The use of analytics by NFL coaches has gotten a great deal of attention in recent weeks—much of it willfully and even aggressively uninformed. Consider this from Phil Simms on CBS Sports, talking about the Baltimore Ravens attempting a 2-point conversion in their Week 14 loss against the Cleveland Browns, rather than kicking the extra point to gain a tie. Said Simms: “We don’t put that into the analytics, I don’t think. I don’t know, I don’t care.” Specifically, Simms referred to how the rest of the game unfolds after a particular play. Does the decision to attempt the 2-point conversion account for the changes in the way the game is played after the conversion attempt if it is unsuccessful? That is a useful question to ask of the people doing the analytics for teams and, in fact, the answer may be yes.
David Romer, an economist at the University of California, Berkeley, used a method known as dynamic programming to study the decision to go for it on fourth down. The method explicitly incorporates what might happen after a decision in determining which is the best alternative for the decision. Romer specifically studies the case of a team with possession of the ball on the opponent’s 2-yard line, facing fourth down. His estimate is that trying for a touchdown raises the probability of winning the game by about 3 percentage points. (Of course, when the opposing quarterback is Tom Brady, that 3% may be overestimated, as the New York Jets learned last weekend.) He also shows “that rational risk aversion about points scored, concern about momentum, and other complications do not noticeably affect the case for trying for a touchdown.” To be clear, Romer’s analysis did consider how the remaining game changes, and there is no reason to assume that the analysts for NFL clubs don’t also do that.
Perhaps, if he cared to know, Phil Simms could ask.
Dennis Coates is a professor of economics, specializing in sports economics, at the University of Maryland, Baltimore County (UMBC).