Overview
Pitching is part art and part science. Pitching analysis is no different. At Brooks Baseball, we strive to present the most complete, accurate, and comprehensive dataset about pitching available on the web. To do that, we start with the PITCHf/x dataset, a public resource made available by MLBAM and Sportvision, and make systematic changes that improve the quality, usefulness, and useability of that data.
Pitch Classifications by Pitch Info
The PITCHf/x data downloaded from MLBAM contains a set of pitch classifications determined by an automated, real-time, neural network algorithm. While this has the advantage of providing very fast labels that can inform broadcasts and products such as MLB Gameday, such automated classifiers have difficulty with certain repertoires and pitch types.
The Pitch Classifications used by Brooks Baseball are manually reviewed by Pitch Info using several parameters of each pitch's trajectory and double-checked against several other sources, such as video evidence (e.g., pitcher grip and catcher signs) and direct communication with on-field personnel (e.g., pitching coaches, catchers, and the pitchers themselves).
Realistic Trajectory Data
Because the PITCHf/x dataset provides the details of each pitch's trajectory, we can slightly alter the default values to align better with reality than what is commonly reported. For example, we slightly shift our reported values back to a release distance of 55' - which more closely reflects the actual release distance of most pitchers - so that release points are more tightly clustered and velocities are slightly increased. We also provide options to visualize pitch movement with gravity added back into the equation or with the effects of air drag removed.
Dan Brooks is an experimental psychologist and statistician. He lives in Tysons Corner, VA with his wife Marilyn (who still
wears her pink Sox hat proudly) and dog Teddy (who wishes they would hold
more Bark in the Park so that he could spend all game eating
peanuts off the floor). For baseball, research, media, or other questions,
he can be contacted at dan@brooksbaseball.net, or followed @brooksbaseball
Harry Pavlidis, a Chicagoan, is the founder of Pitch Info and Director of Technology for
Baseball Prospectus, where he provides a regular column featuring PITCHf/x
based scouting and analysis. Harry is a graduate of Syracuse University
with more than fifteen years experience in developing online and mobile
applications - for things not even related to baseball, too. He can be
contacted at harry@pitchinfo.com, or followed @harrypav
Dan Rozenson is a Washington-based "young professional" and graduate student in international security.
He is a research assistant for Baseball Prospectus and assures the quality of pitch classifications on Brooks Baseball.
Dan is also a writer for Beyond the Box Score on PITCHf/x and other sabermetric topics.
He can be contacted at rozenson@gwmail.gwu.edu, or followed @SixToolPlayer
Kate Morrison is a social media specialist based in the Dallas area. She assures the quality of pitch classifications on Brooks Baseball, is a guest contributor and prospect team member at Baseball Prospectus, and writes about the Texas Rangers' system for WFAA.com. She can be contacted at unlikelyfanatic@gmail.com or followed @unlikelyfanatic.