Which Plays Are Luck?

There are a lot of big plays that occur in football that aren’t a repeatable representation of a team’s skill. The biggie that is always thrown out is the luck involved in recovering fumbles. Who recovers a fumble will alter the course of a game, but teams show no consistency in actually being good at this “skill.” I’m willing to take it a step further. Turnovers in general are almost entirely luck. Sure, you’ll get quarterbacks who are very careful with the ball like Tom Brady and Russell Wilson, and these guys might have a slightly consistent skill in preventing turnovers, but the fact remains that the correlation between turnover margins in current vs. previous seasons is practically nill. The r^2 value is about .02, meaning that only TWO PERCENT of a team’s turnover margin in a given season can be explained/predicted by their turnover margin in their previous season. This implies that turnovers are ALMOST ENTIRELY LUCK.

Speaking of Luck, let’s use him as an example for my next point. Here’s a quarterback who last year showed a very strong disposition toward being careless with the football, turning the football over more than almost anyone else in the league (which can be expected to regress at least partially toward the mean next season), but that’s beside the point. He threw 14 passes last season in which the defense dropped a potential interception (that’s extremely high, even given how often he throws to defenders, in case you were wondering). This is a case where nobody in his right mind would believe that Luck actually has a skill in throwing passes that are hard for defenders to catch. However, all of the Colts close wins are automatically attributed to Luck’s skill. I wonder how many of those games would have gone the other way had defenders been able to hold onto the ball a reasonable amount of the time.  Clearly, wins, and especially close wins, are a terrible measure of skill, given how influenced they can be by matters of randomness, and to me, the bonus we give to our judgment of teams’ skill for winning (even more so for the credit we give to QUARTERBACKS winning) is one of the biggest mistakes we make in our sports opinions in general. This is why my algorithm completely ignores wins. It’s just not worth keeping track of random noise (By random noise, I mean wins, and when I discuss the worth of keeping track of them, I am referring to my effort to measure how good teams are.) when we have the actually meaningful statistics that can explain what causes teams to be winners in the long run.

Despite the importance we place on turnovers and especially wins, I think the hardest “luck play” for us to accept is the touchdown. Get ready for it. This is the case against touchdowns. Now, don’t go crazy just yet. I’m not saying there’s nothing to the ability to put the ball in the endzone or that teams that score more TDs don’t have better offenses. My argument is that TDs are not a skill beyond general offensive skill. If two offenses play equivalently everywhere but the endzone (equally effective at running, passing long, passing short, playing consistently), the fact that team A has scored 45 TDs and team B has scored 30 TDs can be attributed entirely to luck. Touchdowns are simply high variance plays that in the long run can be completely predicted by other measures of offensive skill. Football Outsiders has shown that red zone efficiency is completely random beyond the overall offensive skill of a team (see “The Red Zone Efficiency Myth”, by Aaron Schatz). They created a statistic, “red zone advantage,” for how much better a team is in the red zone than one could expect them to be based on their offensive skill in general, and find that the correlation between “red zone advantage” in the first seven weeks and the final 10 weeks is essentially zero (this is true on defense as well). Beyond skills we see in other areas, teams have no skill in turning good drives into touchdowns. Therefore, my metric also does not consider those when evaluating teams. All that would do is bring unwanted noise into the rankings.

So now that I’ve gotten my football heresy over with, I present the 2013 NFL Team Rankings for Week 1 (next post). Skim all the luck off the top (okay, not all of it, everything is going to have some element of luck involved), and we have our first look at the skill of every team in the National Football League.