Science of Baseball: The Most Deceptive Repertoire for a Pitcher, Chris Tillman

Predicting a Baseball's Path
Bahill et al. 2005
American Scientist

Imagine being at the center of the most dramatic moment in baseball. It's the bottom of the ninth inning of the seventh game of the World Series—two outs, the tying run on second, the winning run on first, and you are the batter. Everything depends on you. The trouble is: The most fearsome pitcher in baseball stands on the mound. He has an awesome assortment of pitches: fastball, change-up, curveball, slider and knuckleball. You want any advantage that you can get in predicting where each pitch will go.

With the crowd going wild and sweat pouring from your every pore, you have to concentrate on the ball that is about to be launched in your direction. You must gather as much information about the pitch as quickly as you can in order to make crucial decisions.

As we will show, you get just a few hundreds of milliseconds to figure out what kind of pitch—perhaps traveling at almost 100 miles per hour—is heading toward the plate. In that instant, you must observe the ball's spin and predict how it will move on its way to the plate. It's a daunting computational task. Luckily, we can describe a few clues for you to use. And you will need them soon, because that fearsome pitcher is rocking back on his pivot leg. In a split second, his arm will swing through a great arc and send a baseball hurtling your way.

Usually, we stick to peer-reviewed original research or, when feeling a bit playful, a thesis.  However, the above article comes from a science magazine and is a piece that is too often forgotten.  The article considers pitch movement, how a pitch chances velocity, and other factors a pitcher can use to deceive hitters.  However, my focus will be on the pitch repertoire they suggest.

4S on left, 2S on right

Much of this is already known, but it is good to get actually empirical data on seam flickers with pitches.  Briefly, what they found was that a four seam fastball and a curveball appear the same.  They look white with a slight red tinge.  Meanwhile, a two seam fastball has two lines on it due to the back seam rolling over and over.  This suggests that batters should have more difficulty discerning differences between a four seam fastball and a curveball as opposed to a two seam fastball and a curveball.  As such, we should expect a larger proportion of pitchers utilizing the four seam fastball / curveball combination and that this pitch selection probably results in a more successful pitcher.

A mix of a four seam fastball and a curveball is actually a rather common one and a pairing that is utilized by many successful pitchers (a third of all starting pitchers who had enough innings to qualify for ERA used this approach).  In fact, 69% of pitcher who throw a four seamer more than 40% of the time also utilized a curveball more than 10% of the time.  The list of pitchers who did that is listed below (note: I combined curveball and knuckle curve designations even though the study did not considered recognition issues with knuckle curves. Locke, Minor, Tillman, and Burnett were noted as knuckle curve throwers).

	4S	CU
Shelby Miller	74	19
Jordan Zimmerman	62	12
Clayton Kershaw	61	13
AJ Griffin	59	16
Jeff Locke	58	19
Matt Harvey	58	13
Mike Minor	57	14
Lance Lynn	56	10
Jose Fernandez	53	34
Jose Quintana	53	19
John Lackey	52	10
Chris Tillman	50	17
Stephen Strasburg	49	23
Julio Teheran	47	13
AJ Burnett	46	35
Felix Doubront	46	14
Eric Stults	45	11
Justin Verlander	44	14

Far fewer starters lean on two seamers, which might be telling in and of itself.  Where the 40% or more four seam use had a group of 26 pitchers, the two seam 40% or more use has a group of twelve (15% of qualified SP).  Only Kris Medlen (44% 2S, 18% CU), Joe Saunders (42% 2S, 11% CU), and Bronzon Arroyo (41% 2S, 12% CU) utilize the curveball to any significant extent.  With such a small sample size, it is difficult to tell whether the group mark of 25% is significantly less than the 69% displayed in the 4S group using their curveballs, but it certainly looks like there is a difference between those populations.

Moving forward, there is a great analytical opportunity I hope to be able to investigate further.  We should be able to take a look at optically similar projectiles and compare that to the flight path in a ball at the last point of batter reaction, which is about a third of the way to the plate.  I would think that the more similar in position a pitch is that looks like each other, the more deceptive it will be.  Likewise, there is probably some sweetspot where dissimilar pitches that take the same path will be as confusing as similar looking projectiles with more different paths.