- Joined
- Apr 7, 2013
- Messages
- 120,131
[h=1]Ted Nguyen 42m ago[/h]
Have you ever watched in frustration as your favorite team inexplicably runs the ball up the middle with little or no success? Do you find yourself thinking, What are they doing? Why don’t they throw the ball more? Am I crazy?
You’re not crazy. The analytics backs up your belief that this type of play calling is illogical.
In the wild-card round last season, for example, the Seattle Seahawks’ offense sputtered as they were eliminated by the Dallas Cowboys 24-22. In reviewing the film, what was most frustrating to watch was Seahawks offensive coordinator Brian Schottenheimer’s blind commitment to running the ball when his team was clearly outmatched up front. The Seahawks’ running backs averaged 2.8 yards per carry while their MVP-caliber quarterback, Russell Wilson, averaged 8.6 yards per pass in the same game.
After the game, coaches didn’t express any regret in sticking with the ground game. Some coaches just can’t help themselves — they grasp on to their ingrained belief that “establishing the run” equates to victory, even though analytics has proved otherwise.
Traditionalists will tell you that the run game is vital to pounding an opponent into submission and that running the ball will lead to wins. But there’s no disputing that the running game matters less in today’s game. The game rules have been changed to make passing easier and more efficient and the CBA rules limit contact in practice, which makes it a lot more difficult for offenses to refine a physical run game.
A yearly downward trend in rushing attempts continued in 2018. Teams only ran the ball 25.9 times per game, which is the lowest average in league history.
The importance of the run game has undoubtedly been diminished, but to what extent? Will the trend continue? What is the run game’s role in today’s NFL? How does it affect strategy and roster construction? How should teams value running backs?
To get the answers to these questions, I collected the research of several top football analytics experts to see what the numbers say and talked to several former and current NFL players and coaches to get their side of the argument. [HR][/HR][h=3]What does the analytics say about the value of the run game?[/h]
Passing is more efficient than running and leads to more points
The most obvious evidence that passing is more efficient than running is simply that pass attempts average more yards than run attempts. Passing attempts have increased for many reasons, including rules changes that limit contact from defensive backs and protect receivers from vicious hits.
Josh Hermsmeyer writes for FiveThirtyEight and has consulted with NFL teams. He did a study on the scoring trend in his book “The League” and found that passing attempts have increased year over year in nearly every season from 2009 to 2016. During that time scoring has increased by 9.2 percent. The increase in passing isn’t the only reason that scoring has gone up, but it’s played a large role.
It’s commonly accepted that a pass play is riskier than a run play because of the chance of a sack, fumble or interception. While it’s inherently true that there is a greater range of possibilities that could lead to negative plays when passing the ball, teams have actually become more efficient as they’ve passed more.
“Touchdown percentage (TD%) has increased slightly while Interception percentage (INT%) has decreased,” wrote Hermsmeyer. “The net effect is that there has been an increase in TD/INT Ratio from 2009 to 2016 of 40%. Passers in the NFL now throw almost twice as many TDs as INTs on average.” Completion percentage has also increased every year in the same sample.
Hermsmeyer’s study concluded that:
Running does not correlate with winning
In their own separate studies, Warren Sharp, who runs his own website and has also been a consultant for NFL teams, and Timo Riske, who is a doctoral student from Mainz, Germany, both found that establishing the run did not correlate with winning. Using data from 2011 to 2014, Sharp ran regressionson yards per rush attempt and wins, total rushing yards and wins, and yards per pass attempt and wins. He was not surprised to find that only yards per pass attempt correlated with winning.
“(Over the) last three years we find no connection, relationship or correlation between teams who run the ball well and teams who win the most games,” concluded Sharp. “Running is used to complement passing, but passing is what determines wins and losses in today’s NFL.”
The Athletic’s Ben Baldwin, who has written for Football Outsiders, talked about Riske’s work in his recent article about the dark web of NFL analytics. Riske found that the best predictor for how good a team will be is how effective its passing game was in the past. He concluded like most analytics experts that:
1. In a given game, passing efficiency predicts the winner to a much greater extent than rushing efficiency.
2. A strong passing offense is much more likely to persist over time than other units.
The analytics suggests that a team’s best chance at sustained success comes by investing in the passing game. This is why a very good to elite quarterback could set a franchise up for success for a long time, but investing in the passing game also means surrounding a quarterback with weapons and protecting him.
The analytics says you don’t have to establish the run for effective play action — though players and coaches still believe it’s crucial
The relationship between play action effectiveness and the run game is a subject that should be further researched in more detail. A play-action pass play’s objective is to get defenders out of position with a run fake. Which means an offense has to run the ball effectively to set up the play action, right? Well, actually … the answer is not as simple as you might think. Research from both Hermsmeyer and Baldwin has shown that not only is play action much more effective than regular drop-back passing, it is also effective regardless of rushing attempts or rushing success.
Baldwin took the difference between each team’s yards per pass on play action and their yards per pass on regular drop backs and related it with total rushes, rushing attempts and rushing success rate. He found that each rushing category did not impact the effectiveness of play action.
The chart shows that a higher number of total rushes, percentage of runs to pass and rushing success rate did not lead to an increase in play-action passing effectiveness compared with a team’s drop-back passing effectiveness.
In the past, we have generally accepted that running the ball more frequently or with more success will cause linebackers to bite harder on play fakes, but player tracking data shows that might not be true. Hermsmeyer did a study using player data to track the distance that linebackers wasted moving up toward the line of scrimmage reacting to play fakes before retreating back to coverage on play-action plays. The average distance remained consistent until the 12th time a play-action play was called in a game. After the 12th time play action was called, the average went up and down but this fluctuation might be a result of a small sample size (play callers haven’t called more than 12 play actions very often).
Hermsmeyer theorized that play action’s sustained effectiveness in deceiving linebackers could be due to the fact stopping the run has been so ingrained in them from youth to the pros that it’s almost second nature. Additionally, during a play, linebackers are focused on reading and reacting to their keys and might not be conscious of how many times an offense has run play action throughout a game.
Former NFL linebacker Kirk Morrison believes there is truth to the fact linebackers almost always have to respect the run fake in certain situations. “You don’t want to get to your gap late,” he explained. For example, linebackers are more likely to bite on first and long than third and long.
But Morrison, as well as other coaches and players, are also adamant that a team that is running the ball well does indeed force linebackers to fill their gaps quicker.
There is validity to Baldwin’s conclusion that play action is still effective without establishing the run, as well as Hermsmeyer’s conclusion that play action doesn’t lose its effectiveness after repeated use. An NFL coach even told me teams aren’t using enough play action. But it would be a mistake to ignore the testimonials and experiences of players and coaches who have seen firsthand how running success can get defenses to bite harder on run fakes.
“If a team is ripping off 5-6 yard gains, you are definitely trying to jump up and get there,” said Morrison. “If you respect the offensive line, you don’t want to get there late. A split second could get you beat in terms of an offensive lineman swallowing up a linebacker.”
There could be other factors at play that haven’t been tested or perhaps can’t be measured. For example, the distance wasted by linebackers on play fakes could be affected by the type of coverage that they are in. If a linebacker has a running back in man-to-man coverage, he might continue to move toward the running back to cover him. Also, the data could be skewed by teams that didn’t use play action enough, resulting in an unreliably small sample size.
The analytics suggests running in short yardage situations and the red zone
Sharp found that running the ball on third- or fourth and short had a significantly higher success rate than passing the ball. League-wide, in 2018, on third- and fourth down with 1 or 2 yards to go, rushes had a 13 percent higher success rate (70 percent) than passes (52 percent). Red zone rushes had an 8 percent higher success rate (50 percent) than red zone passes (42 percent). To be successful in these situations, teams have to be able to run the ball. This leads to the next question: whether teams could do so effectively if they don’t run the ball often. [h=3]Do running backs still matter?[/h]
If you want to find out what people value, follow the money. On average, running backs make less than every position group in football except for long snappers. Todd Gurley has the most guaranteed money on his contract among running backs by a wide margin … and yet he only ranks 36th overall in the NFL in that category. The next running back on the list is David Johnson, who ranks 73rd.
Analytics shows that running backs, even star running backs, have very little control over the outcome of the play over time. Hermsmeyer found that running game success is determined more by the box count, which could be manipulated by an offense’s use of personnel, and offensive line performance, which was measured by his version of adjusted line yards.
PFF’s WAR (Wins Above Replacement) score is a way of assessing player value by essentially answering the question: “if you replaced this player with a player off the practice squad, how many fewer wins would his team have over the course of the season?”
Running backs did not fare well in WAR relative to other positions. Christian McCaffrey was the highest-ranked RB in WAR last season but he was only ranked 36th in the NFL overall.
The WAR score shouldn’t be taken literally but it does provide a measurement of value. The running back position is the most easily replaceable, according to this metric, and many analytics experts have made the same conclusions in their own research.
The 2018 season provided plenty of evidence for the replaceability of running backs. Le’Veon Bell, who is widely regarded as a top-3 running back, sat out the season because of a contract dispute. He was replaced by former third-round pick James Conner, and the Steelers’ rush offense didn’t miss a beat — their yards per carry and rushing success rate actually improved slightly without Bell.
Kareem Hunt, who was the league’s rushing leader in 2017, was cut from the Chiefs right before their playoff run in 2018 after video was released of his physical altercation with a woman in a hotel hallway. The Chiefs’ rush offense dropped slightly but their rush success rate jumped up to 60 percent.
That’s not to say that you can plug any back into these offenses and expect the same level of success, but the numbers suggest that it’s much easier to find a replacement for even superstar running backs. So because running backs are so dependent on outside factors for production, how replaceable they are, and their short shelf lives, teams shouldn’t invest heavily in the running back position. The analytics suggests that teams could better use their cap space investing in assets that directly contribute to the passing game.
Sharp looked at the average cap hit per running back and compared that with team wins and found the relationship is negatively correlated. “What this means is that the more money a team pays to its running backs, the fewer wins they gain,” wrote Sharp. “Look at 2013 as an example: none of the top 5 teams in terms of cap hit per RB (Tennessee, Minnesota, Chicago, Oakland and Baltimore) finished the season with a winning record.”
The Cardinals awarded running back David Johnson a three-year, $30 million extension on the strength of his All-Pro 2016 season, and then he only played in one game in 2017 because of injury. In 2018, the Cardinals’ offensive line was ravaged by injuries and they fired offensive coordinator Mike McCoy during the season. Johnson’s circumstances were terrible and he only averaged 3.6 yards per carry. As skilled and talented as Johnson is, his positional impact wasn’t strong enough to elevate the rest of the Cardinals’ offense.
The Giants drafted Saquon Barkley second overall in 2018 and Barkley had a spectacular season in which he broke a ridiculous 94 tackles — the next-best running back was McCaffrey with 63 broken tackles. Barkley was so good that he accomplished the extremely difficult task of elevating his offensive line.
His line finished 29th in Football Outsiders’ adjusted yard line stat, which is a metric that attempts to separate the yards that an offensive line is responsible for creating in the run game. Yet the Giants finished 10th in the league in yards per carry (4.7 ypc). Even with Barkley’s remarkable season, the Giants only finished 16th in points per game (23.1 ppg) and they only won five games. His playmaking ability elevated the Giants’ run game unit, but even with a top 10 rushing attack, the Giants were still a mediocre offensive unit.
For teams that are rebuilding, it just doesn’t make sense to draft a running back early. Those teams usually don’t have the offensive line to maximize a running back’s impact and they likely won’t play with a lead often enough to run the ball with frequency.
However, though the analytics community might advise against it, I don’t think it’s a bad move for a contending team with a strong passing game or elite defense to draft a running back early. Running backs have shown they can make an easier transition from college to the pros relative to other positions, and when they’re on a rookie contract, they don’t cost much.
With all that said, here are the recommendations from the analytics community:
1. Don’t invest heavily in your run game because it doesn’t correlate with winning
2. Use more play action because it’s more effective than drop-back passing and it has proved to work without having to “establish the run”
3. Don’t run the ball into eight-man boxes
4. Test the limits of passing, don’t force the run “unnecessarily”
5. Use the pass to set up the run (run when defenses start to adjust to pass)
6. Deception in all its forms is the most important element in offense
7. The run game is valuable in short-yardage situations, in the red zone and for running out the clock.
But what about the other side of the argument? [h=3]The case for the run game from NFL coaches and players[/h]
Though it might be hard to define and quantify, the run game is still important, even if the analytics we’ve reviewed challenges traditional views on its value. Though some coaches are simply rigid and unwilling to change their long-held beliefs, some of the most innovative and thoughtful offensive coaches in the league, such as the Rams’ Sean McVay, also build around the run game.
In this section, I’m not attempting to make a case to run the ball more or less; I’m simply trying to clearly define the value of the run game. I talked to several current and ex-coaches and players about the subject and asked them to respond to the arguments that the analytics presents. Some coaches/players asked not to be named, but most of their answers are consistent with each other’s.
Less risk of a negative play
Passing is becoming more efficient but the inherent risk of passing the ball still has to be considered. When passing the ball, there is a greater chance of a negative play than with a run play. A negative play could be a turnover or sack. On a pass play, there is the chance of an interception or fumble. The quarterback could get stripped or an offensive player could fumble the ball after a catch. On a run play, the only significant risk of a turnover is a fumble by the running back.
In 2018, a turnover occurred 2.9 percent of the time during plays in which the quarterback dropped back for a pass (includes pass attempts, sacks, scrambles). On plays in which the ball was handed off or kept by the quarterback by design (example: zone read), a turnover occurred 0.6 percent of the time. Turnovers occurred almost five times more on pass plays than run plays.
Additionally, sacks (including ones that don’t end in turnovers) are often drive killers. Derrik Klassen of Football Outsiders charted the 2016 season and found that only 179/1118 (16.01 percent) drives in which there was a sack eventually got another set of downs. 83.99 percent of drives were essentially killed by sacks. Klassen found that defenses were 70 percent more likely to kill a drive after sacking the quarterback than they were likely to surrender another set of downs.
Running the ball allows offenses to minimize risk. This doesn’t mean teams should continually try to establish the run if they are outmatched up front, but with a higher volume of passes comes a higher chance of turning the ball over, and winning the turnover battle does correlate with winning.
Disarm the pass rush
An offensive line coach told me he might get kicked out of the fraternity for telling me that teams don’t need to run the ball to win, but they need to run the ball to pass. Obviously, not every coach or player would agree with that statement, but every coach and player that I talked to believes that running the ball helps pass protection because it forces pass-rushers to play the run and maintain their gap responsibilities rather than just recklessly fly upfield to rush the passer.
“The biggest mismatch on the field is defensive linemen against offensive linemen,” said 49ers run game coordinator Mike McDaniel. “Defensive linemen are supreme athletes and it’s a hard job to block people. Anything that gets defensive linemen to think about more than sacking the quarterback is a win.”
“The reason why you want to establish the run game is so you can unguard the defense so you can pass block,” said former Bengals and Cowboys offensive line coach Paul Alexander. “You can’t let them think pass, pass, pass. You have to keep them guessing.”
The very nature of pass blocking puts offensive linemen at a disadvantage. They are moving backward and reacting to pass-rushers. By running the ball, you allow offensive linemen to attack and it gives them a psychological edge. Run attempts, even if they aren’t resulting in big gains, can slow down pash-rushers and give offensive linemen a break from pass blocking.
Flawed statistics, unquantifiable effects and game theory
A coach from an analytics-driven team told me that some of the statistics used to prove the passing game’s superiority over the run game could be flawed or don’t tell the whole story because they rely on averages that could be swayed by big plays. He believes that it is better to look at median statistics when looking at the run game. The idea here is that 1) though the run game produces fewer big plays, it produces more consistent yardage; and 2) passing plays have a high variance (explosive plays, incompletes, turnovers, negative plays) that isn’t reflected by averages. There has been some discussion of using median and mode statistics to look at the run game but there haven’t been any thorough studies yet that I could find.
Count McDaniel as skeptical of the analytics: “A lot of stats are not quantifiable in terms of the pass game in regards to the run.” McDaniel and several other coaches explained that a strong run game affects the type of personnel on the field and the structure of the defense. For example, to stop the run, defenses have to keep run-stuffing defensive linemen in the game who aren’t as skilled at rushing the passer, or they have to keep bigger linebackers in the game who aren’t as skilled at coverage.
“A lot of defenses could take away explosive pass games if they were able to play two-high, two-man all day. The reason they can’t: You have run fits that are a hat short in the run game,” said McDaniel. “You have to earn single safety in this league. Single safety opens up some passing holes with five eligibles and three deep. You have to earn that, and to earn that, you have to draw another guy into the box because they’re afraid you’re gonna run it.”
McDaniel was an offensive assistant with the Atlanta Falcons in 2016 and he brought up their game in Week 4 against the Panthers during that season as an example of how the run game affected the defense in ways that are difficult to measure.
In the game before, the Falcons rushed for 217 yards against the New Orleans Saints. The Panthers came into the game knowing they needed to stop the Falcons’ rushing attack. The Panthers had a talented, aggressive defensive line and All-Pro linebackers Luke Kuechly and Thomas Davis attacked the line of scrimmage relentlessly from the very start of the game.
On the first offensive snap for the Falcons, the defense started out in a two-deep coverage, but a safety rotated into the box late to become the eighth man in the box. Quarterback Matt Ryan didn’t fake a handoff but the linebackers moved forward in anticipation of stopping the run. The safety’s rotation left receiver Julio Jones in a one-on-one situation and he took advantage with a nice gain to start the game off.
The Falcons’ success on the ground against the Saints in Week 3 affected the Panthers’ defense in Week 4 and helped the Falcons’ drop-back pass game even without play action. This effect could be difficult to measure by analytics. Against the Panthers, the Falcons only handed the ball off to their running backs 21 times for an average of 3.6 yards. They didn’t have a successful day on the ground because the defense game planned to take it away from the start of the game. But in doing so, they left their secondary vulnerable and Ryan threw for 503 yards while Jones had 300 yards receiving.
If you just looked at the numbers you would think the Falcons won this game solely on the strength of their pass game, but in reality, the fear of the Falcons’ run game left the Panthers’ secondary vulnerable.
McDaniel also said that in that game, they knew they were overmatched up front and called runs just to keep the defense honest. Though they were an outside zone team, they called “safer runs” (inside zone, duo) that are quicker hitting and carried less risk of losing yardage. The objective of rushing in this game wasn’t to necessarily gain yards but to maintain the defense’s respect of the run game and to aid the offensive line.
This pattern of play calling is an excellent example of how game theory could be used to explain why offensive coordinators call runs despite the fact passes average more yards. The goal of every play call isn’t to gain the most yards possible.
“It’s not about just maximizing expected value for any individual play, it’s setting up aggregate expected value/return,” said Chris Brown, the editor of Smart Football. “So a run play might be micro inefficient but macro efficient because of what else it sets up/protects.”
Additionally, offenses only go into a game with a certain number of plays in the game plan. Coordinators can safely call the same run concept over and over again and still safely gain yards, but if they call the same pass concept over and over again and defenders anticipate the concept coming, there is a risk that they’ll jump the pass and intercept it.
On every pass attempt, quarterbacks are computing massive amounts of information in a short period of time while they are getting harassed by pass-rushers. Even with elite QBs, it can be beneficial to give them mental breaks and allow them to hand the ball off once in a while.
Finite resources
Finding personnel for a pass-heavy attack is difficult. The quarterback position might be the most difficult position to fill in all of sports. There just aren’t enough starting-caliber quarterbacks for all 32 teams. It’s also hard to find reliable pass blocking tackles. This is why quarterbacks are the highest-paid players in the league while tackles are among the highest-paid position groups in the league.
It is easier to find good running backs and reliable run-blocking offensive linemen. Plus, most quarterbacks and offensive linemen need the run game to protect them from themselves. Even if teams wanted to, how many teams actually have the resources to pass the ball at a high volume? [HR][/HR][h=3]Final thoughts[/h]
After looking at all the research and talking with experts in the field, it seems clear the hard-line philosophy of the importance of establishing the run is outdated and disproven. But there is still value in being able to run the ball. It’s a tricky balance, but in building a team, general managers and coaches have to put more emphasis on investing in a strong passing game, but at the same time, they can’t lose sight of creating an efficient run game.
In 2018, of the top 10 teams in offensive DVOA (Football Outsiders’ measurement for a team’s efficiency by comparing success on every single play with a league average based on situation and opponent), seven were in the top 10 in rushing success rate. Eight of 10 had winning records and made the playoffs. So while there isn’t a correlation with traditional rushing success stats, there are correlations with running, winning and efficient offensive production.
This research is valuable for starting important conversations, but some analytics experts have reached hard conclusions that are too black and white for an extremely complex game. There is still value in the run game that hasn’t been properly explained by numbers. The subject should be researched further, but it seems experts on both sides of the argument are in a constant shouting match with each other in which nobody ever wins. Analytics experts, coaches, players, analysts and even fans could all benefit from listening to each other and learning from each other in this conversation to advance the game.
Have you ever watched in frustration as your favorite team inexplicably runs the ball up the middle with little or no success? Do you find yourself thinking, What are they doing? Why don’t they throw the ball more? Am I crazy?
You’re not crazy. The analytics backs up your belief that this type of play calling is illogical.
In the wild-card round last season, for example, the Seattle Seahawks’ offense sputtered as they were eliminated by the Dallas Cowboys 24-22. In reviewing the film, what was most frustrating to watch was Seahawks offensive coordinator Brian Schottenheimer’s blind commitment to running the ball when his team was clearly outmatched up front. The Seahawks’ running backs averaged 2.8 yards per carry while their MVP-caliber quarterback, Russell Wilson, averaged 8.6 yards per pass in the same game.
After the game, coaches didn’t express any regret in sticking with the ground game. Some coaches just can’t help themselves — they grasp on to their ingrained belief that “establishing the run” equates to victory, even though analytics has proved otherwise.
Traditionalists will tell you that the run game is vital to pounding an opponent into submission and that running the ball will lead to wins. But there’s no disputing that the running game matters less in today’s game. The game rules have been changed to make passing easier and more efficient and the CBA rules limit contact in practice, which makes it a lot more difficult for offenses to refine a physical run game.
A yearly downward trend in rushing attempts continued in 2018. Teams only ran the ball 25.9 times per game, which is the lowest average in league history.
The importance of the run game has undoubtedly been diminished, but to what extent? Will the trend continue? What is the run game’s role in today’s NFL? How does it affect strategy and roster construction? How should teams value running backs?
To get the answers to these questions, I collected the research of several top football analytics experts to see what the numbers say and talked to several former and current NFL players and coaches to get their side of the argument. [HR][/HR][h=3]What does the analytics say about the value of the run game?[/h]
Passing is more efficient than running and leads to more points
The most obvious evidence that passing is more efficient than running is simply that pass attempts average more yards than run attempts. Passing attempts have increased for many reasons, including rules changes that limit contact from defensive backs and protect receivers from vicious hits.
Josh Hermsmeyer writes for FiveThirtyEight and has consulted with NFL teams. He did a study on the scoring trend in his book “The League” and found that passing attempts have increased year over year in nearly every season from 2009 to 2016. During that time scoring has increased by 9.2 percent. The increase in passing isn’t the only reason that scoring has gone up, but it’s played a large role.
It’s commonly accepted that a pass play is riskier than a run play because of the chance of a sack, fumble or interception. While it’s inherently true that there is a greater range of possibilities that could lead to negative plays when passing the ball, teams have actually become more efficient as they’ve passed more.
“Touchdown percentage (TD%) has increased slightly while Interception percentage (INT%) has decreased,” wrote Hermsmeyer. “The net effect is that there has been an increase in TD/INT Ratio from 2009 to 2016 of 40%. Passers in the NFL now throw almost twice as many TDs as INTs on average.” Completion percentage has also increased every year in the same sample.
Hermsmeyer’s study concluded that:
- Offensive scoring in the NFL has increased over time.
- The rise of passing is responsible for the increase.
- Passing the ball is far more efficient than running.
- As the league has begun passing more it has not seen a decrease in passing efficiency. Instead, teams have become more efficient at gaining yards and scoring points through the air.
Running does not correlate with winning
In their own separate studies, Warren Sharp, who runs his own website and has also been a consultant for NFL teams, and Timo Riske, who is a doctoral student from Mainz, Germany, both found that establishing the run did not correlate with winning. Using data from 2011 to 2014, Sharp ran regressionson yards per rush attempt and wins, total rushing yards and wins, and yards per pass attempt and wins. He was not surprised to find that only yards per pass attempt correlated with winning.
“(Over the) last three years we find no connection, relationship or correlation between teams who run the ball well and teams who win the most games,” concluded Sharp. “Running is used to complement passing, but passing is what determines wins and losses in today’s NFL.”
The Athletic’s Ben Baldwin, who has written for Football Outsiders, talked about Riske’s work in his recent article about the dark web of NFL analytics. Riske found that the best predictor for how good a team will be is how effective its passing game was in the past. He concluded like most analytics experts that:
1. In a given game, passing efficiency predicts the winner to a much greater extent than rushing efficiency.
2. A strong passing offense is much more likely to persist over time than other units.
The analytics suggests that a team’s best chance at sustained success comes by investing in the passing game. This is why a very good to elite quarterback could set a franchise up for success for a long time, but investing in the passing game also means surrounding a quarterback with weapons and protecting him.
The analytics says you don’t have to establish the run for effective play action — though players and coaches still believe it’s crucial
The relationship between play action effectiveness and the run game is a subject that should be further researched in more detail. A play-action pass play’s objective is to get defenders out of position with a run fake. Which means an offense has to run the ball effectively to set up the play action, right? Well, actually … the answer is not as simple as you might think. Research from both Hermsmeyer and Baldwin has shown that not only is play action much more effective than regular drop-back passing, it is also effective regardless of rushing attempts or rushing success.
Baldwin took the difference between each team’s yards per pass on play action and their yards per pass on regular drop backs and related it with total rushes, rushing attempts and rushing success rate. He found that each rushing category did not impact the effectiveness of play action.
The chart shows that a higher number of total rushes, percentage of runs to pass and rushing success rate did not lead to an increase in play-action passing effectiveness compared with a team’s drop-back passing effectiveness.
In the past, we have generally accepted that running the ball more frequently or with more success will cause linebackers to bite harder on play fakes, but player tracking data shows that might not be true. Hermsmeyer did a study using player data to track the distance that linebackers wasted moving up toward the line of scrimmage reacting to play fakes before retreating back to coverage on play-action plays. The average distance remained consistent until the 12th time a play-action play was called in a game. After the 12th time play action was called, the average went up and down but this fluctuation might be a result of a small sample size (play callers haven’t called more than 12 play actions very often).
Hermsmeyer theorized that play action’s sustained effectiveness in deceiving linebackers could be due to the fact stopping the run has been so ingrained in them from youth to the pros that it’s almost second nature. Additionally, during a play, linebackers are focused on reading and reacting to their keys and might not be conscious of how many times an offense has run play action throughout a game.
Former NFL linebacker Kirk Morrison believes there is truth to the fact linebackers almost always have to respect the run fake in certain situations. “You don’t want to get to your gap late,” he explained. For example, linebackers are more likely to bite on first and long than third and long.
But Morrison, as well as other coaches and players, are also adamant that a team that is running the ball well does indeed force linebackers to fill their gaps quicker.
There is validity to Baldwin’s conclusion that play action is still effective without establishing the run, as well as Hermsmeyer’s conclusion that play action doesn’t lose its effectiveness after repeated use. An NFL coach even told me teams aren’t using enough play action. But it would be a mistake to ignore the testimonials and experiences of players and coaches who have seen firsthand how running success can get defenses to bite harder on run fakes.
“If a team is ripping off 5-6 yard gains, you are definitely trying to jump up and get there,” said Morrison. “If you respect the offensive line, you don’t want to get there late. A split second could get you beat in terms of an offensive lineman swallowing up a linebacker.”
There could be other factors at play that haven’t been tested or perhaps can’t be measured. For example, the distance wasted by linebackers on play fakes could be affected by the type of coverage that they are in. If a linebacker has a running back in man-to-man coverage, he might continue to move toward the running back to cover him. Also, the data could be skewed by teams that didn’t use play action enough, resulting in an unreliably small sample size.
The analytics suggests running in short yardage situations and the red zone
Sharp found that running the ball on third- or fourth and short had a significantly higher success rate than passing the ball. League-wide, in 2018, on third- and fourth down with 1 or 2 yards to go, rushes had a 13 percent higher success rate (70 percent) than passes (52 percent). Red zone rushes had an 8 percent higher success rate (50 percent) than red zone passes (42 percent). To be successful in these situations, teams have to be able to run the ball. This leads to the next question: whether teams could do so effectively if they don’t run the ball often. [h=3]Do running backs still matter?[/h]
If you want to find out what people value, follow the money. On average, running backs make less than every position group in football except for long snappers. Todd Gurley has the most guaranteed money on his contract among running backs by a wide margin … and yet he only ranks 36th overall in the NFL in that category. The next running back on the list is David Johnson, who ranks 73rd.
Analytics shows that running backs, even star running backs, have very little control over the outcome of the play over time. Hermsmeyer found that running game success is determined more by the box count, which could be manipulated by an offense’s use of personnel, and offensive line performance, which was measured by his version of adjusted line yards.
PFF’s WAR (Wins Above Replacement) score is a way of assessing player value by essentially answering the question: “if you replaced this player with a player off the practice squad, how many fewer wins would his team have over the course of the season?”
Running backs did not fare well in WAR relative to other positions. Christian McCaffrey was the highest-ranked RB in WAR last season but he was only ranked 36th in the NFL overall.
The WAR score shouldn’t be taken literally but it does provide a measurement of value. The running back position is the most easily replaceable, according to this metric, and many analytics experts have made the same conclusions in their own research.
The 2018 season provided plenty of evidence for the replaceability of running backs. Le’Veon Bell, who is widely regarded as a top-3 running back, sat out the season because of a contract dispute. He was replaced by former third-round pick James Conner, and the Steelers’ rush offense didn’t miss a beat — their yards per carry and rushing success rate actually improved slightly without Bell.
Kareem Hunt, who was the league’s rushing leader in 2017, was cut from the Chiefs right before their playoff run in 2018 after video was released of his physical altercation with a woman in a hotel hallway. The Chiefs’ rush offense dropped slightly but their rush success rate jumped up to 60 percent.
That’s not to say that you can plug any back into these offenses and expect the same level of success, but the numbers suggest that it’s much easier to find a replacement for even superstar running backs. So because running backs are so dependent on outside factors for production, how replaceable they are, and their short shelf lives, teams shouldn’t invest heavily in the running back position. The analytics suggests that teams could better use their cap space investing in assets that directly contribute to the passing game.
Sharp looked at the average cap hit per running back and compared that with team wins and found the relationship is negatively correlated. “What this means is that the more money a team pays to its running backs, the fewer wins they gain,” wrote Sharp. “Look at 2013 as an example: none of the top 5 teams in terms of cap hit per RB (Tennessee, Minnesota, Chicago, Oakland and Baltimore) finished the season with a winning record.”
The Cardinals awarded running back David Johnson a three-year, $30 million extension on the strength of his All-Pro 2016 season, and then he only played in one game in 2017 because of injury. In 2018, the Cardinals’ offensive line was ravaged by injuries and they fired offensive coordinator Mike McCoy during the season. Johnson’s circumstances were terrible and he only averaged 3.6 yards per carry. As skilled and talented as Johnson is, his positional impact wasn’t strong enough to elevate the rest of the Cardinals’ offense.
The Giants drafted Saquon Barkley second overall in 2018 and Barkley had a spectacular season in which he broke a ridiculous 94 tackles — the next-best running back was McCaffrey with 63 broken tackles. Barkley was so good that he accomplished the extremely difficult task of elevating his offensive line.
His line finished 29th in Football Outsiders’ adjusted yard line stat, which is a metric that attempts to separate the yards that an offensive line is responsible for creating in the run game. Yet the Giants finished 10th in the league in yards per carry (4.7 ypc). Even with Barkley’s remarkable season, the Giants only finished 16th in points per game (23.1 ppg) and they only won five games. His playmaking ability elevated the Giants’ run game unit, but even with a top 10 rushing attack, the Giants were still a mediocre offensive unit.
For teams that are rebuilding, it just doesn’t make sense to draft a running back early. Those teams usually don’t have the offensive line to maximize a running back’s impact and they likely won’t play with a lead often enough to run the ball with frequency.
However, though the analytics community might advise against it, I don’t think it’s a bad move for a contending team with a strong passing game or elite defense to draft a running back early. Running backs have shown they can make an easier transition from college to the pros relative to other positions, and when they’re on a rookie contract, they don’t cost much.
With all that said, here are the recommendations from the analytics community:
1. Don’t invest heavily in your run game because it doesn’t correlate with winning
2. Use more play action because it’s more effective than drop-back passing and it has proved to work without having to “establish the run”
3. Don’t run the ball into eight-man boxes
4. Test the limits of passing, don’t force the run “unnecessarily”
5. Use the pass to set up the run (run when defenses start to adjust to pass)
6. Deception in all its forms is the most important element in offense
7. The run game is valuable in short-yardage situations, in the red zone and for running out the clock.
But what about the other side of the argument? [h=3]The case for the run game from NFL coaches and players[/h]
Though it might be hard to define and quantify, the run game is still important, even if the analytics we’ve reviewed challenges traditional views on its value. Though some coaches are simply rigid and unwilling to change their long-held beliefs, some of the most innovative and thoughtful offensive coaches in the league, such as the Rams’ Sean McVay, also build around the run game.
In this section, I’m not attempting to make a case to run the ball more or less; I’m simply trying to clearly define the value of the run game. I talked to several current and ex-coaches and players about the subject and asked them to respond to the arguments that the analytics presents. Some coaches/players asked not to be named, but most of their answers are consistent with each other’s.
Less risk of a negative play
Passing is becoming more efficient but the inherent risk of passing the ball still has to be considered. When passing the ball, there is a greater chance of a negative play than with a run play. A negative play could be a turnover or sack. On a pass play, there is the chance of an interception or fumble. The quarterback could get stripped or an offensive player could fumble the ball after a catch. On a run play, the only significant risk of a turnover is a fumble by the running back.
In 2018, a turnover occurred 2.9 percent of the time during plays in which the quarterback dropped back for a pass (includes pass attempts, sacks, scrambles). On plays in which the ball was handed off or kept by the quarterback by design (example: zone read), a turnover occurred 0.6 percent of the time. Turnovers occurred almost five times more on pass plays than run plays.
Additionally, sacks (including ones that don’t end in turnovers) are often drive killers. Derrik Klassen of Football Outsiders charted the 2016 season and found that only 179/1118 (16.01 percent) drives in which there was a sack eventually got another set of downs. 83.99 percent of drives were essentially killed by sacks. Klassen found that defenses were 70 percent more likely to kill a drive after sacking the quarterback than they were likely to surrender another set of downs.
Running the ball allows offenses to minimize risk. This doesn’t mean teams should continually try to establish the run if they are outmatched up front, but with a higher volume of passes comes a higher chance of turning the ball over, and winning the turnover battle does correlate with winning.
Disarm the pass rush
An offensive line coach told me he might get kicked out of the fraternity for telling me that teams don’t need to run the ball to win, but they need to run the ball to pass. Obviously, not every coach or player would agree with that statement, but every coach and player that I talked to believes that running the ball helps pass protection because it forces pass-rushers to play the run and maintain their gap responsibilities rather than just recklessly fly upfield to rush the passer.
“The biggest mismatch on the field is defensive linemen against offensive linemen,” said 49ers run game coordinator Mike McDaniel. “Defensive linemen are supreme athletes and it’s a hard job to block people. Anything that gets defensive linemen to think about more than sacking the quarterback is a win.”
“The reason why you want to establish the run game is so you can unguard the defense so you can pass block,” said former Bengals and Cowboys offensive line coach Paul Alexander. “You can’t let them think pass, pass, pass. You have to keep them guessing.”
The very nature of pass blocking puts offensive linemen at a disadvantage. They are moving backward and reacting to pass-rushers. By running the ball, you allow offensive linemen to attack and it gives them a psychological edge. Run attempts, even if they aren’t resulting in big gains, can slow down pash-rushers and give offensive linemen a break from pass blocking.
Flawed statistics, unquantifiable effects and game theory
A coach from an analytics-driven team told me that some of the statistics used to prove the passing game’s superiority over the run game could be flawed or don’t tell the whole story because they rely on averages that could be swayed by big plays. He believes that it is better to look at median statistics when looking at the run game. The idea here is that 1) though the run game produces fewer big plays, it produces more consistent yardage; and 2) passing plays have a high variance (explosive plays, incompletes, turnovers, negative plays) that isn’t reflected by averages. There has been some discussion of using median and mode statistics to look at the run game but there haven’t been any thorough studies yet that I could find.
Count McDaniel as skeptical of the analytics: “A lot of stats are not quantifiable in terms of the pass game in regards to the run.” McDaniel and several other coaches explained that a strong run game affects the type of personnel on the field and the structure of the defense. For example, to stop the run, defenses have to keep run-stuffing defensive linemen in the game who aren’t as skilled at rushing the passer, or they have to keep bigger linebackers in the game who aren’t as skilled at coverage.
“A lot of defenses could take away explosive pass games if they were able to play two-high, two-man all day. The reason they can’t: You have run fits that are a hat short in the run game,” said McDaniel. “You have to earn single safety in this league. Single safety opens up some passing holes with five eligibles and three deep. You have to earn that, and to earn that, you have to draw another guy into the box because they’re afraid you’re gonna run it.”
McDaniel was an offensive assistant with the Atlanta Falcons in 2016 and he brought up their game in Week 4 against the Panthers during that season as an example of how the run game affected the defense in ways that are difficult to measure.
In the game before, the Falcons rushed for 217 yards against the New Orleans Saints. The Panthers came into the game knowing they needed to stop the Falcons’ rushing attack. The Panthers had a talented, aggressive defensive line and All-Pro linebackers Luke Kuechly and Thomas Davis attacked the line of scrimmage relentlessly from the very start of the game.
On the first offensive snap for the Falcons, the defense started out in a two-deep coverage, but a safety rotated into the box late to become the eighth man in the box. Quarterback Matt Ryan didn’t fake a handoff but the linebackers moved forward in anticipation of stopping the run. The safety’s rotation left receiver Julio Jones in a one-on-one situation and he took advantage with a nice gain to start the game off.
The Falcons’ success on the ground against the Saints in Week 3 affected the Panthers’ defense in Week 4 and helped the Falcons’ drop-back pass game even without play action. This effect could be difficult to measure by analytics. Against the Panthers, the Falcons only handed the ball off to their running backs 21 times for an average of 3.6 yards. They didn’t have a successful day on the ground because the defense game planned to take it away from the start of the game. But in doing so, they left their secondary vulnerable and Ryan threw for 503 yards while Jones had 300 yards receiving.
If you just looked at the numbers you would think the Falcons won this game solely on the strength of their pass game, but in reality, the fear of the Falcons’ run game left the Panthers’ secondary vulnerable.
McDaniel also said that in that game, they knew they were overmatched up front and called runs just to keep the defense honest. Though they were an outside zone team, they called “safer runs” (inside zone, duo) that are quicker hitting and carried less risk of losing yardage. The objective of rushing in this game wasn’t to necessarily gain yards but to maintain the defense’s respect of the run game and to aid the offensive line.
This pattern of play calling is an excellent example of how game theory could be used to explain why offensive coordinators call runs despite the fact passes average more yards. The goal of every play call isn’t to gain the most yards possible.
“It’s not about just maximizing expected value for any individual play, it’s setting up aggregate expected value/return,” said Chris Brown, the editor of Smart Football. “So a run play might be micro inefficient but macro efficient because of what else it sets up/protects.”
Additionally, offenses only go into a game with a certain number of plays in the game plan. Coordinators can safely call the same run concept over and over again and still safely gain yards, but if they call the same pass concept over and over again and defenders anticipate the concept coming, there is a risk that they’ll jump the pass and intercept it.
On every pass attempt, quarterbacks are computing massive amounts of information in a short period of time while they are getting harassed by pass-rushers. Even with elite QBs, it can be beneficial to give them mental breaks and allow them to hand the ball off once in a while.
Finite resources
Finding personnel for a pass-heavy attack is difficult. The quarterback position might be the most difficult position to fill in all of sports. There just aren’t enough starting-caliber quarterbacks for all 32 teams. It’s also hard to find reliable pass blocking tackles. This is why quarterbacks are the highest-paid players in the league while tackles are among the highest-paid position groups in the league.
It is easier to find good running backs and reliable run-blocking offensive linemen. Plus, most quarterbacks and offensive linemen need the run game to protect them from themselves. Even if teams wanted to, how many teams actually have the resources to pass the ball at a high volume? [HR][/HR][h=3]Final thoughts[/h]
After looking at all the research and talking with experts in the field, it seems clear the hard-line philosophy of the importance of establishing the run is outdated and disproven. But there is still value in being able to run the ball. It’s a tricky balance, but in building a team, general managers and coaches have to put more emphasis on investing in a strong passing game, but at the same time, they can’t lose sight of creating an efficient run game.
In 2018, of the top 10 teams in offensive DVOA (Football Outsiders’ measurement for a team’s efficiency by comparing success on every single play with a league average based on situation and opponent), seven were in the top 10 in rushing success rate. Eight of 10 had winning records and made the playoffs. So while there isn’t a correlation with traditional rushing success stats, there are correlations with running, winning and efficient offensive production.
This research is valuable for starting important conversations, but some analytics experts have reached hard conclusions that are too black and white for an extremely complex game. There is still value in the run game that hasn’t been properly explained by numbers. The subject should be researched further, but it seems experts on both sides of the argument are in a constant shouting match with each other in which nobody ever wins. Analytics experts, coaches, players, analysts and even fans could all benefit from listening to each other and learning from each other in this conversation to advance the game.
