- Joined
- Jun 28, 2010
- Messages
- 49,227
- Name
- Burger man
http://espn.go.com/blog/nflnation/post/_/id/166141/inside-slant-innovating-the-nfl-schedule
(Another in an Inside Slant series on innovation in and around the NFL. For all Inside Slant posts, follow this link.)
In February, a group of researchers revealed a surprising level of systemic disparity in the NFL's yearly game schedule. They presented their paper at the MIT Sloan Sports Analytics Conference, most notably identifying long-term inequities in matchups between teams with dissimilar rest, and soon found -- believe it or not -- that the NFL wanted to compare notes.
And so it came to be that Drs. Murat Kurt and Mark Karwan spent an hour recently speaking with Mike North, the NFL's director of broadcast planning and one of the key members of an executive team that produces the league's game schedule. Their discussion won't impact the 2015 edition, likely to be released next week, but their pending collaboration could help smooth out the edges of future efforts.
"We would look at this as something fun for us and maybe a help for the NFL," said Karwan, who like Kurt is a professor at the University at Buffalo. "You can't get a perfect schedule, but ... our goal would be to show them that a lot of the schedules they generate could be improved."
To be fair, anecdotal schedule inequities have helped evolve the NFL's schedule release into a day of national hysteria. In 2014, for example, we saw the New Orleans Saints play four games in five weeks against teams coming off either their bye week or a Thursday night game. The Tampa Bay Buccaneershad only four home games in a 13-week window, losing 10 of 12. The Baltimore Ravens, on the other hand, played only one divisional game in the strong AFC North after Week 9. They went 4-2 after that point and earned a wild-card playoff bid.
We typically dismiss such disparities with the expectation that they even out over time. But when the Buffalo Bills complained publicly about concerns in their 2013 schedule, Kurt and Karwan teamed with Ph.d. candidate Niraj Kumar Pandey and student Kyle Cunningham to investigate. They emerged with a startling discovery: The league's schedule formula did not correct for the disparities, leaving some teams to absorb significantly more competitive disadvantage than others.
The chart provides a summation of their most notable finding. Since the eight-division format began in 2002, some teams have played more than twice as many games against better-rested opponents -- either coming off a bye or a Thursday night game -- than others. The Bills had the most (29), and the Cincinnati Bengals were given the fewest (14). In 2013, the Bills had five such games; the division rival New England Patriots had none.
It seems obvious that a team with more rest than its opponent has an inherent advantage. To quantify the competitive impact, the researchers looked at a five-season span from 2009-13. Over that period, the NFL's average win percentage when playing a team with more rest was 44.7 -- about four percentage points lower than the average win percentage against all opponents. In 2010, notably, the difference was 14 percentage points.
"The way I put it," Karwan said, "is that it's significant enough that I'm sure Las Vegas would want to know about it."
So the group developed a mathematical method of creating schedules with conditions that address the competitive advantages they identified. The details can be found in their paper via this link, but in essence their resulting formula eliminates the bye week advantage and minimizes the turnarounds caused by Thursday night games. They reduced the instances of playing three consecutive road games and spread divisional games more evenly through the schedule.
In a statement provided to ESPN.com, league spokesman Greg Aiello said the formula "is similar to what the NFL uses but differs in that it focuses only on fairness." Aiello added: "It does not, for example, take into consideration television ratings and other matters. We have always said that we look for the right balance between competitive issues and other considerations such as television. The idea of creating schedules that are supposedly 'fair' to all teams but don't give consideration to television ramifications and other matters, such as stadium conflicts, is unrealistic."
Indeed, Karwan and Kurt acknowledged, as third-party researchers they had no access to the long annual list of television requests and limits to stadium availability. Each year, for example, broadcasters identify about 40 of the 256 games for prime-time or 4 p.m. ET slots. Those requests naturally impact flexibility for the remaining games.
North communicated a message similar to Aiello's but, according to Karwan and Kurt, expressed interest in their methods and agreed to provide the missing proprietary information -- after the schedule has been released -- to see if their model produced better results.
"Right now we can't claim we are going to do it," Karwan said. "We would need a new student to help and it's probably going to take two or three years to expand our method. It's possible we'll have to loosen some of our other conditions to make it work. But we have the ideas, and they said they could get us the data. Maybe it is a situation where we could eventually generate the 2018 schedule after it's announced and then we can compare."
There don't appear to be any conspiracy theories here. A method that gives an advantage to the Bengals, Carolina Panthers, New York Jets and Jacksonville Jaguars -- four of the six teams at the bottom of the chart -- doesn't align with any favored-team conspiracies I'm aware of.
These inequities are mostly a matter of math and economic reality. Nothing will change the latter, but hopefully in a few years, we'll find that the former can be improved.
(Another in an Inside Slant series on innovation in and around the NFL. For all Inside Slant posts, follow this link.)
In February, a group of researchers revealed a surprising level of systemic disparity in the NFL's yearly game schedule. They presented their paper at the MIT Sloan Sports Analytics Conference, most notably identifying long-term inequities in matchups between teams with dissimilar rest, and soon found -- believe it or not -- that the NFL wanted to compare notes.
And so it came to be that Drs. Murat Kurt and Mark Karwan spent an hour recently speaking with Mike North, the NFL's director of broadcast planning and one of the key members of an executive team that produces the league's game schedule. Their discussion won't impact the 2015 edition, likely to be released next week, but their pending collaboration could help smooth out the edges of future efforts.
"We would look at this as something fun for us and maybe a help for the NFL," said Karwan, who like Kurt is a professor at the University at Buffalo. "You can't get a perfect schedule, but ... our goal would be to show them that a lot of the schedules they generate could be improved."
To be fair, anecdotal schedule inequities have helped evolve the NFL's schedule release into a day of national hysteria. In 2014, for example, we saw the New Orleans Saints play four games in five weeks against teams coming off either their bye week or a Thursday night game. The Tampa Bay Buccaneershad only four home games in a 13-week window, losing 10 of 12. The Baltimore Ravens, on the other hand, played only one divisional game in the strong AFC North after Week 9. They went 4-2 after that point and earned a wild-card playoff bid.
We typically dismiss such disparities with the expectation that they even out over time. But when the Buffalo Bills complained publicly about concerns in their 2013 schedule, Kurt and Karwan teamed with Ph.d. candidate Niraj Kumar Pandey and student Kyle Cunningham to investigate. They emerged with a startling discovery: The league's schedule formula did not correct for the disparities, leaving some teams to absorb significantly more competitive disadvantage than others.
The chart provides a summation of their most notable finding. Since the eight-division format began in 2002, some teams have played more than twice as many games against better-rested opponents -- either coming off a bye or a Thursday night game -- than others. The Bills had the most (29), and the Cincinnati Bengals were given the fewest (14). In 2013, the Bills had five such games; the division rival New England Patriots had none.
It seems obvious that a team with more rest than its opponent has an inherent advantage. To quantify the competitive impact, the researchers looked at a five-season span from 2009-13. Over that period, the NFL's average win percentage when playing a team with more rest was 44.7 -- about four percentage points lower than the average win percentage against all opponents. In 2010, notably, the difference was 14 percentage points.
"The way I put it," Karwan said, "is that it's significant enough that I'm sure Las Vegas would want to know about it."
So the group developed a mathematical method of creating schedules with conditions that address the competitive advantages they identified. The details can be found in their paper via this link, but in essence their resulting formula eliminates the bye week advantage and minimizes the turnarounds caused by Thursday night games. They reduced the instances of playing three consecutive road games and spread divisional games more evenly through the schedule.
In a statement provided to ESPN.com, league spokesman Greg Aiello said the formula "is similar to what the NFL uses but differs in that it focuses only on fairness." Aiello added: "It does not, for example, take into consideration television ratings and other matters. We have always said that we look for the right balance between competitive issues and other considerations such as television. The idea of creating schedules that are supposedly 'fair' to all teams but don't give consideration to television ramifications and other matters, such as stadium conflicts, is unrealistic."
Indeed, Karwan and Kurt acknowledged, as third-party researchers they had no access to the long annual list of television requests and limits to stadium availability. Each year, for example, broadcasters identify about 40 of the 256 games for prime-time or 4 p.m. ET slots. Those requests naturally impact flexibility for the remaining games.
North communicated a message similar to Aiello's but, according to Karwan and Kurt, expressed interest in their methods and agreed to provide the missing proprietary information -- after the schedule has been released -- to see if their model produced better results.
"Right now we can't claim we are going to do it," Karwan said. "We would need a new student to help and it's probably going to take two or three years to expand our method. It's possible we'll have to loosen some of our other conditions to make it work. But we have the ideas, and they said they could get us the data. Maybe it is a situation where we could eventually generate the 2018 schedule after it's announced and then we can compare."
There don't appear to be any conspiracy theories here. A method that gives an advantage to the Bengals, Carolina Panthers, New York Jets and Jacksonville Jaguars -- four of the six teams at the bottom of the chart -- doesn't align with any favored-team conspiracies I'm aware of.
These inequities are mostly a matter of math and economic reality. Nothing will change the latter, but hopefully in a few years, we'll find that the former can be improved.
