![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
This study presents three tests of efficiency of the NFL betting market for the years 1994–2000. First, it tests for weak-form informational efficiency of the betting market. Then it examines whether the market incorporates objective information such as power scores and stadium characteristics that might be useful for predicting game outcomes. Finally, it determines whether alternative betting strategies would have yielded a profit. Although there is some indication that differences in the playing surfaces of home and visiting teams were not fully reflected in the betting lines, it is found that there is no conclusive evidence to suggest that the market was inefficient over the period examined.
Acknowledgements
We would like to thank Robert S. Goldfarb for very helpful comments on an earlier draft of this paper.
Notes
1 In 1994, which is the first year in the present sample, there were only 28 teams. In 1999, the number of teams increased from 30 to 31. The teams are divided into two Conferences and within each Conference there are three Divisions. While each team plays 16 games, their opponents are selected on the basis of each team's performance in the previous year.
2 The Green Bay Packers partly attribute their difficulty in winning in the Detroit Lions’ Silverdome (only two wins in their previous nine games) to the noise level in the dome (Silverstein, Citation2000). The Washington Redskins also cite problems with the noise level in the dome (All of the whistles add to decibel level at the Silverdome, The Washington Times, 6 December 1999).
3 Gray and Gray (Citation1997, p. 1734) find a success rate of 50.4% in betting on the home team for the 1976–1992 NFL seasons.
4 Osborne also investigates betting on games in which the predicted margin of victory is five points greater than the betting spread. This five-point filter yields a winning percentage of 54.3%, which is inferior to his three point rule.
5 Presumably, Osborne does not include the first week of each season. His regression equation would predict a zero margin of victory for these games, since the average points scored in previous games by the home and away teams and their opponents would all equal zero for the first week of each season.
6 The power scores are based on data similar to Osborne's independent variables.