Decision making under cognitive pressure: examples from football

One thing I’ve long thought football teams should do is to have somebody on the sidelines whose only job is to tell the head coach what to do in terms of down, score, and clock management.

Nearly 20 years ago David Romer provided an elegant mathematical demonstration of the fact that football coaches were vastly too eager to punt the ball away rather than trying to convert fourth downs into first downs, given the probabilistic outcomes that flowed from those decisions. (Jon Chait and I interviewed Bill Belichick about Romer’s article, which he had read, and subsequently used to become one possibly the first NFL coach to start breaking convention on this score on a regular basis).

Coaching has improved in this regard in recent years, but coaches still make all sorts of questionable to obviously wrong decisions in this area.

Here are three examples from yesterday:

(1) In the Michigan-Ohio State game, OSU coach Ryan Day twice kicked extra points after drawing within nine points of Michigan in the fourth quarter. This is a mistake coaches at all levels continue to make over and over again: In a situation where it’s obvious that you’re going to have to convert a two-point try at some point if you’re going to tie the game, coaches don’t go for two at the first opportunity: instead they wait until there’s no choice but to go for two. This is just obviously the wrong decision: the whole point of going for two early in these circumstances is then you know what you have to do going forward, depending on whether you made it or not. But coaches prefer to put off what seems to them to be an irrevocably fateful outcome for as long as possible. This is pretty much the same cognitive bias that leads them to punt on fourth down rather than going for it.

(2) Late in the game Day broke from conventional thinking, but in a way that was almost certainly the wrong decision. Michigan had just made a first down on the OSU four, up by eight points with three minutes to go. The clock started running, and instead of immediately using one of his two remaining time outs, Day chose to allow Michigan to run it down to 2:20 to go, and then ordered his defense to let Michigan score, in order to get the ball back. This was thinking outside the box, but in a way that was almost surely worse than doing the conventional thing, which would have been to immediately call a time out, and try to stop Michigan from scoring. Now admittedly the likely best case scenario if OSU had done that would have been to get the ball back with about 1:30 to go while down 11 with no time outs left, so it’s not like it made some huge difference at the margin. But running backs fumble, chip shot field goals get blocked occasionally etc. It was a creative move but a bad one on the margin.

(3) By far the most extreme example of bad decision making under cognitive pressure yesterday was Wisconsin’s Paul Chryst’s decision to punt on fourth and one from his own 20 down by ten points with 4:30 to go in the game. This is just an insanely bad call. Now Chryst is not a dumb guy, so I think this is a clear example of somebody doing something that, if he wasn’t thinking about 18 different things at the same time, he would never do. This is exactly where you need Down and Distance Nerd on the Sidelines to tell the coach what to do. (In fact Wisconsin had a penalty on the punt, and after having a minute to think it over Chryst decided to go for it on fourth and six, which was obviously the correct call, but only highlights how terrible his initial decision was, and how he would never make that decision if he were not under extreme cognitive strain).

It’s also a classic example of how coaches are so strongly impelled to try to delay the moment that a loss becomes certain, rather than doing what you need to do now to maximize the probability of winning. That’s what has always driven the conventional wisdom about punting on fourth down, and it’s an obvious cognitive bias that’s extremely suboptimal. Having somebody around whose only job is to counter that kind of bias — also evident in Day’s decision to twice not to go for two in situations where that was without question the wrong decision at the time — would make a lot of sense.