Revenge of the nerds
The NFL postseason contained lots of frothing about modern analytics, but the calculators-and-spreadsheet set hasn't won the battle yet
The post-Super Bowl week tends to contain a sifting-through-the-wreckage vibe to find the narratives to carry us through the early part of the NFL offseason.
This one, mercifully, had plenty: the comparisons between the Chiefs and the Brady-era Patriots, the questions over whether 49ers coach Kyle Shanahan actually knew the overtime rules, the fact that Travis Kelce has anger-management issues.
I used “mercifully” back there because the existence of all this stuff to feed the sports-media beast meant we were spared what might have been a days-long Fourth Down Discourse.
Dodged a bullet there, is what I’m saying.
The Fourth Down Discourse is a thing to be avoided simply because it is dumb. As with almost anything to do with sports and analytics, and even though we are more than a decade into their widespread use, it is still a remarkably short distance from a thoughtful discussion of such things to The Nerds Are Ruining My Sport and You Can’t Measure Heart.
How dumb is the Fourth Down Discourse? When Detroit Lions coach Dan Campbell was burned by two go-for-it calls on fourth down in the NFC Championship Game, he was derided as being a Poindexter beholden to his precious calculator. But: Have you seen Dan Campbell? I am certain that in his day he tossed a nerd or two into a dumpster, or at the very least stuffed them into a locker.
We’ve probably also been spared the worst of the Discourse this week because Shanahan went anti-nerd at the end. Facing 4th-and-four at the Kansas City nine-yard-line halfway through overtime, he opted for the short field-goal attempt and ended up with a three-point lead. Very old-school.
Had Shanahan gone for it and failed, allowing the Chiefs to win with any score, the calls for his head would still be echoing. Entire television segments would be devoted to the malign influence of math on football. Even though going for it in that scenario would have been entirely defensible.
Risky? Yes. But based on sound logic? Also yes.
You will note that I have not proclaimed that going for it was the “right” call. From a pure numbers sense — expected points added, win probability, the nerd stuff — it was a clear go-for-it situation. But having just seen the Chiefs stuff Christian McCaffery on second down for no gain and then pressure Brock Purdy into a rushed throw and never-a-chance incompletion, Shanahan might have felt that he didn’t have a play in his holster that he was confident enough would get the required four yards. I can’t kill him for that. (Shanahan said post-game that he never considered going for it at that point, because the Chiefs were going to get the ball back whether his team scored three or seven. That explanation makes less sense. Wouldn’t you rather force them to match your seven?)
But it’s also notable that Shanahan did go for it on 4th-and-three early in the fourth quarter while trailing by three points. They converted, and scored a touchdown two plays later to take the lead. The successful conversion was worth more than 2.5 expected points added, the most valuable play of the second half for the Niners offence that wasn’t a scoring play.
That is, he trusted the math in the fourth quarter, but didn’t in overtime. Which is the funny thing about the Fourth Down Discourse: the nerds aren’t winning nearly as much as the average fan thinks.
Here is a chart of 4th-Down Recommendations in typical game situations, compiled by the pointy heads at ESPN Analytics:
Note that the go-for-it call is favoured on short yardage almost all over the field, and close to the opponent’s end zone the “go” call is favoured even when there are several yards to gain. There are complex calculations behind all these kinds of decisions, which I will not bore you with and also could not explain if I tried, but the gist is that the model recommends the decision that most increases a team’s win probability on a given play. So if trying a field goal adds 2.9 expected points and trying to convert 4th-and-2 adds 5.1, which would lead to a significant jump in win probability, you get the big GO sign.
But even though NFL coaches are using this kind of guidance a lot more in recent seasons, they aren’t close to giving themselves over to it. If they were, no one would ever punt on fourth-and-short, no one would ever attempt close-range field goals, and no one would ever punt from inside the opponent’s half. Those things still happen with regularity. (Dan Campbell excepted.) Depending on whose numbers you crunch, coaches are going for it on 4th maybe half as often as they should. That’s not exactly a resounding victory for blind faith in the numbers.
That Campbell was widely roasted for his failed fourth-down calls — even though an open receiver dropped an easy catch on one of them — and Shanahan mostly escaped blame for settling for a field goal in overtime shows that the old-school way of doing things remains the safest decision. Even when it’s not the optimal one.