Warning: This post is about football. The ulterior subject is the power of conventional wisdom in decision-making, but to get there I’m going to wander through some pretty dense thickets of football wonkery. So if the sport bores you to tears, or alarms you, feel free to leave the stadium now.
An article in the sports section of the New York Times assessed the increased use that a few NFL teams have made of the two-point point-after-touchdown (PAT) in the last two seasons. For the non-fans still reading, football provides the offensive team a choice for an extra, grace score after every six-point touchdown: either kick the ball for one extra point, or use your regular offense to get it across the goal line, via a pass or run, for two points. Traditionally, few coaches have used the two-point option; the success rate of the short, one-point kick was so high – over 99% -- that coaches opted to pocket the sure one point and disregard a riskier chance at two. Then, two years ago, the league made the choice more interesting. It moved the starting point for the PAT kick back from the two-yard-line to the fifteen-yard-line, making the kick longer. This uptick in difficulty has resulted in a downtick in efficiency, with kickers now succeeding only 93% of the time.
That may not seem a significant decrease, but it’s enough, in fact, to alter the risk-reward calculation on the PAT. Would you rather have a 93% chance at one point, or a 57% chance at two points? (That’s the percentage of success achieved by the handful of teams that have implemented the two-pointer to some degree.) The math yields a 0.93 points-per-attempt ratio via kicking, and 1.14 points-per-attempt via “going for two.” Since such a differential adds up over the course of a game and a season, you’d think coaches would be hastening to switch to the two-point approach. Or, at the very least, that it would be a tossup.
But you’d be wrong.
In fact, very few coaches have embraced the two-point option; only around one in every ten touchdowns is followed by a two-point PAT attempt.
The answer goes to the mistrust of innovation and the powerful role played by conventional wisdom. Sports are said to reveal character, and I believe they do. But sports and games also shed light on how we make decisions – thus the academic discipline of game theory – and on the particular habits of analysis that can prove errant... yet hard to revise. Football, far more than baseball and basketball, is coach-driven, with an enormous emphasis on the strategic decisions, made by coaching staff, that direct each play to its outcome. As such, it’s particularly rich in situations that reveal the power of strategic conventional wisdom in shaping decision-making.
Take punting. The punt (again, for the gridiron neophyte) is the fourth-down play a team uses to unload the football when it has failed to get the requisite ten yards in the prior three attempts, or “downs.” The punt allows you to kick the ball a great distance before it is caught by the opposite team; your team relinquishes the ball, but the opponent must begin its next offensive drive from far down the field. The alternative to punting is to use your fourth (and final) down to attempt to gain whatever yardage you need for a first down that starts the cycle all over again. The problem is, if your team “goes for it” and fails to get that requisite yardage, the other team takes over the ball not far back down the field, but right there. This has been a risk very few coaches have been willing to take; when you have, say, fourth down and one yard to go on your own twenty-yard-line, conventional wisdom, tradition, and a coach’s deep gut instinct all say, Punt.
But is that the correct decision? A few years ago, a coach at my state’s university, UConn, forayed boldly into counterintuitive football. UConn was leading heavily favored Pittsburgh, 30-28, with 2:50 left, and faced a fourth-and-1 on its own 19-yard line. The standard move in this situation, as I noted above, is to punt — since if you don't get that first down, the other team gets the ball right there, and instantly has a short field goal (3-point kick) to win the game. Indeed, most football people consider going for the first down in this situation downright crazy. Why hand your opponent the victory? Almost any football fan reading this will concur.
The UConn coach, however, defied conventional wisdom and went for the first down. And it worked. A running back gained four yards on the play, triggering a new first down. UConn then ran out the clock to end the game without Pittsburgh ever getting the ball back, and afterward the coach won praise for what one sports reporter called “the gutsiest call in his career.”
But was truly it so gutsy, statistically seen? Let's weigh the probabilities. (Warning: here is where the real statistical wonkery begins.)
If you “go for it” on your own 19, what are your chances of winning the game versus losing it? First, assume your team will succeed 75% of the time on fourth-and-1 (the college average). That means a 25% chance of turning the ball over on downs. If this happens, assume the other team has an 80% chance of making that field goal. So going for it fails 25% of the time, and in 80% of those cases you lose the game – or 20% overall. We need to subtract a marginal percentage from these loss scenarios — say, 2% — for the small chance that, even if the other team kicks that field goal, your team roars back for a last-second win. So here is Scenario 1: Going for it means you lose the game 18% of the time and win 82%.
Now look at Scenario 2, punting. A typical college punt from the 19-yard-line, with an average return by the other team figured in, puts the ball at midfield. What are the chances your opponent can get 30 yards or so and kick a field goal? Fifty percent? Forty? Even if we choose a highly conservative 35% — and again subtract 2% for your team's possible last-ditch score — you're still facing a 33% chance of losing the game if you punt. Bottom line: Punting is almost twice as likely to lose you the game as going for the first down. The UConn coach’s choice was no wild gamble, in other words, but strict rationality based on probability.
And yet almost no coach will do it. And I mean ever.
How can conventional wisdom be so out of whack with statistical reality? Keep in mind that a heavily statistics-driven approach to coaching is still pretty new in sports. It has made big inroads into baseball, via the pathbreaking work of numbers guru Bill James and the efforts of such forward-looking general managers as Billy Bean (celebrated in the book and film, Moneyball). But in football, not so much. I think that’s because big-play stakes are so much higher in football — and decisions about those big plays rest wholly in the hands of the coach. So therefore do the consequences. Football, much more than baseball, gives you situations in which fans walk away bitterly saying, "The coach lost the game, and did it on that play." (Ask Seattle Seahawks coach Pete Carroll about the ending of Super Bowl XLIX against the Patriots). Throw in a gotcha!-oriented media, a demanding front office, and a general non-understanding of statistics across all these sectors, and you have a recipe for the conventional wisdom that shrieks, “Go for it on your own 19 — are you crazy?”, or “A two-point conversion, now – are you nuts?” Decision theorists call this kind of decision high-risk/low-reward. The coach may get a sportswriter's accolade if it succeeds, but he faces hell on all sides if it fails.
Humans are risk-averse; we tend to fear losses more than we value gains. Expounding this theme in the Times, Brian Burke, an analytics specialist at ESPN, cites a behavioral model known as prospect theory; essentially it holds that any coach who takes this kind of risk will be two times more disappointed by failure than gratified by success. Burke notes that “It’s easy to put an equation on paper and prove it, but when there’s 70,000 fans screaming and the pressure’s on and your job’s on the line, it’s a very human thing to just fall back on the tried and true.” A coach's gamble in this situation, in other words, is not about game strategy, but job security; the real risk-reward calculation isn’t merely a matter of maximizing scoring, but rather of doing so in a way that makes sense to the people who control your job. If the punishment for violating conventional wisdom is too harsh, you won’t do it. Fear of getting smacked for being heterodox rules the day.
Ten years down the road, I predict that going for two points on the PAT and not punting on short-yardage fourth downs will be the default strategies. But how to get there? We need a few bold coaches to embrace a geeked-out, Freakonomics approach to football and buck the conventional wisdom. Here is where decision-making in sports mirrors decision-making in business and politics. The lesson may be that if you’re going to be innovative, you’ll need to be fearless. Otherwise, when you try to think outside the box, the box will win.