Numbers never lie. You've heard it before and you'll hear it again, and on the surface, it's true. In terms of capturing some aspect of reality, stats don't lie. Math has a way of being flawless like that.
But people lie all the time, and improper interpretations of numbers can lead to horribly inaccurate conclusions. When you hear a stat like "Team X wins 85 percent of games when they run the ball 30 times," the numbers themselves are true. But that doesn't make them useful, and in fact, it can be quite disastrous to accept such propositions without careful examination.
That particular idea—that running the ball often leads to wins—has haunted NFL coaches for years. The teams that run the ball a lot really do tend to win more games, so how can coaches not be tempted to keep the ball on the ground?
The answer lies in notions of cause and effect. With any stat, we need to be careful about assigning it causal properties. Is the stat an explanation of past events, or can it be used to predict future ones? In the case of rushing attempts and winning, it's a mistake to view the former as a cause of the latter. In reality, winning is more often the cause of rushing attempts; teams with the lead run the ball late, creating "balance" that didn't previously exist. Usually, an effective passing game set up the lead.
I showed that in past analysis, and it's particularly true of the Cowboys. While they win more games when their final run-pass ratio is close to balanced, the 'Boys actually win almost twice as often when they pass the ball frequently through three quarters as compared to when they attempt to maintain balance early. That is, Dallas has traditionally won way more often when they pass early in games, then run late once they've already acquired a lead.
But doesn't the run set up the pass?
Last year, I implored the Cowboys to run more play-action passes. The offense attempted way fewer play-action passes than any team in the league, but like most squads, they were really effective when using it. Romo actually recorded a 109.1 passer rating on play-action passes, compared to just 88.3 on straight dropbacks.
Each time I mentioned using play-action, I received e-mails, comments, and tweets telling me that play-action is useless without an efficient running game. And in all likelihood, that's why the Cowboys didn't show much play-action; if you can't run, will defenders still really bite up on run fakes?
Actually, yes they will. Defenders tend to play situations as opposed to past efficiency. And if you think about it, that makes sense. Do you really believe that a linebacker cares (or even knows) that the Cowboys might be averaging only 4.0 YPC instead of the league average of 4.25 YPC? When he sees Romo show run-action on third-and-one, he's going to fly up toward the line, regardless of the Cowboys' past rushing efficiency.
That's evident in the numbers, too. Ryan Tannehill, Matt Ryan, Andrew Luck, Aaron Rodgers, and Tony Romo were all among the most effective play-action passers in the NFL last year. Every one played on a team that ranked near the bottom of the league in rushing efficiency.
As a whole, NFL quarterbacks don't find more success on play-action passes if they have efficient running games. As Scott Kacsmar points out, "The strongest correlation is between play-action passing and all passes, which again goes back to the importance of the quarterback and overall passing game. The correlation of play-action passes to non-play-action passes is the weakest of them all, which speaks to how effective this type of play is regardless of how bad your quarterback or running game is."
It's fun to think that an effective running game can set up the passing game. But it doesn't seem to be true.
Passing efficiency is what matters most in the NFL. The single most important number for predicting team wins is Adjusted Net-YPA—a stat that factors sacks and interceptions into passing totals. It's nearly four times as predictive of team success as rushing efficiency.
Actually, teams that have been highly efficient through the air have been dominant, regardless of their running game. If we look at the most pass-dependent teams since 1970 (in terms of the difference between their passing and rushing efficiency), every single one had a winning record. Every one. And of the 20 teams to be the most run-dependent, just two have posted a winning record. As Kacsmar states, "The top 20 pass-dependent teams still went a staggering 230-69-2 (.767) with six Super Bowl appearances in spite of their imbalance. Meanwhile the top 20 run-dependent teams only went 93-220-1 (.298) with the 2012 Vikings being the lone playoff appearance."
By the way, if you want to know why I'm so high on Romo as a Super Bowl-caliber quarterback, consider that his 7.03 Adjusted-Net YPA is fourth-highest in NFL history—behind Aaron Rodgers, Peyton Manning, and Tom Brady.
No Distinct Dichotomy
As a final note, I think there's a fundamental problem with how we traditionally view the run/pass balance. We split up every play into one of two categories, but that doesn't really capture the essence of football play-calling; no such distinct dichotomy exists.
In reality, plays aren't so absolute; they fall into more of a range. At one end, you might have a Shotgun 5-Wide pass, and at the other might be a traditional option run. But there are all sorts of play types in between: counters, tosses, draws, deep passes, and so on. Are we really supposed to categorize a screen pass the same as a deep post? The two play types are fundamentally different. Actually, in terms of average yards and the consistency of the play, a screen pass is probably more "run like" than it is "pass like." [embedded_ad]
Running and passing can no longer be viewed as opposites. Instead of labeling a particular play as a run or pass, we might want to start thinking of it in a more pluralistic manner. A play-action look is less of a "pure pass" than a straight dropback, for example, but perhaps more pure than a quick screen.
Football is unique game that's holistic in nature; unlike baseball, we can't compartmentalize plays into rigid groups (such as ball or strike, hit or out). Once we start viewing play-calls as exhibiting a range as opposed to a dichotomy, it might change the way we view 'balance.' Instead of striving for a 50/50 run/pass ratio, true balance could be sprinkling runs in with various pass types: screens, deep looks, play-action passes, and so on.
Only then will we be able to move past the erroneous notion that rushing the ball often creates a truly balanced offense.