What does John Smoltz throw?

First of all, I’d like to say thanks to MVN for inviting me to join their team and thanks to Pizza Cutter for sharing this space with me!
Those of you who aren’t familiar with me can check out my old blog for a little background on my previous writing. This summer has marked my return to sabermetrics after a long hiatus. I’ve done a few things here and there over the years and got back into fantasy baseball and vintage baseball drafting a few years back, but it’s been since college that I did any real sabermetric studies, and the information landscape has changed a lot since then.
During the 2007 season, I was introduced by Baseball Prospectus writer Dan Fox to the data gold mine which is MLB’s Enhanced Gameday website and the PITCHf/x system created by Sportvision. MLB acquired all sorts of detailed trajectory data for about half of the 700,000 pitches thrown during the season and provided it for free on the Gameday website for those who wish to download and analyze it. In August, I downloaded the data and created a database, and I’ve been fascinated ever since with the seemingly limitless avenues for inquiry. In fact, it’s been a constant tension between enhancing the capabilities of the database and spending time playing with my exciting new toy.
Mostly, I’ve been analyzing the repertoires of various pitchers, and that’s probably what I’ll start out doing here, too. There are a host of other things that can be done with the PITCHf/x data, but I’ve found that I learn the most from studying the batter-pitcher confrontation and what weapons and strategies various pitchers are bringing to that battle.
I intended to introduce myself here at MVN with a few articles on the slider–quite an interesting pitch–but I quickly found myself sidetracked, as I often do in the PITCHf/x data, by the examination of a particular pitcher’s repertoire. In this case, Atlanta Braves pitcher John Smoltz.
Smoltz has a fairly typical suite of pitches for a starting pitcher, although perhaps not your run-of-the-mill 40-year-old pitcher. Smoltz throws a fastball, slider, split-finger fastball, changeup, and curveball.
My favorite way to begin learning about a pitcher is to graph his pitch speed versus the angle at which the spin on the ball is deflecting the pitch. (You can look here for more information on how I calculate these numbers or what the PITCHf/x data fields mean.)

Smoltz Pitch Speed vs. Spin Deflection Angle
Smoltz’s fastball runs 91-95 mph and looks like a pretty garden-variety four-seam fastball, except for the fact, of course, that he can still get it up to 95 mph as a 40-year-old. Smoltz apparently is quite the athlete. He uses his fastball equally with lefties and righties, about 45% of his pitches.
To right-handed hitters, Smoltz features the slider prominently (44% of pitches), and he throws it in the 85-90 mph range. Lefties don’t see the slider nearly as often, only about 23% of the time.
Instead, to lefties, Smoltz works with the split-finger fastball and changeup. I had a bit of a hard time telling the two pitches apart, but the splitter seems to be thrown a little harder and with the spin axis tilted over a little more such that it has slightly more lateral and less vertical spin deflection than the changeup. The splitter runs 84-88 mph and is used almost exclusively to lefties (15% of pitches). The changeup runs 82-86 mph and is also used more to lefties (11%) than to righties (3%).
Finally, Smoltz mixes in a few curveballs in the 75-81 mph range, making up about 7% of his pitches.
Smoltz Vertical vs. Horizontal Pitch Movement
It’s also interesting to look at how these pitches move after they leave the pitcher’s hand until they cross the plate.

Here we can see the effect of pitch speed since gravity has more time to act on slower pitches like the curveball, causing Smoltz’s curves to drop over two feet between release and the plate, as compared to the initial straight-line path.
Another common way to look at pitch movement is to remove the effect of gravity and simply look at aerodynamic, i.e., spin-induced, deflection in two dimensions. You may have seen this type of graph if you’ve read the work of other PITCHf/x writers around the web.
Smoltz Vertical vs. Horizontal Spin Deflection
There are plenty of ways to dig deeper into this data, and we’ll get to more of that in upcoming articles, but for now, that’s a primer on the world of pitch identification using PITCHf/x data and an intro to the arsenal of John Smoltz.


5 Responses to What does John Smoltz throw?

  1. Pizza Cutter says:

    Finally, I’m not just talking to myself! Welcome to StatSpeak, Mike.

  2. Ryne Crabb says:

    Mike, love your work – welcome to MVN!

  3. Mike Fast says:

    I include postseason games in my data. I don’t know whether John Walsh does, but Josh Kalk does not. Of course, that doesn’t make a difference for Smoltz this year.
    Josh Kalk applies some data adjustments (park corrections, air density adjustments, etc.) that I don’t apply to the data, and I don’t believe John Walsh does any of those adjustments either.
    Josh Kalk has an algorithm to classify pitches automatically. He’s had several versions of this algorithm, and it looks like he has a new version in progress that appears on his query tool but not yet in his player cards. The version that’s in his player cards mis-classifies, IMO, about 30% of pitches, most notably on curveballs (calls them sliders), and also on various types of fastballs (sinker, cutter, splitter). It seems to do a pretty good job with changeups. The new version that’s in his query tool seems to handle curveballs better, but without seeing the player cards, I can’t make a good overall judgment of its effectiveness.
    I, and John Walsh as far as I know, do not use algorithms to classify pitch types. I look at the spin, velocity, and break data for the pitches on a game-by-game basis and develop some rules for separating pitch types on an individual pitcher basis. Some of these rules can be generalized to multiple pitchers, but I haven’t found anything close to a universal set of rules yet.
    John Walsh, Josh Kalk, and most of the other PITCHf/x researchers who have published on the web have presented their data by graphing vertical spin deflection, horizontal spin deflection, and pitch speed. Since that’s the tradition, I present my data that way, too, but I find that calculating the spin parameters for the pitches produces much more illuminating data, so that’s my primary means of classifying and presenting data. As far as I know, Dr. Alan Nathan and I are the only ones to present data in this fashion. (I derived my method from Alan’s work.) I also like to show the effect of gravity on pitches, and that’s not something I’ve seen anyone else do on other than a very occasional basis.
    Then there are strike zone charts, of which I didn’t show any in this article, but I think all the researchers are pretty consistent in how we present the strike zone data. The main variation comes in whether one adjusts the vertical strike zone data for the height of different batter’s zones. I don’t typically do this, and I don’t think Josh Kalk does either, but I’m not sure about that. Some other researchers do make this adjustment when presenting strike zone data.

  4. dan says:

    How is your data presented differently, if at all, than the data/plots pesented by John Walsh and Josh Kalk (mainly at THT)?

  5. dan says:

    Woah. Thanks for the explanation.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: