Run value of a pitch based on its location in the strike zone.

8 months ago centris 37 comments 6 recs |

Story-email Email Printer Print

Comments

Very nice!

I’m surprised batters do that well on up and in.

Apparently AL MVP Dustin Pedroia can’t hit that pitch.

by Dan Turkenkopf on Mar 10, 2009 8:01 AM EDT reply actions 0 recs

That might be a product of older scouting ideas that say that is a difficult pitch to hit. I think that this has changed with a better understanding of the rotational swing in that it is much easier to hit any inside pitch for power. Another observation is that breaking pitches that hang from righthanders are going to end up in that zone, that is easier than batting practise.

What suprised me was that how neutral belt high middle of the plate was.

by BigJawnMize on Mar 10, 2009 8:55 AM EDT up reply actions 0 recs

Huh?

Up and in is a clear negative for the hitter.

Not afraid to nitpick

by joker24 on Mar 10, 2009 11:24 AM EDT up reply actions 0 recs

I beleive that this map is in the eyes of the pitcher

meaning that the up and in pitch results in about a -2 run value for pitchers.

vivaelbeñsheets

by vivaelpujols on Mar 10, 2009 3:25 PM EDT up reply actions 0 recs

Yeah, that was my bad

I’m so used to seeing these kind of heat charts from the batter’s perspective.

by Dan Turkenkopf on Mar 10, 2009 5:15 PM EDT up reply actions 0 recs

I'm just about positive it is the batter's perspective

Bottom right (down away) is the best place to pitch RHP vs. RHB followed by top left (up/in).

All the other graphs make sense—-down and away is the best place to pitch, down and in is bad, up and in is good if you get it there.

Not afraid to nitpick

by joker24 on Mar 12, 2009 3:05 PM EDT up reply actions 0 recs

What are you using for your heat maps?

by Dan Turkenkopf on Mar 10, 2009 8:02 AM EDT reply actions 0 recs

I did it in R, using the command filled.contour.

by centris on Mar 10, 2009 8:24 AM EDT up reply actions 0 recs

Thank you for providing that

I’ve been struggling trying to use hexbin and other scatterplot functions to get something looking like that, now I at least can pretend I know what I’m doing.

Can't get enough of the Oakland A's? Visit Oaktown Awesomer's. For further statistical analysis, Beyond the Box Score.

by iamawesomer on Mar 10, 2009 12:48 PM EDT up reply actions 0 recs

that's half the battle

by Harry Pavlidis on Mar 10, 2009 1:23 PM EDT up reply actions 0 recs

You could also try the commands image or contour in R, or get the lattice package and use levelplot or contourplot.

If you are not as comfortable with R you could do it in matlab and use the command imagesc.

by centris on Mar 10, 2009 1:34 PM EDT up reply actions 0 recs

Question

This graph is only based on hits from the batter, correct? Otherwise, shouldn’t the blue areas on the far outside have some kind of “warmer” color since those pitchers are almost always going to end up in a pitch taken, and thus a pitch closer to a walk, which has a positive run value. Well, unless your Vlad Guerrero that is.

"If Bowden was a general contractor, he'd build houses with nine bedrooms, six garages, no bathrooms, and half a roof."

by DyeLongJustice on Mar 10, 2009 10:21 AM EDT reply actions 0 recs

The graph is based on the run value of all pitches (balls in play, strikes, balls). With ‘warmer’ colors indicating a negative run value (good for the pitcher) and ‘cooler’ colors indicating a positive run value (good for the batter). The link at the bottom is to my blog which goes into a little more depth of the method.

by centris on Mar 10, 2009 10:31 AM EDT up reply actions 0 recs

So is this from the catcher's perspective or the pitcher's?

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 10, 2009 2:37 PM EDT up reply actions 0 recs

Sorry I missed this. It is from the catcher’s perspective.

by centris on Mar 16, 2009 1:35 PM EDT up reply actions 0 recs

So down and away is the best place for the pitcher to pitch.

by centris on Mar 16, 2009 1:35 PM EDT up reply actions 0 recs

Seems to confirm traditional beliefs

-Zach Sanders
MLB Notebook.com
Roto Rat.com

by mlbnotebook on Mar 19, 2009 11:34 AM EDT up reply actions 0 recs

that's poor sabermetrics

---
Juuuust a bit outside!!
http://www.rightfieldbleachers.com

by Jack Moore on Mar 21, 2009 9:47 PM EDT up reply actions 0 recs

So your saying...

…that low and away is the best pitch for hitters in the strikezone?

by BigJawnMize on Mar 10, 2009 2:55 PM EDT up reply actions 0 recs

How is it that low and away is about equal to right down the middle of the zone?

by Scottsdale on Mar 10, 2009 4:31 PM EDT reply actions 0 recs

I like how there's a small portion in between worth almost the same as a pitchfar off the edges.

Graph rec’d

WBULTV - BDale TV Prod

by bs.uf15bosox9bears23 on Mar 10, 2009 5:16 PM EDT up reply actions 0 recs

That should be low and in, right?

Low and in does seem to be a power location for a lot of batters. Although I still would have expected down the middle to be better.

by Dan Turkenkopf on Mar 10, 2009 5:16 PM EDT up reply actions 0 recs

I think you are right, but to rephrase it: Shouldn’t the pitches that are right in the middle of the strike zone be the most hittable?
The only explanation I can think of is that when a pitcher throws a strike down the middle he’s really putting as much on it as he can. However, I’m not sure I really buy that explaination

by Scottsdale on Mar 10, 2009 6:30 PM EDT up reply actions 0 recs

Or those are 3-0 pitches.

Or those are curveballs or something more difficult to hit than fastballs.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 10, 2009 6:33 PM EDT up reply actions 0 recs

I'd love to see these by count

Not afraid to nitpick

by joker24 on Mar 12, 2009 3:06 PM EDT up reply actions 0 recs

How easy would it be to select individual pitch types? Or individual pitchers or hitters?

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 10, 2009 5:27 PM EDT reply actions 0 recs

Individual pitch types is coming soon. Individual batters and hitters is tricky because you just have so much less data. If you don’t bin at all you get a really un-smooth figure (neighboring regions have very different run values), but if you bin and smooth too much you lose too much information. I am still playing around with it to make it really meaningful.

by centris on Mar 10, 2009 5:59 PM EDT up reply actions 0 recs

In addition to binning, how about a bit of regression? Zones with fewer pitches get regressed more.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 10, 2009 6:04 PM EDT up reply actions 0 recs

Individual pitch type

The analysis separated by pitch type is up on my blog.

by centris on Mar 13, 2009 3:51 PM EDT up reply actions 0 recs

Good idea I will give that a try.

by centris on Mar 11, 2009 12:40 AM EDT reply actions 0 recs

I'm trying to use filled.contour with a bunch of pitch locations

and it’s telling me that I need to have increasing x and y values

Any suggestions for an r newbie?

by Dan Turkenkopf on Mar 11, 2009 9:06 PM EDT reply actions 0 recs

Ok, I think I figured it out

I needed to get my px, pz and other value into a matrix.

It’s running now (tons of data), but it appears to have worked.

Did you bin at all for this graph?

If you did use binning, how do you get that into a format for filled.contour?

Thanks for any help and sorry for the basic questions

by Dan Turkenkopf on Mar 11, 2009 9:40 PM EDT up reply actions 0 recs

Hmm.. maybe not

I’m only getting values between 0 and 1 on both axes even though my px and pz are outside that range – so I’m assuming I’m doing something wrong.

Probably time to stop giving a running commentary of my trials with R. I’ll keep playing

by Dan Turkenkopf on Mar 11, 2009 9:51 PM EDT up reply actions 0 recs

Sehr gut, danke schon. Wie ist deine Frau?

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 11, 2009 10:21 PM EDT up reply actions 0 recs

Yeah I did actually bin the data into 2500, 50×50. So I go through put each pitch in a bin and then average the run value for all pitches in a bin, then I smooth the data out a little bit. The result is a 50×50 matrix, let’s say i call it run.value, then I enter the command:

filled.contour(x=seq(-2.5,2.5,length=5),y=(0,5,length=50),z=run.value)

by centris on Mar 13, 2009 3:57 PM EDT up reply actions 0 recs