# This NFL Season Did Not Have As Much Parity As You Think

Over the course of this season, much was made about the parity being exhibited in the NFL. Article after article was written about how this year parity ruled the NFL. This got me wondering, how much parity was there in the NFL? Did these articles have a real basis, or was it just the case that different teams were coming out on top.

To test this, I scraped the number of wins of each NFL team from 1978 (the first season with 16 games) until 2017 from Pro Football Reference. The 1982 and 1987 seasons were not considered because fewer than 16 games were played (1982 played 9 games and 1987 played 15). For the purposes of this study, ties are counted as half a win.

For each season, I calculated two metrics that help paint a picture of the level of parity. The first was the standard deviation of the number of wins of each team. The 2017 season had a standard deviation of 3.20, which ranked tied for 25th out of the 38 seasons that were considered. If we were only to look at the 16 seasons in which there were 32 NFL teams, the 2017 season would rank tied for 10th (the highest ranked was the 2002 season and the lowest ranked was the 2005 season). This doesn’t paint a strong picture for the case that there was a ton of parity in the 2017 season.

However, there are a couple of problems with using standard deviation as a tell all metric. The standard deviation would maximize if half the teams went 16-0 and half the teams went 0-16 (this is impossible with the NFL scheduling formula, but you understand the point). In that hypothetical case, you could make strong arguments both ways as to if there was a lot of parity in the league. However, standard deviation was used to see how much spread there is from the middle.

The second metric calculated using this data is an HSAC favorite, the Gini coefficent. For those uninitiated, the Gini coefficient is a metric often used to describe income disparity in a country. Countries where every single person has the same income has a coefficient of 0, a country where one person has all the money has a coefficient of 100. We calculated the Gini coefficient for every season where the unit of income was wins. We found that of the 38 seasons studied, the 2017 season ranked 24th with a Gini coefficient of 22.14.

Our results can effectively be summarized in the table below:

Overall, our study does not align with the trend that the 2017 season was ripe with parity. It might have been the case that it was different teams (like the Jaguars and Rams) excelling than in past years, but the overall distribution of teams records was wider than the average NFL season.

#### harvardsports

View all posts

• Ed A says:

Another take on this is to look how the favorites did in Vegas this year. After the first few weeks of the season, they did quite well, even as point spreads grew! With more parity, favorites would struggle and point spreads would shrink.

• This website was… how do I say it? Relevant!!
Finally I have found something that helped me.
Kudos!

• Noah says:

The idea behind parity is that poor teams in one year have a good chance of doing well the next. Thus, measuring the gap between the best and worst teams year to year is not a good measure. By pure probability of a small sample size, a 16 game season is subject to greater variation than an 82 game or, God forbid, a pointless 162 game schedule.

The better measure of parity is looking at playoff berth turnover. Most teams who make the NFL playoffs one year will not make it the next. That is massive turnover, and allows perennially poor teams like Buffalo and Cleveland to retain rabid fan bases because hey, there’s a chance.