Butterflies are not bullets
2
|
The Hardball Times found this 11/27/2007 on www.hardballtimes.com [flag] |
Tags:
MLB
Comments (13)
-
-
walshj58 thanks, Tom.
I'm still showing "spin-induced movement" in my plots. I've gotten a little lax with terminology, though, and have been calling it both "movement" and (sometimes) "break".
You can always tell if gravity is included -- you'd never see a positive vertical break in that case.
Of course, all of Wakefield's pitches drop a lot due to gravity, since they are thrown so slowly.
-
tangotiger That's one of my worries about the gravity-time and non-gravity-time graphics: people won't know which one you are really talking about, especially if they come over for the first time. And using the word break to convey spin-less movement would certainly not conform to a typical's reader's definition of what break means. Break can only be what the batter visually sees. If *you* are mixing them up, imagine the poor reader!
I know I sound like a broken record here, but since you are one of the few greats working on this, what you do will be influential.
-
walshj58 I'm not sure I agree with you about the typical reader's interpretation of break. When my brother tosses me a ball in the back yard, does it break? I would say "no", although it drops an awful lot due to gravity. The gravity break on such a throw might be several feet, but it doesn't really indicate the "movement" on the ball, does it?
I have this notion that, since we deal with gravity all the time, the fall of a pitch due to gravity is part of its natural path, in some sense. The spin-induced movement, on the other hand, seems to me something more beyond our normal experience, which is what makes it deceptive. For that reason, it seems to me the best way to show the differences among pitches.
That's my current thinking, anyway.
-
-
-
-
studes Superb article, John. I love the way you covered so many angles to tell the story. -
sdanne Another way to beat the knuckler:
Kevin Millar obviously played for the Red Sox for a few years, but has faced Wakefield 23 times in his career. He is 10 for 21 with a 1.048 SLG, 3 HR, and only 1 strikeout in an obviously tiny sample. In SI this year, he said that he tries to hit Wakefield by swinging for a homerun on every pitch. I dunno, maybe there's something to it.
Outside of that tid-bit, fantastic article.
-
Spinmonkey "Butterflies aren't bullets. You can't aim 'em—you just let 'em go." ... That is simply a beautiful statement about the KB. I'm suprised I haven't heard it before. -
Ike This is an interesting article, however I'm not sure if I would place a whole lot of value in the pitchF/X data for knuckleballs. Because the pitch trajectories, and therefore, the break values, are derived from a fit of a series of points to what is essentially a parabloa (constant acceleration). While this is not exact, for pitches other than knuckleballs, this does seem to be a good enough approximation to obtain useful information. But since the forces acting on a knuckleball can change very dramatically in flight, unlike spinning pitches, I'm not so sure I would put a lot of faith into the results of a fit to a parabola.
A pretty good summary of what is currently known about the physics of a knuckle ball can be found here:
http://farside.ph.utexas.edu/teaching/329/lectures/node45.html
Taking a look at some of the plots of trajectories on that page (they only show movement in the horizontal direction, but the same kinds of movement will occur in the vertical direction, but with the addition of the constant acceleration due to gravity), and it's easy to see that for some trajectories, a fit to a parabola might do OK, and for some it might not. For some of the extreme cases, I would imagine (I haven't actually tried to do these fits, but maybe I might simulate what can happen for these in the future) that all 9 parameters of the fit could be off by quite a bit. If that is indeed the case, It could render a measurement of net break virtually useless.
Anyway, I don't really mean to nitpick, because I think this was a nice article to read. It's just that when I see plots of the break of knuckleballs, I can't help but wonder if I am seeing an accurate representation of what the pitch actually did on it's way to the plate. My gut feeling is that I am not.
-
walshj58 Ike,
You raise a valid point. The key issue for determining the movements (pfx_x and pfx_z, in pitch-f/x parlance) is how well the initial position and direction of the pitch is determined. Those initial conditions will determine the trajectory of the hypothetical spinless pitch, which will then be compared to the final position, which, though determined by the fit, will be highly contrained by the final measurement points.
So, how well will the initial position and velocity be determined? Well, the cameras are measuring the position of the pitch at 60 frames per second, which for Wakefield's knuckler translates to about 6 or 7 measurements every 10 feet. So, looking at the plot you linked to, I estimate that we'll have 3 or 4 measurement points in the first 5 feet of the pitch, before it's really had a chance to do any knuckling. I think those points will do a decent job of determining the initial conditions.
Of course, one would have to do more work to actually demonstrate this, but my intuition tells me that, although not perfect, we can still learn a lot about knuckleballs the from pitch-f/x data.
-
Ike I will agree that it would be possible for the system to make a decent approximation of the initial contidions of the pitch, but it is my understanding that the pitch-f/x system does not work in the way we might hope it to for that case. The way I understand the system, it uses all points it records in a trajectory with equal weight in the fit for the balls trajectory, and that the results of that fit determine the initial conditions that it returns. If I'm not wrong in that understanding (which I will admit that I could be), then having the 3 or 4 points at the beginning won't matter if we have another 10 or 20 points that want to force the fit to some other initial parameters. Those 10 or 20 points will win out over the 3 or 4 every time.
I do not disagree that the we can still learn a lot about knuckleballs from this data. I only point this out so that appropriate grains of salt may be applied.
-
-
-
Nathaniel Dawson I've always thought the best approach when facing a knuckleballer is to lay off the knuckleball all the time until there are 2 strikes (when you know it's coming and you have to go after it then). Make the guy throw strikes consistenly with it before you go whacking away at it. If a whole team was facing a pitcher with this approach, they'd get into a whole lot of hitters counts, and he'd be forced to throw his weak-ass fastball or give up a lot of walks.-
walshj58 I'm not sure I agree. Based on the second table in my article, the knuckleball is not any wilder than an average mlb pitch. I would say that Wakefield does consistently throw it for strikes.
Of course, when he does struggle with control, his backup solutions (the FB and CB) are not very good, that's true.
-
-
Spinmonkey Hopefully this isn't a stupid question and I understand that mathematics simulates the KB very well but has anybody ever setup a super-slo-mo camera with say, a Plexiglas shield and had somebody like Wakefield throw 80 pitches right at the lens to record the ball in flight? Do this in a controlled environment to simulate different air densities, etc. This way they could gather a good sampling of data and have the visuals to back it up. Let me know if this is just ridiculous.
Links (2)
Knuckleballs
Published 11/27/2007 by Tangotiger (tangotiger@yahoo.com) at THE BOOK--Playing The Percentages In Baseball
... John Walsh is back with another PITCHf/x article, this time focusing on Tim Wakefield. Note that all of his graphics refer to the spin-induced movements of the pitch (i.e., gravity and time are not part of the equation). What is interesting about the knuckleball is that it actually changes its spin rotation or axis on its flight toward the batter. So, what John is reporting as “fixed” spin-induced (gravity-less, time-less) movement is really its equivalent. That is, those points he is showing is the equivalent of ...
Ex-White Sox GM Ron Schueler to the Giants?
Published 11/27/2007 by thewizardsofoz <info@southsidesox.com> at South Side Sox
... ***** In other news, The 2008 Hall of Fame vote was released and Tim "The Rock" Raines is among the first-timers. Listen to Jim: Make it Raines. Yasuhiko Yabuta should finalize his 2-year, $5-6 million contract with the Royals this weekend. PMR numbers for pitchers. A Pitch-F/x article on Tim Wakefield: Butterflies are not bullets, and another on Joba Chamberlain: Swinging at Shoe Tops?. Thanks to Tangotiger for the links.
John, great as usual.
The last graphic, is that the "break" (includes gravity and time) or the "spin-induced movement" (like the earlier graphic)?
You talk about break, and it sounds like it in that part of the article. If so, I suggest you use "break" instead of "movement" for that graphic. If not, then you definitely need to use break, since the batter doesn't care how the pitch was released, just its behaviour when he sees it.
Tom