The History of Parity in the NFL pt. 1
2022 had the most parity of any NFL season in recent memory
There have been multiple attempts to measure the parity of the NFL relative to other North American professional sports leagues. In wake of the two conference #1 seeds making SB LVII, I seek to answer if the NFL’s parity has been consistent throughout its history. I measure this by analyzing relative team strength via pythagorean expectations and by observing what percent of games in a season ended as one-score games.In both measures I find that 2022 had the most parity of any NFL season in recent memory. I also find that relative team strength has been on an extremely gradual decline since the salary cap was instituted in 1994, and that between 45-50% of NFL games have ended within one score since the 80s.
There have been multiple attempts to measure some semblance of parity in sports across leagues, notably by Michael Lopez in 2013 and by Kevin Alwell in 2020. Both authors address the issue by looking at the team level. They either evaluate how quickly a team’s fortunes can change in a single year or how much a subset of teams dominate the league’s championships. These attempts well describe the problem in their schema and serve as excellent comparison points for the NFL relative to other North American professional sports. I am going to build on their work by focusing specifically on parity within the NFL over its history.
Instead of looking at high-level team results such as the spread of winning percentages over eras, playoff participants, or championship winners, I chose to look at team-season level and game-level results and relative team strengths in a given season. If a league has high parity, then teams are similarly strengthened and there will be lots of close games. By measuring team strength via expected winning percentage and the amount of close games in a season, we can circumvent several issues in historical comparison in a league with an ever-changing landscape.
We’ll begin by first looking at relative team strength. The obvious option would be to look at win-loss records, however since we want to compare individual seasons against each other this becomes too noisy. Even in a league where every team is exactly the same strength, it’s extremely likely one team wins 12 games due to random chance over a 16 or 17 game season. Although in that same hypothetical even league, we can suppose that most of those random wins would be in close games. Therefore, the teams would likely score and allow a similar amount of points across the league.
Using this logic, we will evaluate the strength of every team in the league in a season by generating a pythagorean win percentage for them. To then get an idea of relative strength across the league, we will get the difference between the team with the highest expected win percentage and the lowest expected win percentage. If this difference is large, this means the best team is much better than the worst team indicating low parity. Thus, a smaller difference indicates a league with more parity.
Note: AFL games from 1960-1969 are considered NFL games for the purposes of this analysis. An exponent of 2.77 was used in the Pythagorean Expectation
This plot displays the expected win percentage range for every NFL season since the league’s inception in 1920. In case it is not terribly clear how to read this, a range of 100% means in that season the best team was expected to go undefeated, while the worst team was expected to go winless. Interestingly, there has only been one significant period of parity increase in the NFL.
From the institution of a championship game in a slightly more buttoned up semi-pro league to the beginning of the AFL, parity increased dramatically. The likeliest explanation in my eyes is that as the league grew more and more professional in its practices, the most professional/best teams had less of an edge and were thus not as dominant as before. This actually reversed somewhat from 1960-1978, but stabilized at a higher-parity state in the following years.
There appears to be a very gradual parity increase starting with the beginning of the salary cap to today, but the correlation is not strong. 2022 did, however, set a 34 year high in parity with the Philadelphia Eagles only being expected to win 51.0% more games than the Indianapolis Colts. Translating to wins, this means the Eagles were expected to win 8.7 more games than the Colts. Additionally, the past 5 seasons were all between 59.3% and 63.1% win percentage (10.0 and 10.7 wins), which is consistently lower than any other 5-year streak in the sample by a thin margin.
This means the 2022 season had the most parity in terms of relative team strength since 1988, and the past 5 seasons constitute the consistently highest parity seen in NFL history. The difference is extremely slight though, and the more important acknowledgement is that relative team strength has been relatively consistent or on a slight decline since the implementation of the salary cap in 1994.
So the best teams today are not as dominant as they have been in the past, however this does not ensure that the season as a whole has had a lot of close games. To measure the amount of close games in a season, we will look at the proportion of games that ended with a one score margin. Since the 2 point conversion was not instituted until 1994 we will be considering a one score margin to be 7 points or fewer.
From this chart we are able to parse out a basic competitive arc of the NFL on the game level. There are fascinating tales within this chart, particularly in the time between the first official championship game in 1933 and the AFL’s maiden season in 1960. But, unfortunately, we will have to revisit these tales another day.
The important part for our analysis is the trendline’s overall shape. In the pre-WWII primordial ooze of professional football, year over year results were extremely volatile with no particular trend. As we entered the 50s, a decade which saw the Lions, Browns, and Colts capturing 8/10 championships, the NFL stabilized around 35-40% of games ending up as one-score games. The AFL’s inception in 1960 and eventual 1970 merger coincided with an extremely gradual increase in the rate of one score games until it stabilized between roughly 45-50% around the early to mid 80s. Most seasons have remained in that one-score range since, with the range maybe tightening following the introduction of the salary cap in 1994 and rookie wage scale in 2011.
In 2022, 52.3% of games ended with a margin of victory within 7 points, the 2nd-highest in NFL history only trailing 1932’s 57.7%. Though, it is by no means an outlier season. 2015 and 2016 each had 51.7% of games end as one-score games and only 6 seasons this century having less than 45% of games in a year end as one-score games. So to answer our original question: the modern NFL has more parity, as measured by the amount of close games in a season, than it did in the 70s, however it has remained in the same range for about 40 years. Around half of all NFL games were tremendously close in the start of the Reagan administration, and it’s remained that way ever since. This is remarkable.
Over the past 40 years the US has gone through 7 presidents. The USSR fell, and along with it all the communist regimes in eastern Europe. The global economy has risen to historic heights only to burst on three separate occasions. The AIDS epidemic ravaged through LGBT communities worldwide. Multiple wars in the middle east began and came to an end. The internet became a ubiquitous part of everyday life and fundamentally changed nearly every aspect of culture, communication, and entertainment. The world has shrunk considerably and allowed people to share their lives and ideas with each other to beautiful and horrific ends. COVID forced governments to lock down for years, and those ramifications will cast their long shadow into the unforeseen future.
Despite all this change and turmoil, humanity has persevered. We’ve adapted. We’ve lived and laughed and loved and cried and cheered. We’ve made families through blood or friendship. We’ve fought for change. We’ve created beautiful, lasting art. We’ve made incredible scientific achievements. We mapped the human genome. We created a vaccine for a disease which fundamentally changed all our lives in only 18 months. We’ve kept on living.
Over the past 40 years the world has changed in some ways to be nearly unrecognizable from what it was before. Yet humanity has remained as it always has. We’ve adapted and thrived and continued to be Sisyphus, smiling, ready to push the boulder back up the hill. And through all of this, the NFL has been an Any Given Sunday league, every fall and early winter, without fail for every single one of the 40 years. I don’t know, I just think there’s something beautiful in that.
Thank you for your patience, and I hope you’ll excuse my non-football ramblings at the end of this post. Next week we’ll be putting the result of SB LVII in historical context, and afterwards we will resume this analysis with a more robust modelling approach. Thank you for reading, it is much appreciated. -CRM