Pitch Identification Tutorial
17
|
The Hardball Times found this 9/19/2007 on www.hardballtimes.com [flag] |
Tags:
MLB
Comments (16)
-
profpepper Wonderful, as always.
One suggestion about the graphs: try to use different colors for speed. "Cold" colors for slower pitches and "Hot" ones for the fastest. R should have ready palettes: some scale from blue to red should be the best.
Max
PS: I forgot to tell you I was in Pisa a few week ago.
-
mikefast The impact gravity has on a changeup does not appear in the data that John presented (pfx_z and pfx_x parameters).
Its effect is in the full set of data, which includes the initial x,y,z coordinates and the final x,y,z coordinates of the pitches, measured at 50 feet from the point of home plate and at the front of home plate.
-
studes Thanks, Mike. What is this data, then? And why not use the "full set of data?"
Also, I agree with Prof Pepper's comments: use ROYGBIV to go from fast to slow. Makes it intuitive. Also, in the legend, it's very hard to see the colors in the circles. Don't know how to fix that.
-
mikefast The full set of data includes a lot of things. You can see the whole list at the glossary on Dr. Alan Nathan's Physics of Baseball website: http://webusers.npl.uiuc.edu/~a-nathan/pob/tracking.htm
The reason not to use the whole movement of the pitch in the x-z plane from start to finish for pitch identification is that it includes too many variables which cloud the picture. As you can see from John's article, typical break on a pitch due to spin is on the order of a foot. Since the strike zone is about two feet tall and 17 inches wide, and pitches often go above and below this, where the pitcher locates the ball has more to do with its total movement than does the break due to spin.
Then there is gravity. Gravity makes a ball drop about three feet, so again that's quite a bit bigger than the break due to spin. I'm not sure whether it makes more sense to include the difference in drop to gravity based on the speed of the pitch together with the break due to spin, or whether these effects are best looked at separately. The data from MLB makes it easiest to look at them separately.
As an aside, in the article John mentioned that a changeup would drop more due to gravity than a fastball would. Using the formula x= 0.5gt^2, we can calculate that an 80-mph pitch would drop about 8.9 inches more than a 90-mph pitch over a distance of 55 feet (42.4 inches vs. 33.5 inches).
There is a previous discussion of some of these topics on Tango's blog: http://www.insidethebook.com/ee/index.php/site/comments/clustering_pitches/-
tangotiger Another way to present it is in terms of "Reaction". Rather than reaction time, and whatever this vertical break represents, show everything based on "Reaction distance".
Let's say we agree that we care about the point the ball crosses the plate at 0.20 seconds prior to that. Then, tell me:
1. how far is the ball (reaction distance)
2. what is its vertical and horizontal position at T minus 0.20
3. what is its vertical and horizontal positon at T (moment it enters the plate plane)
(If you want to make it .25, or .30, that's fine. I'd probably choose a point where the ball is its downward phase, so a look at Zito's curveball might help here at setting a boundary.)
-
mikefast If we're trying to look at this from the pitcher's perspective in order to tell what he's trying to get the ball to do, we get the most information when we find out how hard he's throwing and what kind of spin he's putting on the ball (and secondarily his arm angle and where's he locating the pitch around the strike zone).
This vertical and horizontal break due only to the spin (neglecting gravity and strike zone location) is very well suited for understanding things from the pitcher's perspective. I actually prefer to use some equations to calculate the spin rate and spin direction and feel like I can identify and understand pitches better that way, but the basic perspective is the same--from the pitcher's point of view.
From the batter's point of view, it does not necessarily make sense to look at things this way. The movement due to gravity and strike zone location become more important, as do concepts like reaction time/distance and what happens to the ball after the batter has made his decision to swing.
So far, fewer people are looking at this data from the batter's perspective. There are probably multitude reasons for that. When our understanding is more mature from the pitcher's perspective, it will be easier to understand the data from the batter's perspective, too.
-
tangotiger Mike, excellent point. I agree with you that keeping the speed, spin (and release points) separate, so that you know what it is that the pitcher is doing.
But, a 2D presentation showing veritical and horizontal axises (axii?), and further saying it's from the "catcher's perspective", certainly belies that perspective. Even if you want to show it the mirror-image (from the pitcher's perspective), it still is confusing to show a 2D presentation as is being done. No matter what, the reader will always see a picture something as it relates to the batter.
-
-
-
-
-
-
tangotiger I concur on the colors. The red to blue scale should be used.
I doubly-concur on the WOW.
I concur on the confusion of the vertical parameter being presented, and make a note of it on my blog:
http://www.insidethebook.com/ee/index.php/site/comments/wow/
-
walshj58 Thanks to all for the kind words.
Yes, Max, I should have used a better color scheme, I will try to improve on that. Dave, I can make the circles in the legend solid, instead of hollow. That will probably help in distinguishing the different colors.
As for the vertical movement issue, Mike already said very nicely what I would have said. I've been confining myself to identifying pitches, Tango (as usual) is ahead of the curve, trying to figure out which pitches are hardest to hit. I agree with him that these plots are not so great for that. On the other hand, the 3D trajectory plots shown by Dan Fox and others, while fascinating, aren't very illuminating either, in that respect. Your idea of tracking the last .2 seconds of the pitch in 3D might be a good one, especially since the batter has already committed his swing by that time. (.2 might not be right, need to check my Adair).
Mike, I've been keeping up on your work with rotation axis and speed and it looks promising. I do think there is something to be said for pfx_x and pfx_z, though, from an intuitive standpoint.
-
JinAZ John -- this is fantastic! Thanks so much for creating and publishing figures of all those pitchers. I'm going to have a great time looking at Reds pitchers, even if just for descriptive purposes. -j -
mlentzner I printed it out at work on the color laser printer. I can't wait to get home!
Great work, John.
-
studes That Wakefield graph is hilarious. -
Eric M. Van Awesome stuff.
I agree that a second plot from the batterr's POV would complete the picture. But note that this includes not only the effects of gravity, but also of location. To make this second plot maximally informative, the best approach might well be to use the first plot to decide what pitch is what and then color-code by that rather than velocity.
An example: to my surprise, Julian Tavarez clearly throws two different types of two-seam fatsballs, his bread-and-butter sinker and a harder one that rides in much more. (Removing gravity from the plot makes it look like the two variants have the same vertical behavior, but they don't.) What we don't know from the chart is the extent to which Tavarez uses this to bust guys in on the hands, vs. to backdoor guys. The batter's POV chart would color-code all these riding 2-seamers and show you where he threw them.
Links (17)
Wow
Published 9/19/2007 by Tangotiger (tangotiger@yahoo.com) at THE BOOK--Playing The Percentages In Baseball
... The future is arriving. The pinnacle of sabermetrics is the convergence of performance analysis and scouting. Mariano Rivera is a righthanded pitcher, and for most RHP, they love RHH. Mo’s ...
links for 2007-09-19
Published 9/19/2007 by billfer at The Detroit Tiger Weblog
... Pitch Identification Tutorial — The Hardball Times ...
This is cool
Published 9/19/2007 by Adam J. Morris <info@lonestarball.com> at Lone Star Ball
... Posted on Wed Sep 19, 2007 at 04:16:50 PM EDT Really neat piece at THT on pitch identification with the Gameday fx system. And the author has put together charts to show, for individual pitchers, how much their individual pitches break and at what velocity. He breaks down several pitchers as part of the article, and attaches PDF files at the end for about 150 others. Really neat stuff...for example, with Brandon McCarthy, his fastball has almost no "tailing" action, but is among the most extreme in terms of "rising" (which, of course, is a misnomer, since fastballs don't actually "rise," as the author explains), while his curves almost all have almost the same horizontal break, but with a great deal of variation in their vertical break. C.J. Wilson, meanwhile, has quite a few pitches that are the velocity of sliders, but which have almost no horizontal or vertical break, which (from reading the arti ...
PITCH MOVEMENT
Published 9/19/2007 by The Chronicler at Chronicles of the Lads
... I'll have my Watch List Summary up later this week, hopefully, and will probably be doing some kind of season summary/postseason preview when the time comes. But today I was reading this John Walsh article over at the Hardball Times and figured I'd pass it along. ...
PITCHf/x plots for Harang and Arroyo
Published 9/20/2007 by Justin at On Baseball and the Reds
Today, John Walsh published an excellent "tutorial" on how to identify pitches based on PITCHf/x (Enhanced Gameday) data. That's really nice on its own. But on top of that, he published figures showing vertical and horizontal break of pitches for 150 pitchers! I thought I'd use this opportunity to take a look at Reds pitchers featured among those 150. ...
What Was That Pitch?
Published 9/20/2007 by Geoff Young at Knuckle Curve - Baseball News
... This is sort of, um, what’s the right word — stunning? John Walsh at Hardball Times (full disclosure: I contribute to THT) has done some terrific work with the new Enhanced Gameday pitch-f/x data that is all the rage: ...
Rounding The Bases:
Published 9/20/2007 by Andy Smith at Bugs & Cranks
... September 20th, 2007 Read amongst yourselves. What pitch was that again? [ The Hardball Times ] Mark McGwire gave Matt Holliday some valuable career advice. Before anyone gets all worked up, it was about adding a leg kick to his swing. [ ...
Rounding The Bases: Fun With Graphs
Published 9/20/2007 by Andy Smith at Bugs & Cranks
... September 20th, 2007 Read amongst yourselves. What pitch was that again? [ The Hardball Times ] Mark McGwire gave Matt Holliday some valuable career advice. Before anyone gets all worked up, it was about adding a leg kick to his swing. [ ...
Blog buffet: Texas does the Soulja Boy, Mariotti has a blog, Another ESPN F-up
Published 9/20/2007 by NDub at Construda
... . We had a record-shattering hit day because of them. The Big Lead, Awful Announcing, SPORTSbyBROOKS and Our Book of Scrap all posted to and linked about us that we're aware of and we thank them for it. We've only been blogging Construda for less than two months and hope many of the visitors who came yesterday and the carry-over ones today will continue to visit. Once again, thank you. Attention baseball detail sticklers: Here's a pitch identification tutorial backed by some science. (Hardball Times) I didn't know Donovan McNabb had a blog at Yardbarker but apparently he does, and he's ...
Friday Links (21 Sep 07)
Published 9/21/2007 by Geoff Young at Ducksnorts
... . (See also John Walsh’s Pitch Identification Tutorial at THT.) ...
Could Jared Burton be another Mariano Rivera? And other PITCHf/x stories...
Published 9/25/2007 by Justin at On Baseball and the Reds
John Walsh was incredibly gracious and this morning sent over PITCHf/x plots like those ...
Game 4 ALCS - Pitch f/x Analysis of Wakefield vs. Byrd
Published 10/17/2007 by halejon at The Mockingbird
... ...
Top 20 Sports Blog Stories of The Year
Published 11/10/2007 by airraid81 at Juiced Sports Blog*: Writing Enhanced by Flaxseed Oil
... Journalistic profile on Ron's brother Daniel by Washington Post journalist-turned-sports-blogger Dan Steinberg. The Post is one of few newspapers doing a good job with new media. 12. Hardball Times: Pitch Identification Tutorial The Hardball Times with an insightful report on how to identify various pitches. 13. Deuce of Davenport: ...
Can we classify every pitch?
Published 12/22/2007 by Mike Fast at Statistically Speaking
... for every major league pitcher with at least 100 pitches recorded by the system. Unfortunately, I’m not enough of an expert in clustering algorithms to replicate their work, and neither Joe nor Josh has made their algorithms public. How do we evaluate the effectiveness of their algorithms, or even attempt to construct a different classification method on our own? I should mention that there is another approach to pitch classification that has been applied quite successfully by John Walsh in answering a number of questions, not least in his article in ...
Special f/x
Published 1/3/2008 by Jim McLennan <info@azsnakepit.com> at AZ Snakepit
... - we'll be using his data as the basis for this series of articles. You can use the speed and break (horizontal and vertical) of a ball to work out the kind of pitch thrown - John Walsh wrote about this over at The Hardball Times: his graph for Brandon Webb is shown below: ...
A PITCHf/x primer
Published 1/14/2008 by Mike Fast at Statistically Speaking
... The Hardball Times published a pitch identification tutorial by John Walsh that is a good introduction to the general PITCHf/x topic as well as the specific topic of pitch identification. ...
Johan and the Terrible, Horrible, No Good, Very Bad Day (and a good one too)
Published 2/27/2008 by Tommy at Breaking Balls
... The two large clusters represent his fastball and his notorious change up. To more clearly identify the pitches (and with the help of the ever-informative Hardball Times), I made a chart that plots horizontal vs. vertical break of the pitch, and is color-coded by pitch speed: ...
Wow! Great stuff, John. I especially love the PDF files. Thanks.
Question: I don't understand your comment about the change. You seem to imply that the impact gravity has on a changeup doesn't appear in the data. Is that right?