It's funny how the analytics of a sport can often mirror the sport itself. Sabermetrics are as regimented (or more) as the national pastime ever was; basketball analytics have progressed into lively, intersecting data points thanks to new player tracking technology; and heady football talk often enough comes down to whether or not a team or coach should go for it.
Moment-to-moment tactics in a football game are largely inscrutable to most fans, even the most hyper-educated among them. You might notice a defensive line stunting often, or a tight end running delayed routes, but it's usually easier to pick out quality of play (a defensive lineman blowing past the o-line) than it is to see the framework within which this is all happening. The most visible assertions of decision making are the hero calls. Fourth and 2 was Bill Belichick's Little Big Horn; Sean Payton stole the golden fleece out from under Peyton Manning with an onside kick.
In yesterday's game against Denver, San Diego scored with 10:47 left in the fourth quarter to cut Denver's lead to 28-19, before the extra point. CBS's broadcast booth flashed a graphic saying that "The Book" says to go for two in the situation. Phil Simms burst into his nerdstomper dance, and Jim Nantz came along with him. "If that's what the book says, take the book and burn it," Nantz said. No curiosity about why the book might say that, or what the book is actually saying. Just unexamined disdain for numbers that couldn't possibly be right.
The actual math for the situation is hazier than some clear-cut scenarios. With 12 minutes remaining and down nine after the score, the basic calculation suggests you need about a 39 percent chance of converting to make going for two a winning proposition. With 9 minutes remaining, 38 percent. (The original "chart" by Dick Vermeil just says go for 1.) Denver would remain a heavy favorite regardless. If the score had remained a 9-point Denver lead, its Win Probability would have been 90; at 8 points, 88; and at 7, 84.
F0r all normal two-point conversions between 2000 and 2009, the success rate was 47.9 percent—though if play calls were distributed more evenly, the case could be made that that number would float north of 50 percent. Regardless, that's case closed for The Book, right?
Here's where many of the day-after polemics about this decision or that break down. Can you say with certainty that a team's chance of converting a play is 40 percent, rather than 39? What about 43? And what goes into those calculations, exactly? Even the most thorough looks at the odds of converting allow that there's some amount of bias to the numbers, and that if the offense's tactics changed, the defenses would adjust.
That's not to push off the idea of using all of the information available to make an informed decision. It's just that the overselling of the case—by both sides—is grating. There are dozens, hundreds of things that can go into a coach's estimation of his team's chances of converting a two-point play. Some of those can be unhelpful bits of old wisdom. And while I'm not sure how many coaches have information on their team's offensive success rates on types of play, broken up by down and distance, there are probably a few without it. Still, maybe just as important to the decision are things like the relative health of the left guard, how well the team has been blocking the B gap on short yardage, and, maybe most importantly, how damn tired his team is right now. Some of those aren't things that someone on his or her couch, or sitting up in an announce booth, necessarily knows. Some of those aren't things they can know.
On the surface, this elevates the micro over the macro. But as long as the bottleneck for these decisions is a percentage value of how likely a play is to succeed, I'm not sure you're ever totally going to get away from a coach's gut being at the center of these decisions. The important thing is to make sure it's an informed gut.
The problem, really, is the entire idea of "The Book" or the "Chart." It makes advancing the tactical philosophy guiding these high pressure calls appear to be a matter of passive acquiescence. Just sit back and read the chart, coach, it seems to say. Just remember that we're talking in percentage points and probabilities, and that even when these are based on the best available information, that will always, necessarily be incomplete, whether it's dealing with abstract numbers in a simulated game environment, or a fourth quarter decision on a two-point conversion with tens of thousands of people baying in their seats.