WHICH LEAGUE HAS GREATER PARITY,
THE NFL OR THE NBA?
Sports fans love the NFL because it seems like there is always a surprise team that wins the Super Bowl or challenges for the championship. For example, who expected Tampa Bay to win the Super Bowl in 2002? NBA fans complain the same teams (such as Detroit and San Antonio) are always on top. It is easy to show that the NFL does indeed exhibit more parity and unexpected team performances than does the NBA.
If a league has a great deal of parity you would expect it to be difficult to predict a team's performance based on their previous year's performance. That is, teams that do poorly one season should have a good shot at being above average the following season, and vice versa. If a league exhibits little parity you would expect that it would be relatively easy to predict a team's performance for one season using the previous year's performance. What metric should we use to measure team performance? The simplest metric would be regular season wins, but we have seen that NFL teams play schedules that differ by 7 or more points in difficulty. A team with a tough schedule and 9–7 record might be much better than a team that went 11–5 with an easy schedule. We will use the final Sagarin rating as our metric for team performance during a season.1 These ratings include all regular and post- season games and are quite similar to the least squares ratings described in chapter 40. The file Parity.xls contains the ratings for each NFL team and NBA team for five seasons (for the NBA, 2002–3 through 2006–7 seasons and for the NFL, the 2002–6 seasons). We try to predict each team's Sagarin rating during a year based on their previous year's Sagarin rating. For example, for our NBA data we would have the following four data points for the Spurs: (Spurs 2002–3 rating,
1 See http://www.usatoday.com/sports/sagarin.htm.