Football Analytics

My son suggested a post on football analytics, despite my lack of knowledge.  But it’s such an interesting topic; perhaps others can share their knowledge.  I’ll provide two references plus some questions.

In 2003, David Romer used dynamic programming to address the question: when should one go for it on fourth down.  The surprising answer was that one should go for it far more often than any team would ever consider it including:  fourth and 1 yard to go on one’s  own 20 yard line;  fourth and 5 yards to go on the other team’s 40 yard line.  A copy of his paper is available at

The New England Patriots have a reputation for using analytics for determining what to do on fourth downs, when to go for a 2-point conversion following a touchdown, and what players to choose in the draft.  Technology review published an article on their use of analytics.   

Here are some questions to which I don’t know the answer, and feel that analytics would help.  I also encourage OR Faculty to supervise theses on this topic.  (Then I could impress my son and others with the relevance of O.R. as a field.  I say this only partly tongue in cheek.)

  1. What is the value of “time of possession?”  It is widely believed that extended times of offensive possession tire out the defense, which is an added benefit.  Can this be quantified?
  2. What is the right mix of plays for 2nd and short yardage?  My conjecture is that a mid range pass (10 to 20 yards) is far more valuable on average than a long bomb, and should be used frequently (say around 1/3 to 1/2 of the time).  Long bombs seem mostly like a waste of a play, but seem to be used frequently.
  3. How valuable would it be if a punter could consistently kick the ball 40 yards and then out of bounds (possibly with a kick that is closer to a line drive)?  Is this something that punters could achieve with lots of practice?  Why don’t punters use this strategy?
  4. What is the probability of a kicking team recovering an on-sides kick if the receiving team is not expecting one?  Is there a way of doing an on-sides kick well so as to surprise the receiving team? 

If you know of research that quantifies the answer to any of these questions, you can add a response on my blog, or you can e-mail me directly.  


One Response to “Football Analytics”

  1. rif Says:

    I wanted to add mention my two favorite football analytics related sites.

    Football Commentary hasn’t been updated much in the last year, but it has a lot of great articles using a more sophisticated dynamic programming model than the Romer one. Quoting their description, the primary difference is that “The Carter-Machol-Romer approach isn’t a backward-induction model, but rather a “steady state” model. Instead of supposing that we try to maximize our probability of winning, they assume we try to maximize the expected point differential in an infinite game.” The Football Commentary model instead has time as an explicit state variable.

    Much more active is Football Outsiders, who maintain a ridiculous wealth of useful statistics for geeking out about football.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: