NFL schedules could be fairer, UB analysis finds

The NFL is widely admired for its parity. But could it do a better job promoting fairness, especially when scheduling games?

The answer, according to a new mathematical analysis by UB engineers, is yes.

The study, “Alleviating Competitive Imbalances in NFL Schedules: An Integer-Programming Approach,” was one of eight research papers presented last Friday at the MIT Sloan Sports Analytics Conference in Boston.

The paper’s authors are Niraj Pandey, PhD candidate in the Department of Industrial and Systems Engineering (ISE), and Kyle Cunningham, who earned a BS from UB in 2014. Co-advisers are ISE’s Murat Kurt, assistant professor, and Mark Karwan, Praxair Professor of Operations Research and SUNY Distinguished Teaching Professor.

In the NFL, teams competing against an opponent coming off a bye week or Thursday game are at a disadvantage. This is because the opponent, among other things, has more time to recuperate and make coaching adjustments.

Several teams, including the New York Giants and Buffalo Bills, have filed complaints with the NFL in recent years concerning the issue.

The UB engineers examined NFL schedules from recent years. They found a disparity among teams that played against opponents coming off a bye or Thursday game.

For example, from 2002-14, they found the Bills played more games against rested opponents (26) than any other team. From 2003-13, they found the Atlanta Falcons played more teams coming off a bye (17) than any other opponent.

(Note: Several of the engineers are diehard Bills fans. The study is motivated, in part, upon eradicating scheduling rarities the Bills and other teams face. No person or organization, including Bills, has funded the research.)

“Most of us and our colleagues love sports, and this intellectual curiosity-driven research was a fun way to use our technical skills to see how we could address an important issue in the world’s No. 1 revenue-generating league,” says Karwan. “Besides, our hometown Bills were one of the biggest victims of scheduling anomalies!”

To address the issue, the engineers created a mathematical model called a mixed-integer linear program. The model is similar to what the NFL uses; however, it differs because it focuses solely on scheduling with fairness. It does not, for example, take into consideration television ratings and other matters.

By applying the model to the 2012, 2013 and 2014 seasons, the engineers found that:

• Bye-week disadvantages could be eliminated.
• Disadvantages due to Thursday games could be kept “reasonably low” for each team, while avoiding undesirable road game patterns and long-distance travel in short weeks.
• No team has to play against more than two more-rested opponents per season.
• Games between divisional opponents can always be separated by at least two weeks, and at least one game can occur in the second half of the season.
• The model is stable enough to create hundreds of schedules for future NFL seasons in a matter of minutes.

“In a league of 32 teams, where a 17-week season can be easily called short, the sequence of games is naturally more decisive on where teams end up in late December. The anomalies we found in recent schedules are what drove us to find a better solution and tune them for fairness,” Kurt says.