The Probability of Winning and the Effect of Home-Field Advantage: The Case of Major League Baseball

Levernier, William, Barilla, Anthony G., Academy of Information and Management Sciences Journal

ABSTRACT

This paper examines the factors that affect the probability of a major league baseball team winning a game. The basic hypotheses of the study are that home teams are more likely to win a game than visiting teams, that teams that travel to arrive at a game are less likely to win the game than teams that don't, and that teams having a strong batting performance are more likely to win a game than teams having a weak batting performance. To examine these issues, we estimate five logit regressions from data for all 2,428 regular season games played during the 2004 season. We find that while the strength of a team's batting performance does affect its probability of winning, travel does not affect the likelihood of either the home team or visiting team winning a game. The major finding of the paper, however, is that contrary to the commonly held belief that a home-field advantage exists in major league baseball games, home teams only have an advantage over visiting teams in very close games. In games that are won by more than one run, the likelihood of winning is roughly equal for home teams and visiting teams.

INTRODUCTION

In major league baseball, like most other professional sports, the conventional wisdom is that a home-field advantage exists. Birnbaum (2004, p. 972) reports that home teams have historically won about 54 percent of their games. The difference between a 54 percent winning percentage and a 46 percent winning percentage is substantial since, during a standard 162-game season, a team that wins 54 percent of its games will accumulate 12 more victories than a team that wins 46 percent of its games. Twelve additional wins during the course of a season often makes the difference between a team going to the post-season playoffs and not going to the playoffs. In the two most recent seasons, 2003 and 2004, the first place team won fewer than twelve more games than the second place team in five of the six Major League Baseball divisions. (1)

One reason the home team has the advantage in baseball is the fact that they bat last, which becomes a factor in one-run victories. If a game enters the top of the last inning with the score tied, for example, the manager of the visiting team doesn't know whether his strategy should involve trying to score a single run, since he doesn't know whether or not one run will ultimately be enough to win the game. If the score is tied entering the bottom of the last inning, however, the manager of the home team knows that a single run will be enough to win the game, and he can therefore employ a strategy designed to score just one run. Another possible reason that a home team has an advantage is that the visiting team experiences travel-induced stress and fatigue. Since the visiting team must travel to arrive at a game, it incurs the inconveniences associated with travel, in terms of both the physical act of traveling and the act of staying in an unfamiliar city. In some cases the home team also incurs the inconvenience of travel. (2) If the home team does travel, they would be subjected to the same travel-induced fatigue as the visiting team, but they would not experience the discomfort of being away from the familiar surroundings of home. As such, when both teams travel to a game the visiting team is more likely than the home team to be adversely affected by the travel.

The primary purpose of this paper is to determine the effect that home-field advantage has on the probability of a team winning a major league baseball game played during the 2004 season. We also determine the effect that team batting performance and travel have on the probability of a team winning a game. Specifically, we will determine whether a home-field advantage exists and, if so, whether it exists generally or only in limited situations. To examine these issues we develop and estimate a series of binary logit regressions where the outcome of the game (i.e., win or lose) is the dependent variable.

In the next section we review the literature pertaining to the home-field advantage in major league baseball and to the analysis of factors affecting the run production of baseball teams. In the third section we discuss the data and descriptive statistics and report the probability of victory in various situations. In the fourth section we describe the logit regressions. In the fifth section we report and discuss the regression results. Finally, in the last section we present a summary of our major findings and offer some concluding remarks, including a suggestion for potential directions that future research on the subject of the home-field advantage might take.

REVIEW OF THE LITERATURE

A relatively new and popular field that applies statistical models and methodologies to baseball data is sabermetrics, which derives its name from the Society for American Baseball Research (SABR), an organization devoted to furthering the study of baseball. Birnbaum (2004, p. 963) defines sabermetrics as "the science of answering questions about baseball through the analysis of the statistical evidence." It has also been defined by Bill James, the man who popularized sabermetrics in the early 1980s in the initial versions of the annual The Bill James Baseball Abstract, as "the search for objective knowledge about baseball" (Grabiner).

The scholarly literature has examined several baseball related issues. Lindsey (1963), in one of the earliest academic studies pertaining to baseball performance, derives a formula that explains the number of runs a team scores based on the various components of its hitting production. Albert (1994) employs a Bayesian hierarchical model to determine which game-situations affect players' batting average and determines that several situations affect batting average: the pitch count faced by the batter, facing a pitcher of the opposite arm, facing a groundball pitcher, and playing in a home game. Albright (1993) conducts a statistical analysis of hitting streaks among major league batters during the 1987-1990 seasons and concludes that hitting streaks happen at about the same rate as what would occur in a random model. Gius and Hylan (1996), in a statistical study of the determinants of baseball player salaries, use a fixed-effects multivariate regression model to …

Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: The Probability of Winning and the Effect of Home-Field Advantage: The Case of Major League Baseball. Contributors: Levernier, William - Author, Barilla, Anthony G. - Author. Journal title: Academy of Information and Management Sciences Journal. Volume: 9. Issue: 2 Publication date: July 2006. Page number: 61+. © The DreamCatchers Group, LLC 2007. COPYRIGHT 2006 Gale Group.

This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.