2012 SABR Analytics Conference: Research presentations
Here are abstracts, presenter bios and audio clips from the research presentations delivered during the inaugural SABR Analytics Conference, March 15-17, 2012.
For complete coverage of the 2012 SABR Analytics Conference, visit SABR.org/analytics/2012.
THURSDAY, MARCH 15
3:00-3:45 p.m.: RP1—Dave Studenmund, “Postseason Championship WPA: A Full Accounting”
Win Probability Added and Leverage Index have entered the vocabulary of most baseball analysts, but no one has taken the concept all the way: Championship Probability Added. Dave will combine WPA and something called Championship Leverage Index to survey the entire history of major league postseason play.
Here you'll find the biggest plays, games and series of all time, measured by their ultimate impact on the ultimate goal: winning the world championship. What you'll find will surprise you. Here's a hint: think Bill Mazeroski's 1960 Series-winning home run was the biggest hit of all time? It wasn't even the biggest hit in that game.
Dave Studenmund is the primary manager and owner of the Hardball Times. He's produced eight Hardball Times Annuals and manages Bill James' website. Before diving into baseball statistics, Dave was an executive in the health care industry for many years.
- Listen to Dave Studenmund's presentation here (MP3/intro by Dan Migala)
3:00-3:45 p.m.: RP2—Vince Gennaro, “Top 10 Value Plays for Building a Roster”
Building a roster is all about efficiently acquiring wins—whether it's buying today's wins in the form of signing a free agent or trading for a player who can make an immediate contribution, or tomorrow's wins via the draft or acquisition of a prospect. There are several proven strategies teams employ to stretch their dollars and improve their return on investment. Even high-revenue clubs are turning to value strategies to economize on payroll, particularly under the new CBA, which has onerous penalties for exceeding the luxury tax threshold.
Gennaro's presentation will discuss sources of "value" that fall into four categories: exploiting data biases, inefficient pricing of wins, optimizing timing of transactions and buying risk at the right price. He will show examples of the strategies put into practice and explain why they capitalize on systemic or structural factors that can be exploited by any team.
Vince Gennaro is the President of SABR and author of Diamond Dollars: The Economics of Winning in Baseball. He is a consultant to MLB teams, and he teaches in the graduate programs at Columbia University and Manhattanville College.
3:45-4:30 p.m.: RP3—Brian Cartwright, “Counting Defense: Extending Defensive Efficiency Rating to the Player Level”
Advanced defensive stats are broken — in the sense that sabermetrically inclined baseball fans are expressing a loss in faith of their usefulness. Too often the top providers of these metrics produce a single number out of a black box, which seemingly take years to become reliable. As the creator of The Hardball Times Forecasts, Cartwright refers to his approach as "Counting Defense". If we were starting statistical collection from scratch today, would we choose putouts, assists, errors and double plays as our only measures of defense?
Extending Bill James' Defensive Efficiency Rating down to the individual level, Cartwright has created a fact-based set of counted defensive statistics, in the same way as we have always done for batting and pitching. Every batted ball on the field of play is assigned a fielder who was most responsible for potentially putting that batter out, expanding our current method of only recording which fielder eventually retrieved the ball. We currently count how many errors a shortstop has made, but how many of those allowed a batter to reach base? How many infield hits went to that shortstop? How many groundball hits to the outfield did that shortstop have the best chance to field? A data set of this type allows us to identify specific defensive skills such as range, arm and sure handedness.
After these facts have been compiled, then we can make adjustments to account for context and bias, such park factors, which bases were occupied and the handedness of the batter, resulting in the final overall runs saved rating that has become familiar in recent years.
Brian Cartwright is now in his third year as the creator and developer of The Hardball Times Forecasts, aka Oliver. He got his start in amateur baseball, working from 1978 to 1990 as the statistician for his local college summer league in Johnstown, Pa, and the annual All-American Amateur Baseball Association National Tournament which is held there. Since more recently joining the online analytic community, he has written for SeamHeads, StatSpeak and Fangraphs, was runner up in the 2009 Baseball Prospectus Idol competition, and has consulted for a major league team. For 27 years he has worked as a photogrammetrist for Aerometric Inc, creating digital map products from 3d aerial photography for civil engineers and municipal planners.
3:45-4:30 p.m.: RP4—J.C. Bradbury, “Impact of Pitch Counts and Days Rest on Pitcher Performance”
Many individuals believe that limiting pitch counts and increasing days of rest can improve performance and reduce injuries. Though the belief that overuse can hamper pitchers is widespread, there exists little evidence that adjusting pitch counts and rest has much effect on pitcher performance. In this study, Bradbury uses newly available game-level pitch count data from 1988 to 2011 to evaluate the impact of pitch counts and rest days on future performance. He discusses his employment of linear and non-linear multiple regression analysis techniques to estimate the impact of pitch counts — in recent games and cumulatively over a season — and days of rest on pitcher performances while controlling for the effects of other factors.
J.C. Bradbury is an economist and associate professor at Kennesaw State University in metropolitan Atlanta. He is the author of The Baseball Economist and Hot Stove Economics: Understanding Baseball's Second Season. A lifelong Braves fan, he lives in Marietta, Georgia, with his wife and two daughters.
FRIDAY, MARCH 16
11:00-11:45 a.m.: RP5—Patrick Kilgo, et. al, “Bridging the Gap Between Sabermetrics and Formal Statistical Inference: Toward a Unified Approach to Baseball Research”
With so many sabermetric researchers from so many different types of backgrounds, tensions have been known to arise. The Emory team argues (as have many others) that the tension can be traced to the different approaches taken by two groups that could broadly be classified as classical sabermetricians and quantitative professionals (statisticians, mathematicians, economists, academics, etc.). Increasingly, members of these respective groups are at odds with one another concerning the validity of one another’s research and much confusion has resulted.
Their presentation will focus on the differences in the two approaches as well as the proper roles of classical sabermetric analysis and formal statistical inference, emphasizing the strengths and weaknesses of each. be reconciling the approaches so that they are viewed as complementary, not oppositional, to one another. Examples that bridge the gap between the approaches will be offered and discussed.
Patrick Kilgo is on the faculty in the Biostatistics Department at the Emory University School of Public Health in Atlanta, where he teaches graduate methods courses and performs cardiothoracic surgery clinical research. Jeff Switchenko is currently a Ph.D. student in Biostatistics at Emory. In 2006 he received a BA in Mathematics with a minor in Chemistry from Bowdoin College. Brian Schmotzer, Hillary Superak, Lisa Elon, Paul Weiss, Jason Lee and Lance Waller are Kilgo's colleagues at Emory.
11:00-11:45 a.m.: RP6—Greg Rybarczyk, “Integral Baseball: Comprehensive Performance Valuation via Player Tracking”
Emerging player and ball tracking technologies such as Sportvision's f/x products promise to provide industry insiders and fans alike with a wealth of raw data, documenting the on-field action in more detail, and with more precision than ever. However, to be useful, such data must be transformed into information that clarifies and illuminates the game of baseball to its millions of fans, information that can also guide the actions and decisions of players, managers and executives.
In his presentation, "Integral Baseball: Comprehensive Performance Valuation via Player Tracking", Rybarczyk will review findings from his prior analysis of a sampling of Sportvision's Field f/x data from the 2011 season, and the methods he used to turn that data into information. He will also propose a comprehensive system for apportioning value to all player activity on the field of play, one that is based not on the traditional counting and manipulation of "box score" stats, but on the physical movements of the players and the ball, and the change in run expectancy throughout each play that is brought about by those actions.
Greg Rybarczyk combined his interests in baseball, physics, statistics and automated spreadsheet design to create HitTrackerOnline.com in 2005, collaborating now with ESPN Stats & Information Group to track all MLB home runs. He was an officer in the U.S. Navy for seven years and has worked as a reliability engineer, design engineer and as a Six Sigma “Black Belt” and “Master Black Belt” for two major U.S. and global corporations.
3:30-4:15 p.m.: RP7—Andy Andres and Jason Lefkowitz, “Projecting NCAA Division 1 Player Performance”
There are many baseball performance projections systems that have been developed and refined since Bill James started advanced computation in this field in 1985 with his Brock2 System (The Bill James Baseball Abstract 1985). For all the projections systems that have been developed for professional baseball, we have found none that attempt to project NCAA Division One (D1) baseball performance.
Player-season data for NCAA D1 players was collected since 2002 and analyzed for the following:
1) Season to season variation in the run scoring environment over all of D1 baseball. Recent bat changes from the BESR standard to the BBCOR standard have made the need for such an adjustment obvious.
2) Rudimentary park factors were determined for each D1 home field.
3) Adjustments for the class (freshman, sophomore, junior and senior seasons) of the D1 player were determined, similar to the adjustments made for age (the aging curve) in professional baseball.
Knowing the yearly D1 run scoring environment, park factors, and player-class adjustments needed, Andres then applied a simple projection model using known data to look at future performance. Projecting sophomore performance only used 1 year of freshman data, projecting junior performance only used 2 previous years of data, and projecting senior performance used there years. Andres will report the accuracy of his projection model for NCAA D1 baseball performance.
Andy Andres teaches a course on sabermetrics at Tufts University, the head coach for the MIT Science of Baseball Program and a datacaster for Boston Red Sox games at Fenway Park. He is also a senior lecturer of natural science at Boston University's College of General Studies and an expert in exercise physiology.
Jason Lefkowitz has coached five years of NCAA Division I baseball, at the University of California-Santa Barbara and Brown University. He has helped recruit and develop 13 players who went on to get drafted. He has a bachelor’s degree from Loyola Marymount University and a master’s in Sport Science from the United States Sports Academy. He is the the President of SmallBallStats.com.
3:30-4:15 p.m.: RP8—Graham Goldbeck, “Pitch f/x and Hit f/x Component Aging Curves”
Player projections are a critical part of any baseball front office decision, and one of the more important factors that goes into a projection system is the aging curve. Traditionally aging curves have been based on all-encompassing stats such as WAR or broken down into individual component aging curves such as BB rate or HR rate. But now with four years of full MLB PITCHf/x and HITf/x data, we should be able to delve even further into the processes and explore why players age the way they do.
In this presentation Goldbeck will use PITCHf/x and HITf/x data to assess the aging curves for various skills of both pitchers and hitters including pitch speed/location, batted ball speed/direction, batter eye, and ability to manipulate the strike zone.
Graham Goldbeck is a Baseball Analytics Specialist at Sportvision, the company behind PITCHf/x, HITf/x, COMMANDf/x, and FIELDf/x. In the past, Graham was a writer for the website Beyond the Box Score and also worked as a baseball operations intern for the Oakland Athletics and Tampa Bay Rays.
SATURDAY, MARCH 17
10:15-11 a.m.: RP9—Dave Cameron, “Pitching Injuries and the Free Agent Market”
While pitcher injuries are some of the largest financial pitfalls that teams need to avoid, the price of pitchers with recent medical issues has decreased significantly in recent years. Has this correction gone too far? Are wounded pitchers now an opportunity to find value in the free agent market?
Dave Cameron is Managing Editor and a Senior Writer at FanGraphs.com. He was the co-founder of the U.S.S. Mariner blog and previously worked at Baseball Prospectus.
10:15-11 a.m.: RP10—John Dewan/Ben Jedlovec, “The Fielding Bible III—An Evaluation of the Ted Williams Shift”
Baseball Info Solutions owner John Dewan and analyst Ben Jedlovec will present research from their latest book, The Fielding Bible--Volume III. Their presentation will cover the implementation and effect of the increasingly popular Ted Williams shift. Their research uncovers the most frequently shifted hitters and their performance with and without the shift. In addition, they will discuss the effect of the count, pitch type, and pitch location on the hitter's pull rate and recommend adjustments the defense can make to anticipate the direction of the ball in play thus improving overall team defense.
John Dewan left his highly successful career as an actuary to become one of the founders, majority owner, and CEO of STATS, Inc. because, as he says, “Sports numbers are much more fun than insurance numbers.” Dewan sold STATS to Rupert Murdoch in 1999. He is now the owner of Baseball Info Solutions, which has developed the most in-depth database of detailed baseball information in the industry. The third volume of his highly acclaimed Fielding Bible series, focusing on defensive analytics, is hot off the presses, available in bookstores and websites now.
Ben Jedlovec is a Research Analyst for Baseball Info Solutions, where he plays a large role in the development of new data and analysis for BIS clients. He serves as a defensive analytics consultant for ESPN, contributes regularly to ESPN during the baseball season, and has published research in The Hardball Times Baseball Annual.
For complete coverage of the 2012 SABR Analytics Conference, visit SABR.org/analytics/2012.
This page was last updated February 7, 2013 at 12:43 pm MST.