By Barrett Hansen

As we saw in the Baltimore Ravens vs. New England Patriots game last weekend, facing a fourth down in opponent territory but outside the 35-yard line can make for quite the coaching quandary. You can run out the kicker for a long and improbable field goal attempt that, if missed, would leave your opponent with excellent field position. You can go the safe route and punt the ball away, hoping to pin your opponent inside the ten. Or, given the high likelihood of a touchback, you could roll the dice and go for it, since a failed conversion would give up just 20 yards of field position. Finally, since many punters have trouble with short kicks, you could try to draw the defense offside with the hard count, knowing that at worst you give your punter a bit more space to pin the other team deep.

Baltimore mixed it up with their decisions. Facing a 4^{th} and 5 from the NE 38 in the first quarter and a 4^{th} and 6 from the NE 36 in the third, they elected to punt (resulting in NE ball at the 21) and go for it (converted to NE 1). New England for its part had one 4^{th} and 10 from Baltimore’s 37 in the 3^{rd} quarter and elected to punt, which resulted in Baltimore receiving the ball at the 20. Given how unsatisfying a punt so close to field goal range is, we at HSAC decided to test how to approach this decision.

The decision of whether to go for it been covered before (here as well), with the vast majority agreeing that teams should go for it more on fourth downs. However, thanks to numerous *Thank You for Not Coaching* columns from HSAC friend Bill Barnwell, we know teams are unlikely to be as aggressive on fourth as they should be.

So let’s assume teams are going to punt. The question then becomes where to punt from, since teams can take the delay of game to create more space.

This graph plots a two period moving average (i.e. the value at 35 is the mean of all punts from 34-36 yards) of opponent starting field position against original fourth down field position for the kicking team. The teal line charts punts without a delay of game, while red plots punts taken after a delay of game penalty. Important to note: the x-axis measures *original* fourth down field position (so the punts were actually taken five yards back from there). The y-axis is measured in terms of yards from the opposing end zone (the 90 is the 10, etc).

When considering explanations for the shape of this graph, we must consider selection bias in punters; ostensibly the punters tell the head coach what distances they are most comfortable punting from, which informs the coach’s decision. So the difference from the 35-39 can be partially explained by some punters being “short punt specialists”—if you will—and others “short punt duds”. The punters whose teams took delays are likely to be duds, since otherwise they would have just punted from where they were. The fact that there’s still some gap suggests that some short punt duds might still be too close even after one delay of game (possibly suggesting the need for a false start on top of the delay).

By the 40, the coaches are making efficient decisions with their delay of game penalties; the delay is putting the punter in a spot where he can pin the return team right where league average punters would have from the original yard line. However, go past the 44 and we pass the limits of the punter’s range, which gives the return team an advantage again. As you can see, by the 45 taking a delay costs about five yards, which is exactly what you’d expect from a five-yard penalty.

Before we conclude that you should only take the delay of game between the 40- and 43-yard lines, we should introduce the possibility of drawing the defense offside with a hard count, which—if the distance to go is five yards or fewer—would result in a fresh set of downs for the offense. Let’s place the ball at the 36, where taking the delay of game has the maximum penalty. Returning to AFA’s Expected Points framework, a punt from here on average gives the other team the ball at the 11, which is worth -0.16 points for them (0.16 points for you). Drawing the defense off with the hard count gets you a 1^{st} and 10 from the 31, worth 3.24 points. Failing to draw them off and taking a delay of game gives the return team the ball at the 15 (worth -.11 points for you).

Bringing out the offense therefore represents a possible gain of 3.08 points versus a possible loss of 0.27 points; this places the breakeven probability at 8%. Over the past five years, defenses jumped offside 9 times at this spot on the field, compared to 52 delays of game. This equates to a 17% rate, more than double the required 8%.

Though the sample is quite small, it seems that for now coaches should always at least set up their offense on fourth and short. This strategy doesn’t necessarily have a diminishing marginal return, as mixing strategies becomes advantageous; a team that gets a reputation for trying to draw the other team off could actually catch a team by surprise by running a play. This in turn would help the hard count’s effectiveness.

For those of you thinking we’re splitting hairs here, I have two responses for you: 1) The NFL is a game of inches where every edge matters, and 2) This argument is pointless because teams should just be going for it in these situations.