379
Brewster, MA
I think it does, because the relevant pitch framing question is how catchers compare to each other, not to a hypothetical computer-called strike zone that doesn't actually exist. In the image from Brooks from PW, the problem is doubled because it conflates right- and left-handed batters, who have different strike zones. Without specifically checking, I think it's likely that all three "extra" strikes that PW flags would be called strikes by every umpire, for almost every catcher, because they're in areas where the de facto strike zone bulges out and in compared to a neat rectangle.

(Note that Brooks does partially adjust for reality, in that the zone they use is not the official rulebook zone that some sites use, but rather is a reality-adjusted zone that is pretty close to accurate in the vertical dimensions; but it's still not quite accurate horizontally.)
Do any of these systems adjust the vertical dimensions based on the height of the batter?
 

Saints Rest

Well-Known Member
Lifetime Member
SoSH Member
Would it be possible (for those of us with old-guy vision) to highlight the pitches that are being called out (either pro or con) in a more notable way? Or maybe eliminate (or grey out) all the other pitches where framing did not play a part?

I'm also thinking that down the road, after half a season or more, it would be useful to see all the positive (and all the negative) pitches that a catcher framed. Does CV (or RH or BS) have a tendency to steal pitches of a certain type? More from LHP or RHP? Against LHB or RHB? Inside/outside, up/down?

Thanks for doing this; I find it fascinating and I hope my suggestions wouldn't be onerous to undertake.
 

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
I was struggling with the Brooks Baseball image to try to get it to show only the called strikes or balls, which are the only pitches that really matter for this exercise but if there's a way to do it, I couldn't figure it out. It's also a pain how much the swinging strike color looks like the called strike color. So basically right now I'm limited to what I can get from the PitchFX tool on Brooks Baseball but if people have suggestions for other ways to shake up that data I'm game to try. You can look up the pitch type too if you juggle around the web site a bit but I'm trying to keep this project down to a managable 10 minutes per game max so I'm not going to commit to doing that every time.

Kind of depressing to look at last night's game but Price definitely got hurt by some missed calls:



Lost strikes
- First inning, no one out, runner on first, Brandon Guyer missing. First pitch gets called a ball but PitchF/X has it clearly a strike, a little on the low side but right in the middle. Guyer goes from a .696 OPS hitter after 0-1 to a .894 OPS hitter after 1-0. He was HBP three pitches later, setting up the first Tampa run by moving Forsythe to second base.
- Second inning, two outs, runner on first, Logan Forsythe batting, 0-1 count. Another pitch down in the zone but clearly a strike. Minimal game impact as Forsythe strikes out three pitches later.
- Fourth inning, one out, nobody on, Kevin Keiermeier up. Price falls behind 2-0 with two legit balls and then gets squeezed on two consecutive pitches. The first is that green square at the upper right hand corner of the zone which is close but PitchFX has as a strike. The second is the lower left, more clearly a strike that gets called ball 4. Huge game impact as Keiermeier then scores on the Casali home run to get the fun started in the fourth.

Gained strikes
- Fourth inning, two out, runner on second, Desmond Jennings, count 1-0. Price gets a strike that looks a couple inches high. Jennings is 1.001 OPS career with 2-0 count, .637 with a 1-1 count, but nevertheless hits a double and chases Price from the game.

Barnes gained one strike:


- 8-5 Rays, 5th inning, 2 out, nobody on, Logan Forsythe up with a 2-0 count. Barnes pitch just barely off the plate. Forsythe goes from 1.123 OPS after 3-0 to .905 after 2-1. He walks anyway 2 pitches later then gets stranded

Cuevas gained two strikes:

Gained strikes
- Tampa up 8-7 in 7th, Logan Forsythe up (again with Forsythe!) with 2 outs and nobody on and Cuevas gets a first pitch strike call that looks just an inch low. It's the only strike Cuevas throws as Forsythe walks on 5 pitches. Forsythe is .598 after 0-1 and .818 after 1-0.
- Next batter is Guyer with 2 outs and runner on first. Cuevas gets that solidly outside strike call to make the count 0-2. Guyer OPS actually increases from .612 after 1-1 to .619 after 0-2 (SSS). He singles but Forsythe gets thrown out to end the inning.

So that's 4 gained strikes against 4 losses, but the biggest game impacts were clearly on the pitches to Kiermaier and Guyer that Price didn't get.
 
Last edited:

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
I think it does, because the relevant pitch framing question is how catchers compare to each other, not to a hypothetical computer-called strike zone that doesn't actually exist. In the image from Brooks from PW, the problem is doubled because it conflates right- and left-handed batters, who have different strike zones. Without specifically checking, I think it's likely that all three "extra" strikes that PW flags would be called strikes by every umpire, for almost every catcher, because they're in areas where the de facto strike zone bulges out and in compared to a neat rectangle.

(Note that Brooks does partially adjust for reality, in that the zone they use is not the official rulebook zone that some sites use, but rather is a reality-adjusted zone that is pretty close to accurate in the vertical dimensions; but it's still not quite accurate horizontally.)
I think you're sort of right about this but stating it too strongly. Looking through the charts for a few games my sense is that the area that is outside the legal strike zone but within the area that BB identifies as the "typical zone" (which varies depending on whether the hitter is R or L) - I'll call this area the penumbra, because we're all lawyers here, right - those pitches are getting called strikes maybe half the time, definitely not all the time. I think this area may be where a large portion of the pitch framing action is really taking place. I'll keep an eye on this factor.
 

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
Actually, maybe these are the more useful images. This only shows the called balls and called strikes and shows the typical zone distinction and has the whole game, rather than breaking down by individual pitchers. It also shows both teams data, which isn't what I was thinking originally but maybe makes an interesting comparison with what the other team is getting. Maybe I'll rely on these instead?

Here it is for the 4/20 game, against LHB:


And against RHB:

So 2 of the gained strikes on the 4/20 game were within the penumbra, the other was clearly high. Overall, we got 2 of 3 penumbra calls, maybe 2 of 4 depending on how you feel about that green square right on the line on the RHH map.

In contrast, TB gained 1 and lost 4.
 
Last edited:

simplicio

Member
SoSH Member
Apr 11, 2012
4,763
How does statcorner make their determinations of the zone? I believe they have Vaz with a net loss of 5 calls in yesterday's game (he was at 11 calls gained on the season before, which is down to 6 now), which doesn't correlate with the pitchfx data that I can see.
 

iayork

Member
SoSH Member
Apr 6, 2006
639
How does statcorner make their determinations of the zone? I believe they have Vaz with a net loss of 5 calls in yesterday's game (he was at 11 calls gained on the season before, which is down to 6 now), which doesn't correlate with the pitchfx data that I can see.
I believe StatCorner uses a similar approach to mine, though it's not identical. He doesn't identify a "zone" from the rules, but instead uses PITCHf/x data to identify where pitches are and are not called strikes. He then compares individual catchers' balls and strikes to this typical zone. He has an explanation here though I think there's a more detailed explanation somewhere else on the site. That's the same as my approach, though I do a little extra work trying to get percentages more accurately than he does, so it's not surprising we end up with similar numbers as a rule.
 

iayork

Member
SoSH Member
Apr 6, 2006
639
Do any of these systems adjust the vertical dimensions based on the height of the batter?
I don't think any do, and it turns out to be surprisingly unimportant. PITCHf/x offers information that purports to give the individualized strike zone based on batter height, but when I looked at it it turned out that apparently some batters are nearly 8 feet tall, so I ignore it (and so does everyone else, I think). Others have also found that umpires make only fairly minor changes to their zone based on batter height.
 

phenweigh

Member
SoSH Member
Aug 8, 2005
1,379
Brewster, MA
I don't think any do, and it turns out to be surprisingly unimportant. PITCHf/x offers information that purports to give the individualized strike zone based on batter height, but when I looked at it it turned out that apparently some batters are nearly 8 feet tall, so I ignore it (and so does everyone else, I think). Others have also found that umpires make only fairly minor changes to their zone based on batter height.
Thanks, and sucks for Pedroia.
 

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
Playing some catch up. I've always suspected that knuckleballers get screwed by human umps and Steven Wright certainly was on Friday. Against LHB:


And against RHB:


Lost Strikes
- Red Sox up 2-0, Inning 1, 1 out runners on first and second for Rasmus. First pitch is clearly a strike, nearly right down the middle, gets called ball 1. Rasmus goes from a .835 career after 1-0 to .654. He lines out.
- Next batter is Tyler White who also gets ball 1 on a pitch that was clearly a strike. I'm not going to bother with his SSS count based PAs but he grounds to 2nd to end the inning.
- Still 2-0 2nd inning, Kratz up with 2 outs nobody on and gets a ball 1 on a pitch that was well within the zone. Krats is .689 after 1-0 and .510 after 0-1. He gets all the way to 3-0 but Wright battles back to strike him out anyway.
- 3rd inning, runner on first with 2 outs, Colby Rasmus gets a gift ball 1. That takes him from a .654 OPS hitter to a .835 OPS hitter. He grounds out.
- 4th inning, nobody on, two outs, Carlos Gomez takes a pitch on the corner that looks in the zone. He goes from 598 to .823 career. He lines out.
- 5th inning, leading off. Luis Valbeuna walks with the help of a first pitch ball that actually caught the outside of the plate. Valbuena is .783 1-0, .614 0-1.
- 8th inning, 2-1 on Tyler White with 2 outs and nobody on, Hembree doesn't get a call at the knees. White later walks.

Gained Strikes
- Red Sox up 2-0, inning 2, Gattis leading off. Wright gets strike one on a pitch a little high. Gattis goes from .789 at 1-0 to .703 at 0-1. He later strikes out.
- 5th inning, Wright gets a makeup call on Valbuena on the 3-1 pitch, but walks him anyway.
- 5th inning, Wright again gets a generous call up in the zone on the first pitch against Gattis, who hits into a double play on the next pitch.
- 7th inning, Red Sox up 5-0, runners on 2nd and 3rd with 2 out, Jose Altuve batting. Wright throws a pitch that looks way outside and gets a called strike. Manages to get the strike despite missing well outside the zone or the penumbra and despite the fact that Hanigan can't catch it and it goes for a PB, scoring the first Houston run. Altuve later walks.

So if you're wondering, do umpires fuck up calls more with a knuckler on the mound, the answer from this game appears to be yes, and it's not surprising that at least in this game it cost Wright more than it helped him. I wonder how many strikes Wake lost over his career due to human umps. I bet it's a lot.

Steven Wright is also a factor that is going to affect Hanigan's pitch framing numbers.
 

Hoodie Sleeves

Member
SoSH Member
Nov 24, 2015
1,204
Gained strikes
- 3-0 Red Sox, Second inning, Corey Dickerson leading off, count 3-0. Porcello's pitch missed a little bit outside on the upper end of the strike zone. Big game impact here as instead of a leadoff walk the count went to 3-1 and Dickerson would eventually fly out to left..
One thing to note is that this pitch may not have anything to do with pitch framing. The defacto strike zone actually gets larger in 3 ball counts and smaller in 2 strike counts - umpires seem somewhat reticent to end at-bats with calls.
 

Rice4HOF

Member
SoSH Member
Jan 21, 2002
1,887
Calgary, Canada
It's not that umpires are reluctant to end an at bat with their call. It's that we use the batters motivation and action as an influence as to what we call. A borderline pitch on 3-0 we may call a strike because we know the batter was taking no matter what. On 0-2 we know they'll swing at anything close, so if it's the same borderline pitch and they don't swing, we second guess ourselves and think "must have been outside" and call it a ball.

There's a myth that almost any 3-0 pitch is a strike. That's not true, just that if it might have touched the paint, we are more likely to call it a strike than in a swinging situation. That is what leads to what appears to be different strike zones depending on the count.
 

iayork

Member
SoSH Member
Apr 6, 2006
639
It's not that umpires are reluctant to end an at bat with their call. It's that we use the batters motivation and action as an influence as to what we call.
I don't know what the umpires' psychological motivation is, but the difference is very real and very large. The strike zone is dramatically smaller with an 0-2 count, and dramatically larger with a 3-0 count. I wrote it up here, and this is the key image:
Red regions are areas where umpires are more likely to call strikes, blue are where they're less likely to call strikes; the key on the right side shows intensity vs probability; more explanation in the article.

I was surprised by the size and intensity of the difference.
 

Rice4HOF

Member
SoSH Member
Jan 21, 2002
1,887
Calgary, Canada
Yes, yes. I think we are agreeing on the results. I frequently refer to the summary charts on this fangraphs article that shows much the same thing. I know I call less strikes on 0-2 than in other counts. I was just disagreeing on the reasoning. Umpires do not mind ending an at bat with their call - I personally love ringing guys up on strike 3, gives me a chance to do my best Leslie Nielsen impersonation, which I'm sure is the main reason fans paid money to come to the ballpark - but when a batter takes an 0-2 pitch, we sometimes doubt ourselves and think the pitch missed the plate, whereas on another count we're more confident it's a strike and will call it so.
 

whatittakes

Member
SoSH Member
Apr 11, 2016
215
This is just anecdotal, but I thought I noticed CV pulling at least 3-4 strikes back into the zone, including the last pitch Price threw which his framing even fooled Joe Castiglione. Price had a banner night, and CV seemed to be a big part of that.
 

kieckeredinthehead

Member
SoSH Member
Jun 26, 2006
8,635
This is just anecdotal, but I thought I noticed CV pulling at least 3-4 strikes back into the zone, including the last pitch Price threw which his framing even fooled Joe Castiglione. Price had a banner night, and CV seemed to be a big part of that.
Never ascribe to Vazquez that which is adequately explained by Joe West
 

ehaz

Member
SoSH Member
Sep 30, 2007
4,948
I didn't watch the game but saw the Price K reel highlights - were the Braves pitchers getting those same inside calls?
 

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
So doing this for every game turned out to be a bit more than I can handle but here's the charts for the Price game.



Looks like 7 gained strikes (4 in penumbra) against 5 lost strikes. Overall I count 4 for 7 in penumbra. So +2 net, although some of those misses are brutal.
 

koufax37

Member
SoSH Member
Jul 20, 2005
2,472
I don't want to go too far from the world of factual analysis into the dark realm of human psychology, but there is also a confidence boost for pitchers (who trend toward delusionally overconfident when allowed to) in seeing borderline pitches presented well, even in the cases where the call isn't impacted. Much like when Tobias used the expensive shampoo and started to receive compliments. No big conclusion to draw on this yet, but muddying the water that the benefits of pitch framing will exceed the specific calls changed.
 

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
I suspect that is true too. I also wonder about how easy it is to account for the accumulative increased pitch counts from missing calls, especially if it puts you into more high-stress counts.

FWIW, taking out the time consuming game situation task, Steven Wright and co lost 7 strikes and got back 4, all but 1 in penumbra, a -3 net:


 

PrometheusWakefield

Member
SoSH Member
May 25, 2009
10,441
Boston, MA
Zimmerman at fangraphs puts the Red Sox at -9 overall net strikes for the year. Some of that is certainly Wright, but it does indicate that Swihart is probably not a plus player in this regard yet:

 

Jnai

is not worried about sex with goats
SoSH Member
Sep 15, 2007
16,123
<null>
Brooks does not adjust the zone very well in those main pages. It's just a guide.

For BP, we have a GAM model fit by pitch type for all combinations of pitcher and batter handedness that includes the count as a covariate.

We assign partial framing credit based on the calls compared to the expected value from that model, and then use a mixed effects model to assign credit to separate pitcher and catcher random effects.

For what it's worth, I think modern framing measurements are some of the most statistically rigorous, but also heavily derived measures in baseball. Grain of salt suggested, especially early in season.

Also, if there are particular visualizations at Brooks that would be useful, I can try to make them.
 
Last edited: