The Sabermetric Revolution: Assessing the Growth of Analytics in Baseball

THE SABERMETRIC REVOLUTION

ASSESSING THE GROWTH
OF ANALYTICS IN BASEBALL

BENJAMIN BAUMER

AND

ANDREW ZIMBALIST

Copyright © 2014 University of Pennsylvania Press

All rights reserved. Except for brief quotations used
for purposes of review or scholarly citation, none of this
book may be reproduced in any form by any means without
written permission from the publisher.

Published by
University of Pennsylvania Press
Philadelphia, Pennsylvania 19104-4112
www.upenn.edu/pennpress

Printed in the United States of America
on acid-free paper

2 4 6 8 10 9 7 5 3 1

Library of Congress Cataloging-in-Publication Data

Baumer, Benjamin.

The sabermetric revolution : assessing the growth of analytics in baseball / Benjamin Baumer and Andrew Zimbalist. — 1st ed.

p. cm.

Includes bibliographical references and index.

ISBN 978-0-8122-4572-1 (hardcover : alk. paper)

1. Baseball—Statistical methods. 2. Baseball—Mathematical models. I. Zimbalist, Andrew S. II. Title.

GV877.B38 2014

796.357021—dc23

2013026520

For all the left arms that made it,
and all those that didn’t

Preface

1. Revisiting
Moneyball

2. The Growth and Application of Baseball Analytics Today

3. An Overview of Current Sabermetric Thought I: Offense

4. An Overview of Current Sabermetric Thought II: Defense, WAR, and Strategy

5. The Moneyball Diaspora

6. Analytics and the Business of Baseball

7. Estimating the Impact of Sabermetrics

Appendix

The Expected Run Matrix

Modeling the Effectiveness of Sabermetric Statistics

Modeling the Shifting Inefficiencies in MLB Labor Markets

Notes

Index

Acknowledgments

Michael Lewis wrote
Moneyball
because he fell in love with a story. The story is about how intelligent innovation (the creative use of statistical analysis) in the face of market inefficiency (the failure of all other teams to use available information productively) can overcome the unfairness of baseball economics (rich teams can buy all the best players) to enable a poor team to slay the giants. Lewis is an engaging storyteller and, along the way, introduces us to intriguing characters who carry forward the rags-to-riches plot. By the end, the story of the 2002 Oakland A’s and their general manager, Billy Beane, is so well told that we believe its portrayal of baseball history, economics, and competitive success. The result is a new Horatio Alger tale that reinforces a beloved American myth and, all the better, applies to our national pastime.

The appeal of Lewis’s
Moneyball
was sufficiently strong that Hollywood wanted a piece of the action. With a compelling script, smart direction, and the handsome Brad Pitt as Beane,
Moneyball
became part of mass culture and its perceived validity—and its legend—only grew.

This book will attempt to set the record straight on
Moneyball
and the role of “analytics” in baseball. Whether one believes Lewis’s account or not, it had a significant impact on baseball management. Following the book’s publication in 2003, team after team began to create their own analytics or sabermetric sub-departments within baseball operations. Today, over three-quarters of major league teams have individuals dedicated to performing these functions. Many teams have multiple staffers creatively parsing numbers.

In a world where the average baseball team payroll exceeds $100 million and the average team generates $250 million in revenue each year, the hiring of one, two, or three sabermetricians, at salaries ranging from $30,000 to $125,000, can practically be an afterthought. (Sabermetricians is what Bill
James called individuals who statistically analyze baseball performance, named after the Society for American Baseball Research, SABR.) Particularly, once the expectation of prospective insight and gain is in place and other teams join the movement, a team that does not hire a sabermetrician could be accused of malpractice. In baseball, much like the rest of the world, executives and managers are subject to loss aversion. Many of their actions are motivated not by which decision or investment offers the highest potential return, but by which decision will insulate them best from criticism for neglecting to follow the conventional wisdom. So, to some degree, the sabermetric wildfire in baseball is a product of group behavior or conformism.

Meanwhile, the proliferation of data on baseball performance and its extensive accessibility, as well as the emergence of myriad statistical services and practitioner websites, have imbued sabermetrics with the quality of a fad. The fact that it is a fad, much like rotisserie baseball leagues, fantasy football leagues, and video games, does not mean that it doesn’t contain some underlying validity and value. One of our tasks in this book will be to decipher what parts of baseball analytics are faddish and what parts are meritorious.

Some of the new metrics, such as the one that purports to assess fielding ability accurately (UZR), are black boxes, wherein the authors hold their method to be proprietary and will not reveal how they are calculated. The problem is that this makes the metric’s value much more difficult to evaluate. Of course, fads, like myths, are more easily perpetuated when it is not possible to shed light on their inner workings.

Here are some questions that need to be answered. What is the state of knowledge and insight that emanates from sabermetric research? How has it influenced the competitive success of teams? Does the incorporation of sabermetric insight into player evaluation and on-the-field strategy help to overcome the financial disadvantage of small market teams and, thereby, promote competitive balance in the game? Lewis’s account in
Moneyball
exudes optimism on all counts.

Beyond the rags-to-riches theme, Lewis’s story echoes another well-worn refrain in modern culture—the perception that quantification is scientific. Given that our world is increasingly dominated by the TV, the computer,
the tablet, and the smartphone—all forms of electronic communication and dependent on binary signaling—it is perhaps understandable that society genuflects before numbers and statistics. Yet the fetish of quantification well predates modern electronic communications.

Consider, for instance, the school of industrial management that was spawned by Frederick Winslow Taylor over a hundred years ago. Taylor argued that it was possible to improve worker productivity through a process that scientifically evaluated each job. This evaluation entailed, among other components, the measurement of each worker’s physical movements in the production process and use of a stopwatch to assess the optimal length of time it should take to perform each movement. On this basis, an optimal output expectation could be set for each worker and the worker’s pay could be linked, via a piece rate system, to the worker’s output.
¹
The Taylorist system was known as “scientific management” and was promulgated widely during the first decades of the twentieth century. The purported benefits of scientific management, however, proved to be spurious and the school was supplanted by another—one that emphasized the human relations of production. Thus, obsession with quantification at the expense of human relations met with failure.
²

Baseball, much more than other team sports, lends itself to measurement. The game unfolds in a restricted number of discrete plays and outcomes. When an inning begins, there are no outs and no one is on base. After one batter, there is either one out or no outs and a runner on first, second or third base, or no outs and a run will have scored. In fact, at any point in time during a game, there are twenty-four possible discrete situations. There are eight possible combinations of base runners: (1) no one on base; (2) a runner on first; (3) a runner on second; (4) a runner on third; (5) runners on first and second; (6) runners on first and third; (7) runners on second and third; (8) runners on first, second, and third. For each of these combinations of base runners, there can be either zero, one, or two outs. Eight runner alignments and three different out situations makes twenty-four discrete situations. (It is on this grid of possible situations that the run expectancy matrix, to be discussed in later chapters, is based.)

Compare that to basketball. There are virtually an infinite number of
positions on the floor where the five offensive players can be standing (or moving across). Five different players can be handling the ball.

Or, compare it to football. Each team has four downs to go ten yards. The offensive series can begin at any yard line (or half- or quarter-yard line) on the field. The eleven offensive players can align themselves in a myriad of possible formations; likewise the defense. After one play, it can be second and ten yards to go, or second and nine and a half, or second and three, or second and twelve, and so on.