Using a Binomial Logit Model to Estimate and Explain the Home-field Advantage in Major League Baseball

In a sporting event involving two teams, the home team generally has a higher likelihood of winning the event than the visiting team. This advantage is known as the home-field advantage.  This paper develops a statistical... [ view full abstract ]

In a sporting event involving two teams, the home team generally has a higher likelihood of winning the event than the visiting team. This advantage is known as the home-field advantage.  This paper develops a statistical model that estimates the magnitude of the home-field advantage and identifies the factors that affect the advantage in major league baseball.  The model is estimated using all 24,297 regular season games that were played during the ten seasons from 2006 through 2015 as observations.  Post-season games (i.e., the playoffs and World Series) are excluded from the analysis.  The data were obtained from the Baseball Reference website (www.baseballreference.com).

The dependent variable in the regression model is a binary variable that indicates whether the home team won or lost a particular game.  Several independent variables that are hypothesized to affect the likelihood of the home-team winning a game are included in the model.  The independent variables include the number of runs scored by the home team, the number of runs scored by the visiting team, how close the game is, whether the game goes into extra innings, and whether or not the game is played with a designated hitter.  Also, a series of team dummy variables are included to determine whether certain teams have a larger home-field advantage than other teams, and a series of year dummy variables are included to determine whether the home-field advantage is larger in some years than in others.  The regression equation is estimated as a binary logit model, which is a model that is commonly used in economics when the dependent variable is a categorical variable.

Some of the major findings of the paper are: 1) A home-team’s likelihood of winning a game increases as it scores more runs, but the increase is at a decreasing rate; 2) The home-field advantage is larger in games that are relatively close than in games that aren’t close, with the advantage being largest in games where the winning team wins by one run; 3) The magnitude of the home-field advantage varies by team and by season; 4)  The home-field advantage is affected by whether or not the game goes into extra innings; and 5) The home-field advantage is larger in games where the home team is a National League team (the designated hitter isn’t used a these games) than in games where the home team is an American League team.