Last year, I wrote a piece on why penalties don't matter much. They can matter greatly in certain situations, like if a block in the back negates a punt return touchdown or a hold calls back a 60-yard run. However on the whole, penalties do not correlate much with winning. Just look at that piece I wrote or Matt Hinton's recent bit on passing efficiency (which mentions penalties in its table of stats).
So with the matter of penalties as a whole not meaning much being pretty well settled, what else is there to write about? Today, I'm trying to tackle why it is that penalties don't matter much. I've seen a number guesses as to why it is, but not a whole lot of real analysis. If you've seen something, please let me know in the comments. Let's start with the top answer.
As far as I can tell, this is the canonical answer as to why penalties in aggregate mean little. Good teams tend to be aggressive, and aggression can lead to penalties.
That's not a bad answer, but I've never been satisfied with it. How aggressive is a substitution infraction? Was the left tackle holding the defensive end because he was being aggressive or because he got beat? Did the receiver line up incorrectly out of aggression?
Being aggressive might explain some penalties, but some teams are plenty aggressive without committing many penalties and vice versa. I want better explanations.
Yards per play
Let's focus on offense for the moment. College football had 1,294 discrete teams across the last 11 seasons. Only 369 of them (28.5%) averaged fewer than five yards per play. Just 20 of them (1.5%) averaged fewer than four yards per play.
That first stat means that for roughly three quarters of college offenses, the expected value of running three plays already overcomes a five-yard penalty. For most of the rest of them, they would need to manage just one extra yard per play than expected to overcome a five-yard penalty.
Of those 1,294 teams, only 474 of them (36.6%) averaged fewer than 6.67 yards per pass attempt. Just 203 (15.7%) of them averaged fewer than six yards per pass attempt.
That first number means that roughly two thirds of college offenses could be expected to overcome a 10-yard penalty merely by passing the ball three times. The second one means that for a lot of the rest of them, they would need to get just one extra yard per pass attempt to overcome a 10-yard penalty.
Obviously the degree of difficulty for an offense increases when it commits a penalty. However, overcoming that penalty to get a first down is not that daunting when you break it down like this. Getting a 15-yard penalty and racking up multiple penalties are still not recommended, but only the very worst offenses are rendered dead in the water by five- and ten-yard penalties. And certainly, committing penalties when facing a great defense is much worse than against a poor defense.
With most offenses having the capacity to overcome a lot of their own penalties, it's not reasonable to expect most defenses to overcome their own. When it comes to defensive penalties, damage control is the bigger question.
Unfortunately, no college football stat source keeps numbers on drives. I did my best to approximate defensive drives for the teams in the last five seasons by adding up TDs allowed (without including kickoff/punt return TDs), field goal attempts, turnovers gained, punts forced, and fourth down conversions stopped. I then added an adjustment of 0.75 per team per game to account for safeties and drives ended by time running out in a half. I also calculated out a points per drive figure by assigning seven points to each touchdown (I know that's not exact, but it's close) and three to each field goal made.
Based on that, the average team in the last five years had 12.9 drives per game and scored 1.9 points per drive. When a defense commits a penalty that leads directly to a first down or sets it up as a gimme first down, it's like taking a drive that perhaps was going to result in zero points (because of a stop) and resetting its chances for scoring. Imagine the worst-case scenario: a stop negated by a penalty. On average, you just gave the offense an expected value of 1.9 points.
Here's the thing. In the 7,630 college football games between I-A teams since 2000, only 565 (7.4%) were decided by three points or less. Only 1,158 games (15.2%) were decided by seven points or less. Now the margin in the game at a given point affects strategy and will make a difference in the final score, but with most games not ending up decided by a slim margin, a couple defensive penalties here and there are not going to be the deciding blow.
If you toss the number of first downs a defense allows into those drive numbers, you get roughly the number of sets of downs that defenses face. The NCAA publishes first down stats for the last three seasons, so that's all I got there. Anyway, first downs caused by penalties account for the outcome of 4.52% down sets on average. That's not a tremendous figure, and while it doesn't account for penalties that make first downs easier without causing them directly, it shows how relatively rare penalties causing first downs are.
No stat source I've seen provides penalty data broken down by offense, defense, and special teams, and none provide a breakdown of how many are five yards, 10 yards, 15yards, and half the distance to the goal in length. What's here is about the limit of what I can do as I don't have play-by-play data.
However, I hope it casts some light on why it is that penalties have a very weak correlation with winning percentage. It's not just that the best offenses stand a good chance at overcoming the five- and 10-yard penalties, but nearly all offenses stand a decent chance at overcoming them unless they're facing a top flight defense. It also has to do with the fact that defenses not getting a stop on any particular set of downs doesn't automatically lead to disaster.
So if your team commits a really dumb penalty this fall, or several in a row, go ahead and get mad. Just don't fret if you look at the stats and see your team among the worst penalty offenders in the country. That simply doesn't matter much.