Brandon League has possibly been the most squeezed pitcher in baseball to date

We're dealing with a tiny tiny sample here, so we'll see if this persists, but League has already had 18 pitches within the PITCHfx strike zone called balls. Considering batters have only taken a total of 42 pitches from League within the PITCHfx strike zone on the year, that's a crazy number, and it may speak to the fallibility of human umpires who just don't know what to do with pitchers who impart tons of spin on everything they throw. A League who can consistently get the low strike is a very different pitcher from a League who cannot.

about 1 year ago Jeff Sullivan 23 comments 0 recs |

Story-email Email Printer Print

Comments

My favorite part is that we have three relievers in the top 50

and yet somehow they have all been wildly effective

by seattlebruin on May 12, 2011 12:22 PM PDT reply actions

Thanks

you can go ahead and send my BrianL award for most comments to my work address

by seattlebruin on May 12, 2011 12:40 PM PDT up reply actions

Now if the umps could just figure out a way to call swinging strikes balls against our relievers they would have it all covered.

At first glance, this would seem difficult to accomplish. However, they have shown that opposing catchers don’t have to actually tag our runners to call them out. They have proved that the ball does’t have to reach first base on a force play ahead of the runner to call him out. It doesn’t even have to be close. The home plate umpire could get the first and third base umpire involved. Think how much fun it would be to see the batter take a full blown cut at the ball only to have the home plate ump appeal to the base ump and have him give the did not go sign.

by Droid Rage on May 12, 2011 1:01 PM PDT reply actions

I have to question the accuracy of that data

Particularly since they show that most of the difference is at the top and bottom of the zone. The PITCHf/x operators are positively awful at setting the top and bottom boundaries of the zone. With a small sample size of 42 pitches, that could have a huge effect.

Winner, Beyond the Box Score 32 Predictions Contest, 2009

by Mike Fast on May 12, 2011 1:02 PM PDT reply actions

Clearly you're not looking at the numbers the right way.

Any set of numbers and description that Jeff puts up, that helps explain how this team may not suck as bad as they appear, must be accurate. In both quantity and interpretation!

by Nate_ on May 12, 2011 1:34 PM PDT up reply actions

It is interesting when you look at the Texas Leaguers data

here. There’s some disagreement between this data and the data from Baseball Analytics.

by Jeff Sullivan on May 12, 2011 1:56 PM PDT up reply actions

I prefer Joe Lefkowitz graphs for stuff like this

Since he doesn’t normalize by the sz_top and sz_bot that the PITCHf/x operator set.

Brandon League pitcher card for 2011

Joe’s graph draws the strike zone box with the bottom at 1.5 feet and the top at 3.5 feet, but the actual average called by umps is 1.75 and 3.45 feet.

Brandon League may still be getting squeezed a little bit on that basis, but it’s not nearly as bad as the Baseball Analytics numbers made it seem.

by Mike Fast on May 12, 2011 2:14 PM PDT up reply actions

Tell me if I'm interpreting this correctly

Using Lefkowitz’s plot that you posted, it looks like League is getting squeezed quite a bit by the Umps. But he uses 1.5 feet as the lower boundary, while Umps usually call it more like 1.75. If you move that lower boundary up to 1.75, about a dozen of those mis-calls disappear. Why the preference for Lefkowitz’s plotting if it’s different than what the Umps call?

Also, the top and bottom of the strike zone are going to vary based on hitter height/stance. How much does this really differ from batter-to-batter, and how many of those low borderline calls might be explained by this?

by nathaniel dawson on May 12, 2011 3:24 PM PDT up reply actions

I like Joe's plot because he plots the actual height of the pitch

You can then decide for yourself what boundary value you want to use. Trip’s graphs at TexasLeaguers, and the Baseball Analytics article presumably, use a normalized height. If you don’t like how the data was normalized (and I don’t), then you can’t make much use of the data.

Trip’s graphs (and btw, I love his site) normalize based upon the sz_top and sz_bot parameters that the PITCHf/x operators measure, supposedly based upon the individual batter’s stance. The problem is that they do a horribly inconsistent job of it, and in a non-random fashion.

Also, the top and bottom of the strike zone are going to vary based on hitter height/stance. How much does this really differ from batter-to-batter

One standard deviation for the MLB population is about one inch at the top and at the bottom boundary.

by Mike Fast on May 12, 2011 3:42 PM PDT up reply actions

Thanks.

I wouldn’t have thought that there was a huge difference in variance of strike zone height for batters, but I’m a bit surprised that it’s as low as it is. And not too surprising that PitchF/X operators would have a problem with consistency trying to record it. It seems like it would be a bit of a by-guess-and-by-golly thing.

by nathaniel dawson on May 12, 2011 4:12 PM PDT up reply actions

By "non-random"

do you mean that there’s a lot of variance by park/stringer/press box?

by marc w on May 12, 2011 10:20 PM PDT up reply actions

I don't completely understand the cause, but I can give my speculation

Certain values of sz_top and sz_bot seem to be “sticky”, as if there was a default that gets stored in a database somewhere and used when the operator doesn’t have time to make a new measurement for that pitch or that at-bat. Those “sticky” values get reset to new values from time to time in a way I don’t understand. Sometimes the “sticky” values are shared between parks and sometimes not. There also may be park/camera vantage point effects, but that’s unclear.

Here’s a chart I made for Placido Polanco a while back:

by Mike Fast on May 13, 2011 4:32 AM PDT up reply actions

This seems like a good time for a question I've had for a long time.

On what plane does PITCHf/x read the strike zone? The front edge of the plate?
I don’t know if I phrased that question properly. But I’m always curious how/where the cameras read pitches, and how/whether that differs from how umpires are trained to read pitches.

by Pete_ on May 12, 2011 1:43 PM PDT reply actions

Front edge

I’ve always assumed it’s the front edge of the plate.

This is a great question though, because this information is being used more and more to determine pitchers and umpires skill. It ought to be specified and understood.

by Nate_ on May 12, 2011 1:48 PM PDT up reply actions

PITCHf/x data reports the location at the front of the plate

by Mike Fast on May 12, 2011 2:07 PM PDT up reply actions

Umpiring has to be nearly impossible with a pitcher like League

“Buuuhhh… I donno!!”

by Pete_ on May 12, 2011 1:44 PM PDT reply actions

Another PITCHf/x question, unrelated to Brandon League issues

Is it correct to say that they don’t adjust the strike zone batter to batter like an umpire would? Richie Sexson and Chone Figgins have significantly different strike zones. When we watch MLB Gameday, are we looking at a generic-height player, or a strike zone unique to each hitter as they step to the plate?
Seems to me, if they use some generic strike zone, we have to give a little leeway up and down on the chart. There is variability in height and stance (some squatty stances, some more upright, etc.)…

by Pete_ on May 12, 2011 5:32 PM PDT reply actions

They attempt to adjust the strike zone to the height of the batter

They just do an extremely inconsistent job with it.

by Mike Fast on May 12, 2011 6:45 PM PDT up reply actions

In your opinion, could you do a better job by using the batter height data

and ignoring “stance” altogether? Or are there enough Carlos Quentins in the league to screw this up?

by marc w on May 12, 2011 10:25 PM PDT up reply actions

That is what I do

I use a percentage of batter height if I’m trying to grade umps. It will be off a little bit for the Carlos Quentins of the world, not at the bottom so much, but at the top. I have not been able to come up with a systematic way to measure that, though, because the PITCHf/x-measured top is too inconsistent to be of help, even if regressed to a percentage of batter height. At least that was true in 2008-2009. I haven’t looked at final 2010 numbers to see if they improved the quality of that data in 2010.

by Mike Fast on May 13, 2011 4:35 AM PDT up reply actions