THE BOOK--Playing The Percentages In Baseball

Do fastballs rotate differently (RPM or direction) from change-ups? I.e. can spin be used a proxy for picking up speed?

I’m sure change-ups rotate at a different RPM than fastballs, but I’m not sure if that’s detectible.  I do know that one of the main reasons they’re so effective is that the direction of spin is the same or similar.

On sliders, the ball rotates around the z-axis (imaginary line from batter to pitcher), which causes the seams to form a visible dot/circle in the middle of the ball that batters like Gonzalez pick up.  If they don’t see the dot, they assume the ball is spinning on the x-axis (line from first base to third), and is therefore a fastball.

The book the Psychology of Baseball mentioned a study where batters were thrown balls with the same direction of spin at five different speeds.  Even the best hitters struggled to make contact 20% of the time.  When the machine was modified to throw at only two speeds (similar to a real-life fastball/change pitcher), the batters’ contact rates more closely reflected game performance.

Fangraphs has him positive against all pitch types except knuckleball (SSS) and Slider.

kanka: you are saying that that test concluded that if a pitcher can throw his 4 seamers (while maintaining the same rpm and spin angle) at 96, 94, 92, 90, and his change at 88, that that would be more effective than throwing his fastballs all at 96 and his change at 88?

I don’t know what my prior expectation would be of that, other than pitchers don’t seem to do that, and so I’d say that that can’t be true.

Was this tested against MLB hitters, or college hitters?

I’m sure change-ups rotate at a different RPM than fastballs, but I’m not sure if that’s detectible.  I do know that one of the main reasons they’re so effective is that the direction of spin is the same or similar.

Agreed.  Changeups do spin somewhat slower than fastballs, but not by a large amount.  Spin is picked up by the batter seeing the flicker of the seam patterns as the ball rotates, and a four-seamer or a changeup look pretty much identical in that regard.  You can’t see much of a seam pattern with the four-seamer or changeup.  A cutter is similar in that regard.

The seam pattern looks a little different on a two-seamer, but it’s subtle, so I don’t know whether batters can pick it up well or not.  I believe it was Baldwin and Bahill that demonstrated this in the lab, but picking it up in game action might be a different matter.  I have not heard anecdotally of anyone doing that.

On sliders, the ball rotates around the z-axis (imaginary line from batter to pitcher), which causes the seams to form a visible dot/circle in the middle of the ball that batters like Gonzalez pick up.  If they don’t see the dot, they assume the ball is spinning on the x-axis (line from first base to third), and is therefore a fastball.

This is the theory and what some batters have commented about, but it’s been interesting to me to watch a few sliders now from the YES-MO slow motion where you can see the seam rotation.  The “dot” does appear on a slider, but it’s nowhere near as clear as I had imagined it based upon what I had read.  It usually only shows up as a passing thing for part of the trajectory.

For a splitter, the ball spins a lot slower--a half or third as fast as a fastball--and some batters seem to be able to pick up on that.  Wade Boggs, for example, commented about this as how he could tell when his eyesight had deteriorated.

I don’t know whether a batter can pick up the topspin on a curveball.  It doesn’t have the characteristic “dot” that a slider does, and otherwise its seam pattern when rotating should not be particularly distinguishable, I don’t think.  I haven’t seen a curveball yet on YES-MO so that I can verify that.  What I have heard is that batters pick up the curveball because the pitcher has to launch the curve going up in order for it land in the strike zone.

Great interview by David, as usual, btw.

Mike/3: that wasn’t what I was asking.

You have to be careful when you interpret those numbers.  I don’t particularly like how those numbers are shown anyway.

The question I am asking is Adrian is more successful against one type of pitcher than another type of pitcher (not the pitch… the pitcher).  And of course, it would be relative to all hitters against those types of pitchers.

Tango/4: The machine in the book varied even more than that.  I believe the gap was somewhere between 5 and 10 MPH.  (So it was either throwing 80, 85, 90, 95; or 65, 75, 85, 95.)

The study used batters with varying levels of experience, even people who have never hit before.  I can’t remember exactly, but I want to say the best hitters either had college experience or maybe even as high as AA experience.

As far as why pitchers don’t do that: I’m sure it’s difficult enough to throw two different pitches (a four-seamer and a change-up) with the same mechanics and release point, much less throw four different pitches that vary in speed that greatly.

Yes, this interview is a great example of how scouting and performance analysis meet.  Both sides are after the same thing, but just viewed from different angles.  And David draws that out with his questions.

The standard gasbags who ask questions, even from former players (which is discouraging) simply miss the point completely.  “How do you feel” or variations thereof are the most banal and useless of all questions.  Darren Pang, a former NHL goalie, and now the guy who does the on-ice interviews, always asks that question.  Can’t he ask a question based on the fact that he used to be an NHL player, and so has insights to ask probing questions?

What is it that David Laurilia has that the former players don’t have?  Maybe he can teach them…

"On sliders, the ball rotates around the z-axis (imaginary line from batter to pitcher), which causes the seams to form a visible dot/circle in the middle of the ball that batters like Gonzalez pick up.”

Sliders do not rotate around the z axis.  If a ball were to rotate around the z axis, it would not break at all.  That is “gyro” spin, which produces no magnus force at all.  In order for a ball to break, the axis of rotation has to be at some non-zero angle to the direction of movement.

A slider is a tight curve ball (and faster of course).  It breaks down and away, which means that the spin is forward like a curve ball, and the axis is between horizontal (like an overhand curve ball) and vertical (like a sidearm curveball).

MGL/9, for all intents and purposes, kanka is correct about the slider.  It mostly rotates around the direction of travel, i.e., bullet spin.  For a number of pitchers, e.g., Francisco Liriano, it’s pretty close to exactly bullet spin on the slider.  Other pitchers have a few inches of break up or down and to the glove side, but the vast majority of the movement on a slider comes from gravity.

I’m not sure what “mostly” means, but are you saying that Liriano’s slider has virtually no or no horizontal movement? I’ve never seen a slider that did not break down more than gravity and away from the pitcher’s arm side.  But you are the pitch f/x expert.

A ball thrown with “bullet spin” would have no magnus force acting on it, therefore it would break like a splitter or overhand change up, but with more downward break (since it has no back spin at all to force it up) - basically it should drop exactly the same distance as a spinless ball…

MGL - Mike’s “mostly” means that a ball doesn’t have a point like a bullet so a pitcher’s slider doesn’t always have its spin axis perfectly aligned with its line of travel.  Also, because the pitcher is on a mound and his release point is necessarily higher and to one side of where he wants the ball to end up, he is going to throw a pitch with velocity vectors in both the Z and X directions as well as the Y direction toward the plate.  So there will be some small magnus induced movements induced by the balls spin axis not being perfectly aligned with those X and Z axis velocity vectors.  But compared to any other type of pitched ball the slider will have “mostly” no magnus induced movement.

What is the average or a typical horizontal (compared to a no-spin pitch of course) and vertical movement of a slider, according to pitch f/x?

Using MLB pitch classifications for 2009 data I calculate that the magnus force moves the average slider UP 5 inches compared to the same pitch with no spin.  Gravity creates all the sink.

Peter/12, yes, good explanation.

MGL/13, using MLBAM’s pitch classifications for 2010-2011, the average slider from a right-handed pitcher had pfx_x = +2.5 inches and pfx_z = +1.3 inches, and from a left-handed pitcher, pfx_x = -1.6 inches and pfx_z = +0.9 inches.

The PITCHf/x coordinate system has the origin at the point of home plate, the z axis pointing up, the y axis pointing toward the pitcher’s mound, and the x axis pointing toward the first base side.

The pfx_x/z parameter is the spin+drag-induced movement measured from y=40 feet to the front of home plate (y=1.4 feet).  (The drag force usually contributes around an inch or so of upward movement to pfx_z because the pitch trajectory is somewhat downward.  Similarly it can contribute an inch or so of lateral movement to pfx_x depending on the lateral component of the speed.)

Peter/14, I get an average for 2009, using MLBAM classifications, as follows:

RHP: pfx_x = +2.3 in., pfx_z = +2.3 in. (n=85348)
LHP: pfx_x = -1.5 in., pfx_z = +1.7 in. (n=33262)

Oh, Peter, are you calculating the Magnus force over the whole 53-foot trajectory from release to front of plate?

That’s certainly a legitimate way to do it, but it’s different than the “conventional” 40-foot break numbers from PITCHf/x that I reported.

Mike - Yes, I was calculating the numbers myself using a longer distance and some rough calculations.  I forgot the there were already pre calculated numbers that would more accurate than what I was doing.  Glad I got as close as I did!

I guess there is no need to quibble with the word “mostly” or “for all intents and purposes.” To me, 2 inches of break up and to the side is quite different from gyro spin and no break…

I suppose “mostly” is in the eye of the beholder.

By way of comparison, an average curveball:
RHP: pfx_x = +5.3 in., pfx_z = -5.6 in.
LHP: pfx_x = -4.0 in., pfx_z = -6.2 in.

And as I mentioned before, make the pfx_z numbers an inch more negative and the pfx_x numbers an inch closer to zero in order to remove the effect of drag.

An average fastball:
RHP: pfx_x = -5.2 in., pfx_z = +8.6 in.
LHP: pfx_x = +5.8 in., pfx_z = +9.0 in.

An average changeup:
RHP: pfx_x = -6.7 in., pfx_z = +4.0 in.
LHP: pfx_x = +7.9 in., pfx_z = +5.1 in.

I mentioned Francisco Liriano’s slider earlier.  Here are a few similar sliders in terms of average break.

Francisco Liriano: pfx_x = -0.4 in., pfx_z = +0.4 in.
Ryan Dempster: pfx_x = 1.6 in., pfx_z = +0.8 in.
Edwin Jackson: pfx_x = +0.3 in., pfx_z = +2.4 in.
Johnny Cueto: pfx_x = +1.8 in., pfx_z = +0.6 in.
Luke Gregerson: pfx_x = +1.6 in., pfx_z = 1.5 in.
Brad Bergesen: pfx_x = +1.9 in., pfx_z = -0.7 in.
C.J. Wilson: pfx_x = -1.1 in., pfx_z = +1.0 in.
Tim Hudson: pfx_x = +1.6 in., pfx_z = +0.6 in.
Aaron Harang: pfx_x = +1.9 in, pfx_z = +1.7 in.
Scott Baker: pfx_x = +1.0 in, pfx_z = +1.8 in.
Max Scherzer: pfx_x = +1.8 in., pfx_z = +0.7 in.
Jeremy Bonderman: pfx_x = +1.5 in., pfx_z = +0.3 in.
Brad Lidge: pfx_x = +0.5 in., pfx_z = +0.9 in.
Felipe Paulino: pfx_x = -0.8 in., pfx_z = -0.6 in.
Joel Hanrahan: pfx_x = +0.2 in., pfx_z = +0.7 in.

...and the list could go on of pitchers with minimal spin movement on their sliders.

A big-moving slider is one like Carlos Marmol’s:
pfx_x = +5.0 in., pfx_z = -1.9 in.

Hmmm, interesting. I did not realize that there were many pitchers who indeed have very little (spin induced) break on their sliders…