For decades, the NFL has been influenced by general managers and coaches who grew up with Football in the 1970s and 1980s. Those guys have a certain expectation of Football – ground-and-pound. They think old-fashioned and rarely want to have to do anything with modern approaches like analytics. In former times the run game has been dominant, and that has set the mind of many people. That’s why you hear sentences like “they need to establish the run game” so often. But has the run game been useful back in the 1980s? Furthermore, how relevant is it nowadays? What’s the importance of the passing game? This article will clarify this with an analytical approach.
If you take a look at the Super Bowl winners of the 80s and 90s, you will notice that many teams get to the same denominator called pass efficiency – whether it’s been an elite offensive, defensive or combined pass efficiency. MVPs like Joe Montana, Phil Simms, Richard Dent, Mark Rypien, Jerry Rice, John Elway, Steve Young – all those players had something to do with the pass.
Bill Walsh was Three Steps Ahead
The legendary coach and founder of the West Coast Offense, Bill Walsh, had then already realized that passing the ball can be much more useful than running it. His 1981 49ers team that won the Super Bowl had the fewest yards per rush in the league at 3.5. They went 13-3 and won the Super Bowl. When almost everyone still believed in the run game as the standard to succeed in the NFL, Bill Walsh and his West Coast offense were rocking the NFL, and only a few teams covered up. John Madden once said about Bill Walsh:
Bill’s legacy is going to be that he changed the offense. Offense before Bill Walsh was run, run defense, establish the run. Run on first down, run on second down, and if that doesn’t work, pass on third down. Bill Walsh passed on first down, passed on second down and used that to set up the run.
Even though the run game was dominant back in that time, it doesn’t mean it was more effective than passing – as Bill Walsh understood.
In statistics, you can calculate the correlation between two variables. For instance, the more horsepower a car has, the faster it can drive and the quicker it can accelerate. Therefore, the correlation would be high. You calculate the basic correlation R and also R-squared. R^2 is the coefficient of determination which describes how strong the connectivity is. You can use both numbers, but the ranges of significance are different. R can range from -1 to +1, depending on negative or positive correlation. R^2 ranges from 0 to 1. As a rule of thumb: measurable or “weak” correlation starts with an R^2 minimum of 0.25, significant correlation starts at 0.5. For R it’s 0.5 and 0.75, or -0.5 and -0.75.
The Run Game in the 1980s
I went through the seasons from 1980 – 1987 without the strike season of 1982 and collected total points scored, yards per
|Metric||R (Total Points)||R^2 (Total points)|
|Yards per Carry||0.14||0.02|
An R^2 of 0.55 represents a significant correlation for the passing game, and it is much more significant than the running game. As a matter of fact, the correlation of the running game is non-existent. You will notice the correlations much better visually in the following point graphs:
You notice that the correlation for yards per rush isn’t existing. The numbers spread a whole lot more than for net yards per pass attempt. Statistically, it’s irrelevant. Back in the 80s, the passing game was much more effective towards scoring than the run game. The average Net Yards Per Pass was 5.9, the average Yards Per Carry was 4.0. This results in a difference of 1.9 yards per play.
The Run Game for Super Bowl Teams in the 1980s
In the 1980s there have been ten Super Bowl matchups with – logically – 20 different teams. Here is how those teams ranked in net yards per pass and yards per run during the respective regular seasons:
|NYPPA offense||YPC offense||NYPPA defense||YPC defense|
During the 1980s, 14 Super Bowl teams ranked in the top-10 in NYPPA on offense and 16 listed in the top-10 on defense. Only two Super Bowl champions ranked in the top-10 in YPC while ranking lower in NYPPA. As a conclusion, passing mattered in the 1980s.
Bill Walsh is quoted with the following statement:
I do not believe in necessarily establishing a run or establishing a passing game. I believe more in fully dimensional football where you are establishing your offense.
That statement describes how smart Bill Walsh was. Coaches like Chuck Pagano think that they must run the ball a certain amount of times, no matter how good the opposing run defense or how good the own run efficiency is. On the other side, coaches like Bill Belichick only run when they can. In the 2014 AFC divisional round, the Patriots faced the Ravens and their stout run defense. The Pats passed the ball 51 times and gave their running backs only seven carries. They knew they couldn’t run, so they didn’t try to in the first place. That’s understanding how the game works and what efficiency means. The Pats won 35-31.
Passing efficiency in the modern NFL era
Over the past three decades, the pass has been becoming more and more efficient. A little numbers example: from 2011 to 2017, teams have gained 2.5 yards more per pass than per run. That fact alone should open some eyes. In a game that is won by gaining yards efficiently which should lead to scoring, a pass play gains 2.5 more yards than a run play on an average down.
You get a lead through the air, and you secure it on the ground. You run when you win, not win when you run – that sentence is written at the beginning of the Football Outsiders Almanac, and that is entirely true. Since 1990, 71.9% of the winning teams have more rush yards than their opponent. That’s because those teams were leading by an average of 4.5 points at the half and 6.2 points after the third quarter. They had an average of 9.8 rush attempts more than their opponents because they were milking the clock in the second half.
The interesting part comes into play when we take away the number of attempts and look at yards per play. All 7,098 winning teams since 1990 had 4.1 yards per carry whereas the losing side had 4.02. That’s a difference of 0.08 yards which is zero. But those winning teams had an average of 7.25 yards per pass whereas the losing team had 5.8 YPA – a difference of 1.45 yards! Since 2011 it is 7.5 yards per pass to 6.05 and 4.18 yards per carry to 4.19. Losing teams had 0.01 yards more per carry! Another mind-blowing fact: Since 2011, when teams have more than 6.0 yards per carry, and their opponents have 4.0 or less, they went 56-84 (40%).
Deep dive into the numbers
Over the past two decades, we were able to collect and process data more efficiently. For pass efficiency, I use metrics like Pass DVOA (Football Outsiders), Passer Rating (Teamrankings), ANY/p (Pro Football Reference) and NYPPA (Pro Football Reference), the one we used for the 1980s data. For rushing purposes, I use yards per carry and rush DVOA. Rush DVOA is down-and-distance based: DVOA measures a team’s efficiency by comparing success on every single play to a league average based on situation and opponent. For points scored I use offensive points per game (Teamrankings). The data goes from 2011 to 2017, because I wanted to use the data after the significant rule changes for the passing game in 2010. Each number applies to a team’s full season (n=225). Here are the correlation values of the passing & rushing metrics against offensive points per game:
|Metric||R (Offensive Scoring)||R^2 (Offensive Scoring))|
|Yards per Carry||0.22||0.05|
As you can see, R^2 for passing efficiency metrics ranges from 0.64 to 0.75 whereas the correlation of yards per run is non-existing. Rush DVOA has a slight relationship towards scoring, and it barely makes the measurable range. But at 0.25 it is almost insignificant. Statistically, it is just a lot more significant than mere yards per rush.
Yards per Carry has to Die
Running can be very valuable in short yardage situations, but you don’t need a superior run offense to convert a 2nd& 2. ANY/p has the best correlation to scoring because it weighs in touchdowns and interceptions. Pass DVOA, which doesn’t weigh in drive results like touchdowns and interceptions as heavily, also shows a remarkable correlation, whereas rush DVOA spreads much more. See the following point charts:
Here is ANY/p:
It gets way more apparent when looking at yards per carry, which has zero relationship with scoring. Yards per carry shouldn’t be used at all when talking about Football:
The charts are speaking for themselves. Let’s take a look at the defensive side of the ball. Here are the correlation values for the metrics above, but against defensive scoring:
|Metric||R (Defensive Scoring)||R^2 (Defensive Scoring)|
|Yards per Carry||0.09||0.01|
The scheme is the same; the correlation is not as high as on the offensive side. Run defense is mostly irrelevant, and yards per carry shouldn’t be used.
The Impact of LaDainian Tomlinson
LaDainian Tomlinson was one heck of an RB. He went 5th overall in the 2001 draft. Over the first three seasons, he averaged 1,512 rushing yards, 12.3 rushing TDs
The 2017 Cowboys had the same rush DVOA (11.9% to 11.5%) in 2017 as they had in 2016. However, they scored fewer points, and the offense as a whole was worse because the passing game wasn’t as efficient as in 2016. Wade Philips has built his defenses towards defending the pass in recent years, and he doesn’t care about the run defense. His units are good. This off-season, the Rams let their worst coverage defender in Alec Ogletree and their worst pass rusher in Connor Barwin go, extended the contracts of their two best coverage guys Lamarcus Joyner and Nickel Robey-Coleman and went after cornerbacks Marcus Peters and Aqib Talib. Suh can stop the run and rush the passer. I could go on, but you are getting my point – passing matters.
The biggest myths about the run game
So often we listen to commentators, coaches or casual fans talking about how important it is to run the ball, so a team can set up the pass and can also set up play action more efficiently. The truth is, there is no proof of that being true, just the opinion of most people. Here are the correlation numbers for rush DVOA and YPC against the ANY/p metric which weighs passing touchdowns and interceptions. Remember it has a very high correlation to scoring:
|Metric||R (ANY/p)||R^2 (ANY/p)|
The correlation between the run game and ANY/p is non-existent. There is no significant statistical correlation. Here are the charts:
The Truth about Play-Action
What about play-action? Football Outsiders Premium provides us with play action yards per play going back to 2015. Here is the correlation between the rushing metrics from above and play action yards per play from 2015-2017:
|Metric||R (PA yards per play)||R^2 (PA yards per play)|
Yes, your eyes aren’t lying: there is zero correlation between the run game and its efficiency and using play-action efficiently — just zero. At the same time, the few percents we are getting are negative. For instance, in 2017, the Detroit Lions and Washington Redskins ranked bottom-5 in yards per rush and bottom-5 in rush DVOA. On the contrary, they ranked top-5 in play-action yards per play. The Raiders, Browns, and Packers ranked bottom-3 in play-action yards per play but ranked top-15 in rush DVOA and yards per rush. Using play-action efficiently is all about situational play-calling, creativity, and execution. It doesn’t have anything to do whether you run the ball efficiently or not. As a matter of fact, the pure presence of an RB in the backfield faking a handoff is enough. To visualize it, here is the point graph for the correlation between rush DVOA & play-action yards per play:
Passing matters, running does not
The passing game is more relevant than ever. But furthermore, it is the most relevant thing in today’s NFL. Running, in general, has no impact on scoring and winning Football games, no matter how stupid that may sound. It is largely irrelevant. It doesn’t matter whether you run for 3.5 or 5.1 yards per run. But it matters whether you pass for 5.9 or 7.5 yards per pass.
It matters whether your offensive line excels in pass protection, not in run blocking. It matters whether your RB can catch the ball, create after the catch and run some routes, not whether he gets three breakout runs on the season. Recent drafts have shown that you can find a lot of these guys in the third round or later. And it matters whether your WRs create separation and make contested catches instead of blocking for runs. If you cannot pass the ball efficiently, you aren’t going to score many points – that’s a fact.
Teams should still Run the Ball
However, the run can still be beneficial, for instance in short down or goal line situations. According to Football Outsiders research, running the ball on 3rd/4th& 1-2 yards to go is more effective than passing. Moreover, it is also highly active on 2nd& short. But you don’t need to build a strong run game to get one or two yards on the ground. It’s much more about situational play-calling and knowing when to run and when not. Being unpredictable is important. Against good run defenses, you can abandon the run game and still win comfortably. But you can rarely abandon the passing game and win.
The question remains: why don’t teams pass more often? Why don’t all teams build towards the passing game? Why do Chuck Pagano and Mike Mularkey put such a high emphasis on the run game? Because as I said in the first paragraph, they probably don’t know better. Or they are too stubborn. They aren’t analytics guys. Tape guys. They have a certain expectation of Football, and that’s what they want to get on the field. But the good thing is that more and more teams are putting their focus on pass efficiency rather than on improving the run game.