Will Passing Still Be King in 2024?
The premium on airing it out has been reduced in recent years. Will this NFL season add to that trend?
One aspect of football strategy that I’ve long been fascinated by is the complex playcalling dynamic between running the ball and passing it. What is the right mix of each play type?
Traditionally, coaches often tended to aim for balance — if not an exact 50-50 split — presumably to make the offense less predictable, while also taking advantage of whatever play the defense was less ready for. Running was often used to set up the pass, lulling the defense to sleep with three yards and a cloud of dust before heaving it over the top for a big gain.
Empirically, though, it’s also true that passes gain more yards on average than runs. (A lot more, in fact.)1 As rule changes made passing easier, the NFL started taking advantage by throwing the ball more and more — but from a game-theoretical perspective, as long as a gap exists in efficiency between the pass and the run, you should pass the ball even more. For years, this made for a puzzling disconnect in which teams weren’t testing the limits of the passing game enough to make runs and passes equally worthwhile.
That was, until recently. Over the past few seasons, the premium between passing and running has shrunk considerably, if not outright disappeared, depending on how you measure it. This suggests teams have finally found an optimal mix under which running is no longer massively overvalued — though it is unclear whether that will remain true going forward.
At a glance, it may not seem like anything has really changed in the interplay between passing and rushing. The average pass in 2023 yielded 6.0 net yards, while the average run gained 4.2 — good for a gap of 1.8 yards per play, the same as it was in 2006 (or 1992, or 1979):
But as Chase Stuart pointed out in a seminal 2008 PFR Blog post entitled “Why do teams run the ball?”, simply looking at raw averages misses the potential for passing plays to end in events like interceptions or sacks. (Both play types can also end in fumbles, but Chase didn’t have that data at the time; as it turns out, they have a pretty equal rate of happening on each play type anyway.)2 So Chase used adjusted yards per play in his analysis,3 and he found that running the ball was actually more efficient than passing until 1980 — at which point a gap began developing, though a smaller one than suggested by pure yards per play.
Now, of course, we have even better tools to analyze the run-versus-pass dynamic, which I want to dig into some here.
First, let’s look at Expected Points Added (EPA) — which measures how much a team’s chances of scoring changed on any given play — for each play type, according to nflfastR’s model. We’ll make one further adjustment for our purposes here, setting the average play from scrimmage to zero net EPA, so we can see how both runs and passes compare to the overall average offensive snap. Here’s how that adjusted EPA per play has evolved since the stat was first calculated in 1999:
Now that is a remarkable change! As recently as 2015, the per-play gap in EPA between passing and rushing was as wide as it had ever been this century, but in 2022 the two play types had essentially achieved parity — runs were as efficient as passes and vice-versa. (Granted, in an environment where 57 percent of all plays were passes.) Passing took a slight lead again in 2023, though the gap was as narrow as it had ever been before 2022, tying the leaguewide differential from 2003.
I’m a fan of breaking down football stats by situation when possible, so I wanted to see whether this trend also held true specifically in “neutral” circumstances: Early downs (1-2), reasonable yardage to go (10 or fewer), between the 20-yard-lines, with the margin within a TD, in quarters 1-3.4 And as it turns out, passing still carries a huge premium in those types of situations:
But rushing is an increasingly useful situational tool outside of those neutral conditions. Interestingly, this is something coaches always seem to have intuited, with one answer to the “why run the ball?” question typically being that it’s a low-risk way to burn clock when the goal is to maximize winning rather than points or yardage. And in fact, we can see this in action when we look at Win Probability Added (WPA) instead of EPA, to account for how each play type contributes to winning the game.
By this accounting, which is probably the most holistic way of looking at the pass-run efficiency gap (because it naturally accounts for the situational value of rushing), running the ball was actually a more efficient path to victory than passing in 2022, and passing only re-took the narrowest of leads in 2023. That’s a very different story than you might get from simply looking at yards per play, or from the narrative that running the ball (or stopping the run) are largely irrelevant in today’s NFL.
That doesn’t mean there aren’t still imbalanced teams with big gaps between how effective their passing and running games are. Here, for instance, were the WPA passing premiums for every team a season ago, which really illustrates how much variation there still is among teams in that regard:
Leaguewide, though, we’re in an age of parity between passing and rushing efficiency that we haven’t really seen before in the play-by-play stats era — and likely not since the late 1970s, if we’d had that data back then. Of course, this all comes with the caveat that it’s a recent development (things were tilted toward passing in a relatively normal way within the past 5 years) and that passing already made a slight comeback in 2023 compared with 2022.
So the past few seasons may well end up being a blip on the league’s long-term trend toward aerial supremacy. But that’s what makes this such an interesting statistical storyline for 2024 — we’re truly in a place where the league could go in either direction from here this season.
Filed under: NFL
This truth was frequently pointed out by early “sabermetric” football resources like The Hidden Game of Football, Football Outsiders and the Pro-Football-Reference Blog.
Since 1991, the first year of SportRadar’s fumble data, a dropback results in a fumble 1.85% of the time and a rush attempt leads to a fumble 1.92% of the time.
Adjusted yards gives a 20-yard bonus for touchdowns and a 45-yard penalty for interceptions. It also subtracts sack yards — which, to be fair, are already removed at the team level — and divides by dropbacks instead of regular pass attempts.
These are plays under which the conditions of the game aren’t going to distort a team’s strategic choices.
The King is alive and well - long live the King! An important and terrific discussion to start the season. There's a lot here that I could be misinterpreting, so apologies in advance if I am misunderstanding some elements. With that in mind, let me offer the following reactions.
How you "run" the football matters immensely to the numbers. To that end, I am concerned that the definitions here are possibly too broad and bit confusing on the issue. What constitutes a "run" and what a "pass?" For decades that was easy to discern. Outside of direct snaps, running was when the quarterback turned around and immediately gave the ball five yards behind the line of scrimmage to a smallish running back in full view of the defense who ran into a wall of bodies in front of him hoping to gain yardage. It was the least deceptive and efficient strategy in sports - and arguably the dumbest. Not surprisingly, its YPA was very low and very poor. This is what people think of when they hear "running the football."
Today, it isn't so easy. Runs by quarterbacks and not traditional RBs - either as RPOs, read options, designed runs or scrambles - constitute a significant and increasing portions of offenses that count as runs too. Yet those feel different to fans, and for good reason. They are very different and far more deceptive and efficient than the ineffective traditional runs that immediately come to mind. Put another way, Lamar Jackson is way more efficient running the ball than John Riggins ever was.
Same with passing. Passes behind the line of scrimmage count as "passes" - but really aren't, and their extremely low YPA misleadingly gets labeled as "passing." Because these plays are initiated so quickly, without any reads or consideration of other plays, they are extended handoffs designed to avoid the hazards of passing you reference - sacks and interceptions. While the ball is technically in the air, it's not really a "forward" pass as the definitions used might suggest.
Incredibly, offensive coordinators continue to call these awful plays week after week despite never working or gaining yards. A testament to the level of delusion and analytical ignorance still roaming the sidelines in the NFL. One way we can all advance is to demand that the NFL stop calling these dreadful plays that are unfairly maligning the reputation of the forward pass.
So, our definitions and how we bucket the data could very likely be misleading people into thinking that somehow - miraculously - the data proves that Jonathan Taylor is every bit as valuable as Joe Burrow. He's not. More data segmentation would likely present a more accurate picture. Take out the ever-increasing quarterback runs from the "rushing" category (I suspect the likely cause of the recent shrinkage in the delta between the strategies), take out passes behind the line from the "passing" category and add shovel passes to the running category and reshuffle. I bet the charts look a lot different.
The King is still the King.
Second, the focus on turnover risk has always falsely slandered the forward pass for decades. You would routinely hear people say "when you pass, three things can happen and two of them are bad." Yet, definitions are a huge part of the problem here too. By any reasonable interpretation, a punt is a turnover. A team surrenders possession to the other team on its own downs - i.e. a turnover.
We hear that an interception far downfield is "as good as a punt" and a failed 4th down is a "turnover on downs" - yet, we seem inexplicably unable to equate punts landing into the arms of the opponent as turnovers as well. In fact, for decades the punt was somehow heralded as a "smart" strategy because it exchanged yardage and points for turnover ratio and field possession...but I digress.
Guess what strategy I suspect yield more punts - more total turnovers - overall? The traditional running of the football. When you ignore punts, running falsely looks less risky than passing. It isn't. Giving away possessions to your opponents hurts you, regardless of how its done.
Finally, one way I look at the power of an offense is by examining the percentage of first downs an offense achieves without using 3rd or 4th downs or penalties. In other words, which offenses get more first downs utilizing only two plays? While an examination of the total number of first downs generated is certainly valuable as a broad offensive measurement, for me incremental insight can be gained by examining the first down percentage gained on the first two downs. Stripping out first downs achieved on 3rd and 4th downs as well as those gained by a penalty will tell you how powerful an offense truly is at quickly moving the sticks and gaining first downs more efficiently. Invariably, passing teams are more powerful.
Teams that consistently need to string together plays by going deeper into their allotted downs to achieve incremental first downs are likely running teams - and accordingly, less powerful, less likely to score and more likely to punt than those that can get to the sticks more efficiently. This has been particularly true when teams face stiffer defenses such as in the Playoffs.
May you and King Forward Pass have long and fruitful reigns!
Just eyeballing the matrix, it looks like good teams have better WPA/Pass and bad teams have better WPA/Run.