SABR 50 convention in Baltimore

SABR 50: Poster Presentations

Join us this summer in Baltimore to celebrate a special occasion this summer: SABR’s 50th annual convention, scheduled for August 17-21, 2022, at the Hyatt Regency Inner Harbor!

Here is the list of SABR 50 poster presentations that will be on display throughout the convention:


P01: Tad Berkery, “Grouping for Predictability: Clustering Batted Balls to Project Next-Year wOBA”

Increasing the predictability of performance is critical to the success of front offices in their evaluation process. Baseball has a well-established accounting system for batted balls in singles, doubles, triples, and home runs that is wonderfully easy to see and is also very intuitive. However, while reasonable at the box score and casual fan levels, this accounting system doesn’t maximize year-to-year predictability. To accurately project future performance, is it possible to design an accounting system that groups batted balls in a predictive manner? Analytically, can we group batted balls into clusters and assign a weight to the frequency of batted balls in each category as a means for projecting a given batter’s future weighted on-base percentage? Using Baseball Savant fields such as launch angle, exit velocity, and pitch motion data, Berkery identifies six groupings of batted balls that are highly predictive in a year-to-year context. Using a weighted average of a batter’s batted ball count in each cluster, he can explain 10% more of the variation in next-year weighted on-base percentage than current year on-base percentage alone.

Tad Berkery <tberker2@jhu.edu> is a computer science and economics major at Johns Hopkins University who will be working in the Washington Nationals front office on their Research & Development Team this summer. He has previously conducted projects for the Baltimore Ravens, Baltimore Orioles, and University of Maryland women’s softball team in the sports analytics realm and has also performed analytics in the corporate realm, spending a summer working as a Data Analyst for &pizza. Along with his interest in analytics, Tad also cares about making analytics and general ideas impactful through the art of storytelling. He is a sports writer for the Johns Hopkins News-Letter and the published author of What’s the “Right” Career?</tberker2@jhu.edu>


P02: Jaspar Carmichael, “Baiting and Ambushing in Baseball: A Game Theoretic Approach”

Carmichael explores the game-theoretic relationship between the hitter and the pitcher in a baseball game. The idea of ambushing is one that has been present in the sport for as long as the game has been around, but hasn’t been formally acknowledged. A player can theoretically deviate from their expected strategy to surprise, or “ambush”, their opponent, effectively gaining an advantage over them. For example, a pitcher may elect to throw a fastball down the middle in an 0-2 count when the hitter is expecting him to throw a “waste” pitch, catching the hitter off guard. Carmichael establishes a game-theoretic model along with a simulation environment written in Python to capture this relationship and investigate questions regarding when, how, and for how long a player should deviate to gain the largest possible advantage from “ambushing” their opponent in certain situations.

Jasper Carmichael <jcarmic6@jhu.edu> is a senior at Johns Hopkins University pursuing a degree in Applied Mathematics with a minor in Computer Science. Jasper is a member of the Sports Analytics research group headed by Dr. Anton Dahbura. Additionally, he is a member of the JHU varsity baseball team as a relief pitcher. An avid sports fan, Jasper also enjoys the outdoors, watching Marvel movies, and playing video games.</jcarmic6@jhu.edu>


P03: Alan Cohen, “The Slow Rate of Integration as Shown by Lists”

When Jackie Robinson first played for the Brooklyn Dodgers on April 15, 1947, integration of the major leagues began – or did it? Notoriously, the Boston Red Sox did not field their first players until 1959, and it was not until 1966 that ten players (yes, 10 players in 20 years) had worn the team’s uniform. As the major leagues integrated, the Negro Leagues crept along to a slow death in front of an ever-decreasing audience. From 1949 through 1962, the annual All-Star game (the East-West Game) continued to be played. These games drew upwards of 50, 000 fans to Chicago’s Comiskey Park during the years of World War II. Well under half that number came to the games after 1948. Yet, 301 men were on the team rosters for that game, and several of them made it to the major leagues, most notably Ernie Banks of the Chicago Cubs. Through lists, Cohen provides insight into the pace of integration and the players of color signed by each team.

Alan Cohen <adc0317@comcast.net> has been a SABR member since 2010. He chairs the BioProject fact-checking committee and serves as Vice President-Treasurer of the Connecticut Smoky Joe Wood Chapter. His stories have appeared in over 60 SABR publications. His 2020 contribution to the Baseball Research Journal catalogued Josh Gibson hitting a home run in virtually every big-league ballpark in which he played from 1930 through 1946. Alan helped launch the First Games Back project on the SABR website in 2020. He has four children and eight grandchildren and resides in Connecticut with wife Frances, their cats Morty, Ava, and Zoe, and their dog Buddy.</adc0317@comcast.net>


P04: Kevin Michael Doyle and Edward T. Doyle Jr., “Boats & Baseball – The 1943 Saint Augustine Coast Guard All-Stars Baseball Team”

World War II necessitated the expansion of Army, Navy and Coast Guard bases around the United States, along with the creation of numerous new bases. On these posts, Service men and women participated in on-base sports as part of their training regimen. In addition, many bases formed teams that played in off-base regional leagues, state championships and national tournaments. One such traveling Service team was the All-Stars baseball club from the Coast Guard Training Base in Saint Augustine, Florida. In 1943, the All-Stars played 37 games with civilian and military opponents in the greater Jacksonville – St. Augustine, Florida area, St. Mary, Georgia and Parris Island, South Carolina. The All-Stars included some players with minor League experience, and one pitcher who eventually reached the majors. Doyle and Doyle describe St. Augustine’s exploits, honoring these Coast Guard trainees who were preparing for service in WWII.

Kevin M. Doyle of Fort Myers, Florida, was born in 1959 and grew up in Delaware County, Pennsylvania as a Phillies and Reds fan. Kevin played organized baseball through high school. Earned Geology degrees from the University of Dayton (1981) and the University of Wisconsin-Milwaukee (1983). Afterwards he married fellow geology student, Alison Schappach. Raised two sons together while working for Unocal and Chevron in Casper, Los Angeles, London, Bangkok and Houston. He coached Little League in Bangkok! He retired from full-time work in 2016 and settled in Fort Myers. Currently teaching part-time and online for Wichita State University. Kevin formed the Fort Myers Vagabonds Vintage Base Ball Club in 2019.

Edward T. Doyle Jr. of Hockessin, Delaware, was born in 1949 and grew up in Delaware County, Pennsylvania, as a Phillies and Red Sox fan. Edward played organized baseball through high school. He earned degrees from the University of Wisconsin-Superior, West Chester University and the University of Delaware. He has been married to his wife Becky for 48 years and has two children who played/coached baseball/softball. He worked in the oil industry and owned a Safety, Health and Emergency Management company. He coached Piedmont League Baseball (Hockessin, DE). Edward is currently a Piedmont board member, Assistant State Commissioner for Delaware Babe Ruth Baseball and Mid-Atlantic Babe Ruth Hall of Fame member. He teaches community college-level Leadership and Emergency Planning.


P05: David Gordon, “Quantitative Assessment of Career Value in the Negro Leagues”

When the Hall of Fame (HOF) opened its doors to the Negro Leagues in 1971, electors relied heavily on oral history due to gaps in the statistical record. Even now, with more complete and reliable statistics, we still lack a robust objective metric to evaluate players in the Negro and traditional major leagues on a universal scale. Gordon proposes Career Value Index (CVI), based on Baseball Reference’s Wins Above Replacement (bWAR). CVI values stars who were at or near the top of their league for several years over lesser players who accrued similar career WAR via exceptional longevity. It also addresses historical opportunity inequities, such as the threefold disparity in IP per season between 19th and 21st century starting pitchers. Using Negro League statistics from the Seamheads database, which incorporates stats from pre-1920 leagues, Latin American leagues, and independent teams that faced high-quality competition, in addition to the seven MLB-certified “major” Negro Leagues, Gordon assesses career value in leagues of all shapes and sizes, allowing for comparison between Negro League players with other major leaguers, while recognizing that these Negro League records will continue to evolve as additional data comes to light.

Dr. David Gordon <gordondj811@gmail.com> is a cardiovascular epidemiologist and clinical trialist, who retired in 2016 after a distinguished 42-year career at the National Institutes of Health, where he authored or co-authored nearly 100 articles and book chapters. Since joining SABR in 2016, four of his research articles have been published in the Baseball Research Journal in 2018-21; two more articles are under review. Dr. Gordon has recently written two books, both published in 2021: Baseball Generations (Summer Game Books) and The American Cardiovascular Pandemic: A 100-Year History (McFarland Books).</gordondj811@gmail.com>


P06: Tom Hanrahan, “The Dodgers’ Unique and Incredible Accomplishment, 2018-2021”

In 2021 the Dodgers led the National League in scoring runs, and also led the NL in allowing the fewest runs. This was the fourth consecutive year the Dodgers accomplished this. Hanrahan will show how astounding this feat is, through a list and discussion of other teams that led their league in both categories in consecutive seasons, showing the rarity of this accomplishment, especially in a 15 -team lead. He will examine the odds of this happening by random chance, and the uniqueness of the accomplishment.

For 40 years, Tom Hanrahan <han60man@aol.com> has been a St. Mary’s County, Maryland resident; an employee of the US Navy; and a dedicated husband to his wife, Jan. He has been a servant to others in his local church for almost that long as well. But those credentials pale in comparison to his 55 years of being a baseball fan.</han60man@aol.com>


P07: Chuck Hildebrandt, “The Peters Principle: Multiple Rookie Seasons”

Named in honor of Gary Peters — who seemed to have risen to the level of his incompetence by struggling as a major league rookie for four seasons before finally breaking through and winning Rookie of the Year honors in his fifth — Hildebrandt discusses the phenomenon of major league players who maintained rookie status across multiple seasons, by dint of not having met the established rookie limits of 130 at bats for position players, or 50 innings pitched for pitchers. He explores several questions related to this phenomenon: What percent of rookies experience multiple rookie seasons? How has that circumstance trended across time? Which players had the most rookie seasons? Who had the best rookie breakthrough season among those with multiple rookie campaigns? (Besides Gary Peters!) How many Hall of Famers had multiple rookie seasons?

Chuck Hildebrandt <chuck.sabr@gmail.com> is a two-time Doug Pappas Award winner for his oral presentations “’Little League Home Runs’ in MLB History” (2015) and “Does Changing Leagues Affect Player Performance, and How?” (2017). Chuck has authored three articles for the Baseball Research Journal, including the cover story, “The Retroactive All-Star Game Project”, in the Spring 2015 issue. Chuck founded the Baseball and the Media Committee in 2013. A Detroit native who is a proud Tigers fan even through their too-frequent rebuilding phases, Chuck lives with his lovely wife Terrie in Chicago, where he plays in rec softball leagues while recovering from multiple knee surgeries.</chuck.sabr@gmail.com>


P08: Anthony W. Hughes, “Dave Duncan’s Revenge”

Hughes reviews pitcher performance and the Fielding Independent Pitching (FIP) construct initially described by Voros McCracken to determine if it is possible for a pitcher to consistently outperform the FIP metric. FIP assumes that balls put in play have random results (except for home runs); removing balls in play (except for home runs) rewards pitchers with higher strikeout totals. Given equal results across HRs, walks and IP, the pitcher with the higher strikeout totals will produce a lower FIP. The “pitch to contact” pitcher is “penalized” for his lower strikeout totals. Evaluators who use FIP have gravitated towards strikeout pitchers and may discount skills of contact pitchers. However, recent trends in baseball, such as the increased use of fielder shifts, may work against McCracken’s assumptions in defining FIP – that balls in play are random and the pitcher has no control over the results. Hughes evaluates pitchers using exit velocity, ball movement, ground ball and fly ball rates and pitch profiles to determine whether there are traits endemic to pitchers who consistently outperform their FIP.

Anthony Hughes <thughes9302@msn.com></thughes9302@msn.com>

is a lifelong baseball fan and retired US Air Force officer. Tony grew up with two loves, all things St Louis Cardinals and the Missouri Tigers. At least the Cardinals have been pretty good. Tony’s interests include his family, BBQ and analytics. Currently he works as an analyst/model developer supporting the USAF.


P09: Adam Korengold, “Re-Visualizing Baseball Cards and Data: New Ways of Seeing”

Baseball cards follow design trends, from photorealist approaches from the 1940s through the late 1950s and weaving through pop art, color field painting, and modern trends in computer-aided and generated graphics and design. Korengold walks through research in twentieth-century art and relates it to trends in baseball card design on the front of the baseball card. The revolution in data visualization techniques and the widespread availability of large volumes of baseball data also has made it possible to re-imagine the back of the baseball card. Statistics printed on the backs of cards have changed little since the 1950s. With new tools like Tableau and R enabling data to be presented on cards as historical line graphs or scatterplots to represent differences between players and notable performances, in a compelling and new visual way, Korengold suggests several new ways of re-conceiving statistics on the 2 ½ by 3 ½ inch space on the back of a baseball card. He expands the meaning and use cases for new kinds of baseball statistics and presentations for today’s visually-focused media environment.

Adam Korengold <akorengold30@gmail.com> has been a SABR member since 2020. He is an Analytics Lead at the National Library of Medicine in Bethesda, Maryland, and a professor in the graduate data analytics and visualization program at the Maryland Institute College of Art in Baltimore, teaching courses in case studies and visual storytelling. In addition to his work to expand data visualization practice, he creates original baseball cards combining original paintings based on vintage cards with visualizations presenting player statistics, and paintings in acrylic and gesso on vintage cards. He has also presented on the relationship between baseball card design and art/design trends to several SABR chapters.</akorengold30@gmail.com>


P10: Herm Krabbenhoft, “Sac-Fly Rules: Roadblock or Gateway to a Career .300 Average?”

“My biggest regret was letting my lifetime [batting] average drop below .300. I always felt I was a .300 hitter, and if I could change one thing, that would be it.” That quote by Mickey Mantle, who compiled a career batting average of .298, raises the historical significance of achieving a lifetime .300 batting average. The official rule for sacrifice flys — which varied over time — affects whether a player achieved a career .300 batting average. Depending on when a player played, the Sac-Fly rule could be a roadblock or a gateway to the cherished lifetime .300 batting average. When classifying players from different time periods as lifetime a .300 hitter or not, it is appropriate to hypothetically employ the same RBI flyout rule (i.e., at-bat or non-at-bat). Krabbenhoft leverages the Stathead Search Engine from Baseball-Reference, Retrosheet’s Daily Records and Play-By-Play narratives, and newspaper game accounts to identify players from 1920-1953 who would hit .300 career under the current Sac-Fly rule (established in 1954) and players from 1954-2021 who would not have hit .300 had the pre-1954 no-Sac-Fly rule been operative.

Herm Krabbenhoft <bqr9343@aol.com> is a retired research chemist who has been a SABR member since 1981. Among his numerous baseball research accomplishments are: (a) Restoring the 1912 NL Triple Crown to Heinie Zimmerman; (b) Establishing, in collaboration with Keith Carlson, David Newman, and Dixie Tourangeau, the accurate ML record for most runs scored in a single season by an individual player — Billy Hamilton, 196 runs for Philadelphia in 1894; (c) Determining the longest consecutive games on base safely streak in ML history — 84 games by Ted Williams in 1949. Herm is the author of Leadoff Batters published by McFarland in 2001. Krabbenhoft has been the recipient of three SABR Baseball Research Awards (1992, 1996, 2013).</bqr9343@aol.com>


P11: Allison R. Levin and Mitchell Daniels, “Not Your Typical Team: What MLB Can Learn from The Savannah Bananas”

Every year we hear that major league baseball is dying and that to survive, it must attract young fans and casual fans. This fear that baseball has lost its allure to all but the most diehard of fans has led MLB to experiment with controversial ideas geared towards speeding up the game with the goal of bringing in younger and casual fans. While MLB struggles to find a happy medium between satisfying both new and established fans, in the collegiate summer Coastal Plain League, the Savannah Bananas sell out every game. They have found a way to bring in young and casual fans and maintain national excitement around their team year-round—their TikTok account has more followers than any major league team. Much of their popularity stems from the Bananas’ antics before, after and during games. Levin and Daniels examine the underlying rationale of why the Bananas are successful and explore how the marketing tactics can be adapted and adopted to attract new fans who love the game. Levin and Daniels provide examples of how MLB can tap into a version of the game that resonates with both traditional fans and younger and more casual audiences.

Allison R. Levin (MA, JD) <allison.levin@gmail.com> is a Professor of Sports Communication at Webster University. whose research centers around social/cultural issues of sports fandom, particularly baseball. She is a SABR Board Member and is Vice President of the St. Louis Bob Broeg Chapter.</allison.levin@gmail.com>

Mitchell Daniels <daniels.mitchell99@gmail.com> is a senior Sports Entertainment and Management major at Webster University and a member of the baseball team.</daniels.mitchell99@gmail.com>


P12: Dale Mahoney, “Was Shohei Ohtani’s 2021 Season the Best Two-Way Season Ever?”

Shohei Ohtani had a tremendous 2021, hitting .257/ .372/.592 for an OPS of .965, an OPS+ of 158, and a Baseball Reference Wins Above Replacement (bWAR) due to hitting of 5.1. He was the starting pitcher in 23 games with an ERA of 3.18, an ERA+ of 141, and a pitching bWAR of 4.0, for an overall bWAR of 9.1. He won the American League MVP unanimously, the first since Bryce Harper in the NL in 2015 and fellow Angel teammate Mike Trout in 2014. Only 18 other players in history have been voted unanimous MVP. He was the most valuable player in the AL by 1.8 WAR, and the best in MLB by 1.4 WAR, the largest margin. But how great was it all-time? Was it the best two-way season in history? Mahoney investigates if it was, using a variety of methods based on bWAR.

Dale Mahoney <dalemahoney@gmail.com> has been a SABR member since his first SABR convention in Denver at SABR 33 in 2003, which got him hooked. He has been engaged in sabermetrics ever since sabermetrics hit the Internet in the mid-1990s.</dalemahoney@gmail.com>


P13: Bill Marston, “I’ll Do It Myself – The Stories Behind Unassisted Triple Plays”

One of the most unusual plays in baseball, an unassisted triple play has occurred only fifteen times in the World Series era of Major League Baseball, with a possible sixteenth one occurring in the nineteenth century. That’s fewer than other rare and notable defensive achievements such as immaculate innings, four strikeout innings, and perfect games. Marston reviews a few of those plays starting with the controversial triple play in 1878 by outfielder Paul Hines. He also describes the first one turned in the twentieth century by Neal Ball; the only one to occur in the World Series, turned by Bill Wambsganss; and the most recent one turned by Eric Bruntlett in 2009. Fun facts about unassisted triple plays will also be shared.

Bill Marston <billm2126@gmail.com> is a longtime baseball player and coach, and an avid St. Louis Cardinals fan. As a lifelong math teacher, his love of statistics was put toward understanding the game of baseball better. Before advanced metrics became popular, he tried to develop a statistical tool for evaluating pitching performance. He was fascinated by the impact the count on a batter had on the result and collected data from the games he watched and listened to. In recent years, he has become an avid reader and developed a passion for learning more about the historical aspects of baseball, with a special interest in the World Series.</billm2126@gmail.com>


P14: Roberta J. Newman, “Koshien Dreams: Yuuji Terajima’s Ace of the Diamond and the Reinforcement of Cultural Values in Baseball Manga”

Consistently ranked as one of the top-selling manga since its introduction in 2006, Ace of the Diamond (ダイヤの A エース, Diaya no Aesu), written and illustrated by Yuuji Terajima, is an exemplar of its type. Manga, an extremely popular form of serialized graphic narrative, accounts for more than half of published material in Japan. Unlike baseball and other sports fiction aimed at a U.S. readership, in which the underdog generally triumphs, despite the odds stacked against it, baseball manga also features its fair share of failure, though it is often followed by victory, but only after protracted struggles and a great deal of practice. Ace of the Diamond is the story of Eijin Sawamura, an ambitious young player from the provinces, recruited to pitch for a baseball powerhouse high school. At the outset, his unusual talent is almost eclipsed by his arrogance and unpracticed manner. Newman examines the first three volumes of the original forty-seven volume set. She demonstrates the ways in which it uses baseball not only to reflect, but to reinforce, Japanese cultural values.

Roberta J. Newman <rn5@nyu.edu> is a Clinical Professor of Liberal Studies at New York University, whose work focuses on the intersections of baseball and popular culture. She is co-author of Black Baseball, Black Business: Race Enterprise and the Fate of the Segregated Dollar (2014), and author of Here’s the Pitch: the Amazing, True, New, and Improved Story of Baseball and Advertising (2019).</rn5@nyu.edu>


P15: Ashwin Pasupathy, “A Markov Model Using Batted Ball Profiles for Lineup Optimization”

Every day, MLB managers are faced with how to create a robust lineup that can generate the most runs. An optimized batting order theoretically can improve the expected wins in a season by three, so strong lineup creation can often be the factor in the final push to the postseason. Previous works involving lineup optimization have used discrete state-space Markov Models to obtain lineup run distributions by treating baseball as a Markovian process. Pasupathy presents an alternative approach, using a batted ball derived Markov Model for robust creation of MLB lineups. Using batted ball data for players obtained from Statcast stratified by pitcher type, he samples each player’s expected Weighted On-Base Average (wOBA) distribution using Monte Carlo methods to obtain hit and out probabilities. To validate the model, he runs simulations on MLB lineups from games in 2019-2021 then compares the results to the true outcomes. The work creates a Markov Model that accurately simulates the run output of a specified lineup of players.

Ashwin Pasupathy <apasupa2@jhu.edu> is an undergraduate junior studying Applied Mathematics and Statistics at Johns Hopkins University. Ashwin is a part of Dr. Anton Dahbura’s Sports Analytics Group working with the Orioles on baseball research problems. He is from San Jose, California and is a huge San Francisco Giants fan.</apasupa2@jhu.edu>


P16: Marty Payne, “Jim Crow Plays Hard Ball”

Payne surveys black baseball on the Eastern Shore of Maryland, including significant new research since his virtual presentation to the Babe Ruth Chapter meeting in April 2021 that has shifted the emphasis. Themes include 19th century activity, baseball and migration during Jim Crow, the importance of baseball in black communities, local and regional newspapers and the Afro-American, the Negro Leagues on the Eastern Shore, and players of the Negro Leagues that came from the Eastern Shore of Maryland. Payne looks at black baseball in a rural environment that was once a slave holding region. It was a region considered by the national press as a “hot bed” for white baseball. And it was near enough to Negro League markets in Baltimore, Philadelphia, and New York to be impacted by them. As such, he provides a unique perspective of baseball in African American communities.

Marty Payne <martyp453@gmail.com> is a native of the Eastern Shore of Maryland and lives and works from his home in St. Michaels, Maryland.</martyp453@gmail.com>


P17: Tom Thress, “The First No-Hitter in World Series History”

The recent declaration by MLB that the Negro Leagues were Major Leagues has introduced a new set of firsts in major-league history. One new first took place here in Baltimore on October 3, 1926, when Claude “Red” Grier threw the first no-hitter in a World Series game. Thress will show a brief overview of the teams who met in the 1926 Negro League World Series, the Atlantic City Bacharach Giants and Chicago American Giants. He will review the 1926 World Series, which went a record 11 games, provide a brief biography of Claude Grier, and review Grier’s no-hitter in some detail, taking advantage of Retrosheet’s recent expansion into Negro League games, including a full play-by-play account of Grier’s big game.

Tom Thress <tthress@retrosheet.org> has been a SABR member since 2003. He presented a poster presentation at SABR 48 and gave an oral presentation (over Zoom) as part of SABR’s virtual convention in 2020. Mr. Thress has had research published in the SABR publications By the Numbers and the Baseball Research Journal and is the author of two books, Player Won-Lost Records in Baseball: Measuring Performance in Context (McFarland, 2017) and Baseball Player Won-Lost Records: 150 Players, 50 Seasons (self-published, 2018). Mr. Thress is treasurer of Retrosheet and is heading up Retrosheet’s efforts to add Negro League games,which he presented at the 2022 Jerry Malloy Negro League Conference. In his day job, Mr. Thress is a professional economist with a master’s degree in Economics from the University of Chicago.</tthress@retrosheet.org>


P18: Gregory H. Wolf, “A Fighting Spirit: The 1901 AL Charter Member Baltimore Orioles”

Wolf examines the 1901 AL charter member Baltimore Orioles, an often-overlooked club in baseball history. Two years later, the club relocated to New York City and became the Highlanders (and eventually the Yankees), relegating the complex and fascinating story of its inaugural 1901 season to the margins. The Orioles moved to the National League in 1892 upon the American Association’s demise. The poster focuses on the 1901 AL charter member Orioles, commencing with the competing plans in mid-1900 to establish a new professional team in the Monument City. After presenting the success of Ban Johnson’s plan and overtures to McGraw and the formation of a Baltimore baseball company, Wolf explores how the team and its roster of players came into rapid existence in less than four-months. He elucidates the construction of an entirely new ballpark, Oriole Park, located just blocks from Union Park, where the Orioles had played in 1899, chronicles the team’s spring training in Hot Springs, Arkansas, and concludes with highlights from the regular season.

Holder of the Dennis and Jean Bauman Endowed Chair of the Humanities at North Central College in Naperville, Illinois, Gregory H. Wolf <ghwolf1967@gmail.com> has served as co-director of SABR’s BioProject since 2017. He has edited 15 books for SABR and has written approximately 200 biographies of players for the BioProject and approximately 250 game stories for the Games Project.</ghwolf1967@gmail.com>


P19: Peiyuan Xu, “Introducing a Reinforcement Learning Framework for Dynamic Baseball Scheduling”

Scheduling sports matches is still a challenging problem, where the goal is not only to create robust tables that are both fair and suitable to all players while satisfying a variety of constraints. Although many advancements have been made via various optimization approaches such as semi-definite programming, heuristic methods, and hybrid approaches, they are based on a model which the central scheduler creates a schedule for all of the teams in a league. Xu presents a dynamic scheduling process in which teams select series against other teams in a round-robin fashion. Xu introduces a reinforcement learning (RL) framework that consists of a dynamic scheduling process between the baseball league scheduler and the RL agent to achieve individual teams and league goals, which is modeled as a Markov decision process (MDP). He employs deep Q-learning to have the agent learn this framework as well as its interaction with the league scheduler, using the High-A Central Baseball League as an example.

Peiyuan Xu <pxu11@jhu.edu> is currently pursuing his Bachelor of Science degree double majoring in Computer Science and Applied Mathematics and Statistics (AMS) at Johns Hopkins University, where his current research interests include machine learning, reinforcement learning (RL), data science, and optimization research. He has performed research in several areas including modeling and simulation of community-wide infectious diseases; data mining and performance analysis of cloud computing applications; and combining optimization and reinforcement learning approaches to sports scheduling applications. Mr. Xu will pursue his Master’s degree in Computer Science, where he will explore incorporating RL methods in sports scheduling.</pxu11@jhu.edu>


To go back to the SABR 50 convention main page, click here.

For more information, or to register for the SABR 50 convention in Baltimore, visit SABR.org/convention.

© SABR. All Rights Reserved