Driving Question

It's football season again, so let's talk about professional basketball for a minute.  That's right, basketball.  In order to make games more exciting, NBA rules don't allow teams to hold onto the ball for too long during any given possession.  Currently, the shot clock counts down 24 seconds, meaning that's the amount of time a team has to attempt a score before they're forced to turn the ball over.  That's the law of the land.

Now imagine everybody gets together next week and agrees that they'll instead turn the ball over after 18 seconds.  Not that the rule will change — there will still be 24 perfectly good seconds available for legal use — but that the new prevailing convention will suddenly be to give up 25% of your available possession time.  Everyone would think that's nuts.

And yet that's more or less what happens in every single NFL football game.  A team has four chances to advance the ball a minimum of 10 yards, but if they haven't succeeded after three plays, they typically sacrifice the last down in order to punt the ball over to the opposing side.  Pretty much everybody thinks that's reasonable.

This is a weird situation.  Why do coaches so willingly hand (foot?) over the one truly valuable thing in the game of football, i.e. the football?  Conventional fourth down wisdom goes something like this.1  If you fail to pick up the required yardage, then the opposing team gets the ball in roughly its current position.  If you punt, the opponents still get the ball, but generally much farther away from your own goal line, which makes it easier to defend.  Ostensibly it's better to have this defensive cushion than even the possibility of further offensive possession.

If you think that seems pretty risk-averse for a game where hypertrophied men routinely smash their faces together with brain-bruising force, you're not alone.  Back in 2005, UC Berkeley economist David Romer published a paper that examined the first quarters2 of 732 regular season NFL games and noted the outcome of every single play.  Based on his analysis, Dr. Romer was able to assign values to different field positions: given a particular starting situation, how many points, on average, the offensive team earned over the course of a drive.  Armed with these expected point values, he claimed teams punt too much.  Way too much.

Without getting into the nitty-gritty about calculating expected points, they essentially represent a weighted average based on the different types of scores (touchdown, field goal) and the probability of scoring from different places on the field.

Here's a graph showing two quick example situations: fourth and two from your opponents' 40-yard line, and fourth and five from your own 40 (you can toggle back and forth with the button).

A conservative coach might punt in the first cast, and almost everybody would punt in the second case, barring a desperate end-of-game situation.  But notice that, in both cases, punting actually leads to a worse result on average!  It would be better to take the risk of going for it.

Think this is all  hand-wavy  professorial nonsense?  Fair enough.  Listen to Kevin Kelley, head coach of the Pulaski Academy Bruins in Little Rock, Arkansas.  He's taken this strategy to the extreme...and to the state championship.  Notice who embraces this strategy immediately, no mathematical proof required: the kids always want to go for it.  And they do.

Teachers, want to have this conversation in class?  Check out the Driving Question lesson materials on our site!


1.  Assuming your field position is such that a field goal isn't a realistic option.

2.  Since, in practice, coaching decisions can be heavily influenced by specific game situations (the score, how much time is left in the half/game), restricting the analysis to early-game plays helps remove some complicating variables and leads to a more general — if slightly idealized — result.


Leave a Reply

Your email address will not be published. Required fields are marked *