December 8, 2024 The temperature is pretty mild in Kansas at this time of year. With 10°C the weather is far from the coldness with which the Chiefs have just won their 12th match. In the center of Arrowhead stadium, Matthew Wright has just hit a ball 31 yards from the posts. The ball hits the right post, but the echo is quickly drowned out by the clamor that rises in the aisles of the stadium. Falling on the right side of the posts, this field goal gave Andy Reid's men their 12th victory of the season. Their 10th by less than a possession.
Consistently this season, the Chiefs seemed in the hot seat. Consistently, they managed to win. But are such close victories something that can be reproduced from one season to the next? Could any team other than the Chiefs defy logic to this extent?
Wins by less than one possession
This article defines a close game as one with a margin of 7 points or less. Wins by 8 points are excluded, as the probability of making a 2-point attempt is approximately 50%.
Using data since 1999, the teams that have won the most games by one possession are as follows:
Kansas City has just established a historic mark with 10 victories in close games. Record to be put into perspective with a 17-game season, as for the 2023 Steelers.
For these 6 teams with 9 victories, none managed to reproduce the feat during the following season. Aside from Seattle, none came within 4 wins of their total the following season.
Since 1999, the league average of close wins has been 3.8 per team. This value is not particularly stable but fluctuates between 3.25 and 4.25.
Since 2022, the average of close games has reached heights reflecting the current homogeneity of the league.
Stability over time
For each team the evolution of the number of close victories from one season to the next is far from stable. The following season, there was a gap of almost 68% (in absolute value) in the total number of close victories.
By assigning to each team its number of close victories in year n and its number of victories in year n+1, this lack of stability is obvious.
The size of the points is proportional to the number of occurrences of binomial n/n+1. For example, winning 4 close games one season and winning 4 the next season is the most likely occurrence (39 occurrences). The next ones are 3/3 (37 occurrences) and 4/3 (35 occurrences).
The red curve represents the relationship between the number of close wins from year to year (a linear regression for the more math-savvy readers, with a slope of 0.04).
What this graph says is that no matter how many close matches a team wins in a given season, the total tends to fall into line the following season and therefore get closer to the average of the league.
The best example remains the Panthers of 2003. After a season with a record of 11 victories including 9 by less than one possession (concluded with an unfortunate final against the Patriots), they followed up with a season with 7 victories including only one by less than a possession. In 2004 their total of 11 victories identical to 2003 was only obtained thanks to 4 close matches.
Win rate
The analysis is currently incomplete because it focuses on victories. Between 1999 and 2024, the average success rate in the regular season for one-possession games is… 49.3%! Almost a one in two chance of winning this type of match.
And of course, as with the number of victories, the success rate in these matches is not stable from one season to the next. A team will always tend to move closer to the average value after an abnormally good or bad season.
With 10 victories in 10 close games, the Chiefs have probably had a historic season which is a statistical anomaly. Their average win rate in these games has been above average since Mahomes arrived (65% from 2018 to 2023). It is unlikely that in the future they will maintain the level of 2024 standards.
Football Fiction
One-possession games therefore have high variance results. The randomness is very important in the final result. This observation is logical in a sport where the key statistic for predicting the outcome is the points differential. If you dominate your opponents, you put yourself out there and are probably a better team than those who sometimes win narrowly.
To do a bit of football fiction, let's try to eliminate randomness by applying a 50% win rate to each team in close matches. Applying that the league would have looked like this this season:
The Chiefs find themselves in the hot seat, while the Bengals reach the play-offs. On the other side of the country, the NFC West has an even tighter standings. The Cardinals move past the Rams and Seahawks to win the division.
Conclusion
So while some teams always seem to work magic to win close games, it's unlikely they'll repeat that from season to season. The Chiefs, for example, give the image of an invincible team in these situations, particularly thanks to their play-off scenarios.
However, without an improvement in their level of play, their win total may decrease. Their number of close victories will eventually normalize.
The best teams will always tend to fall on the right side of these close games. On the other hand, it seems wrong in preparing for the following season to be satisfied only with the final results. A team whose win total is positively influenced by its results in one-possession games undoubtedly has a lot to work on to hope for the same record the following season.
