Written by: Eric Zhou
In Correspondence to Sam Ventura’s class: 36-144 (Winning with Statistics in Sports)
Football is a relatively new sport when it comes to sports analytics - especially compared to another major American pastime: baseball. Football by nature is more difficult to analyze in-depth because of its ever-changing landscape on the field. There are hundreds of possibilities that could happen when you combine field position, number of downs, weather, and much more. And taking into account so many positions that are kept track of with unique metrics, it is always challenging to determine the effect each factor or player can have on the game.
But, baseball has a metric where roughly every 10 runs a team scores over the course of a season equates to a win. This magic number is a rough estimate, but is still profoundly useful in the “big-picture” or “long-run” point-of-view. So why hasn’t football come up with a similar number? That question is hard to answer, but, I can say that with the help of my professor Sam Ventura, we have come up with football’s magic number.
Using data from the last three regular seasons of the NFL, we looked at the score differential (score differential is the number of points scored in a season minus the number of points that were scored against a team in a season) of every team and graphed it against their corresponding number of wins. This graph showed an extremely positive correlation between the two variables, as expected.
The line-of-best-fit above has a slope of 0.030336. Which means that for every unit increase in score differential we have 0.030336 wins. If we want an easier number to understand then we take the reciprocal of our decimal and round it to get 33. 33 is our magic number. Take that Schoolhouse Rock. And so to summarize, over the course of a season, every 33 points of score differential should get a team one win.
This graph shows that if a team’s score differential is 0 then their record should be 8-8, and that roughly every 33 points from then on will get 1 more win. So a team with a season score differential of 100 should be going 11-5. An important thing to note, however, is that this doesn’t mean a team needs to win each game by 33 points, that’s extremely unlikely. It means that over the course of a season, a team needs to score or prevent a total of 33 points to snag one more win above .500.
Since 2002 over 400 games have been won by 3 points, and more than 250 have been won by 7. That equates to roughly 26 percent of games being won by a field goal or touchdown. (info. From Chad Langager, sportscharts.com)
Looking at football as a game of maximizing gains and minimizing losses takes away much of its romantic aspects in place of a calculating way to win. So to recap, for roughly every 33 points of score differential a team has over the course of a season that team should have one win. This stratifies the contribution of offense and defense to a team’s chances of victory even more. Applying this knowledge to the game would mean that player recruitment could be completely changed. Now, coaches and managers can look at specific statistics when looking at potential players. A more clear-cut method of evaluating players and their contributions could be made. And the ripple effect would affect contracts, salary caps, and even superstar status. Salaries for players would change, possibly increasing overall which would change the value of the salary cap that the NFL put in place. And players that regularly would not get much media coverage and reputation overall could be recognized for their unsung contributions to the team as the new system would make clear who and who is not performing well.