Stanford Graduate School of BusinessEp82 “Should NFL Teams Be More Aggressive on Fourth Down?” With David Romer
Stanford Graduate School of Business
Setting the question 0:06
The hosts introduce the NFL as a case where competition should push teams toward the best choice, yet they suspect teams still get important decisions wrong. They explain that the episode will focus on David Romer’s paper and on fourth down, when a team must choose between going for it, kicking a field goal, or punting.
How football works 2:01
They lay out the basic rules for listeners who do not know the game. A team has four downs to gain 10 yards on a 100-yard field, and if it succeeds it gets four new downs. A touchdown is worth six points, with an extra point after the kick, while a field goal is worth three. If a team fails, the other side takes over at the spot where the play ended, and a punt gives up the ball but pushes the opponent back downfield.
Paper and surprise 16:32
The discussion turns to how the paper came about and what was surprising in it. David says he already expected the broad result before doing the full dynamic programming. Simple examples in the paper were enough to make the point. The main surprise was not the direction of the answer but the numbers. Fourth-and-one was not as automatic as many people assume, and even fourth-and-ten had a meaningful chance of success. That made going for it on longer distances more sensible than it first sounds.
Why teams changed 19:31
The conversation then asks why NFL teams changed their fourth-down approach. David points to three forces. Competition should eventually reward the better choice. Analytics became more common across sports and teams began hiring staff for it. Then Super Bowl 52 in February 2018 gave the issue a visible push. The Eagles converted an unconventional fourth-and-one in their own territory and later used another bold call on fourth and goal. After that, teams seemed to wake up and question old habits.
Paper and reaction 32:00
The paper became a way to show how dynamic programming and linear algebra can answer a question many people would dismiss. Teams mostly did not adopt the recommendation, and the usual reaction was to point to what the analysis left out and then stop there.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
