A few years ago I happened upon a now forgotten web site that attempted to prove that yards don’t matter in football. All that mattered was points. My first thought was, “Well, yards are how you get points most of the time”.
Turns out I was right (for once).
The site went on and on with examples of how and why yards don’t matter. As I recall it was the heyday of the Honey Badger and LSU seemingly won week after week by scoring on special teams and defense.
Great, that’s one team out of 120 (at the time). For the vast majority of the other 119 teams and in the vast majority of games yards do matter. As a matter of fact, they matter more than anything else.
For 2011 and 2012 in games between FBS teams (1,404 games in all), 76% of the time the winning team has gained more yards than the losing team.
It’s not rocket science.
What I didn’t analyze at that time was the correlation between yards and points scored.
Points seem kind of important, so I decided to find out which of the statistics I track most likely leads to points.
Out of the 12 (other than points scored) numerical categories I track, there is no greater correlation to points scored than total yards and, with the exception of yards per play, it’s not even close.
Using the Pearson Coefficient I came up with a correlation of 0.766324 between total yards and points. You can read more about the Pearson Coefficient here, but the basic concept is a coefficient of 1 means there is a perfect positive correlation between the two variables and -1 means there is a perfect negative correlation between the two variables.
What does this tell us? There’s not a better predictor of points scored than yards gained.
Anything above 0.5 can be considered a high correlation.
The interesting numbers to me are the ones that are typically associated by coaches, announcers and writers (and bloggers) as important to winning with low correlation to points scored like plays run (technically this falls in the medium correlation range), time of possession and penalty yards.
Remember that correlation should not be confused with causation. For example, the number of passes thrown has a low (and negative) correlation to scoring points, but the number of passes a team throws may not necessarily be the cause of the number of points scored.
Finally, when I do an analysis of this type invariably someone asks, “What about defensive statistics?” Defensive numbers are included in these statistics in this way: For every yard an offense gains a defense gave that yard up. For example, one would expect a given team to have less yards against Alabama and therefore less points. If an offense averages 500 yards a game and then plays Alabama, for example, they’re likely to gain less yards and score fewer points.
What we know from this analysis is that over the last two years a point is scored (or given up by a defense) for every 14.2 yards gained (or given up by a defense) on average.
This analysis included both the statistics of the winning and losing teams in 1,404 games, 2 teams in each game and 2 variables (points and total yards gained) for each team per game.