Stat Analysis
Advanced analytics on player and team performance

Finding the Optimal Pass/Run Ratio

L.A. Rams QB Jared Goff and RB Todd Gurley
Photo: USA Today Sports Images

Guest column by Lau Sze Yui

Every football fan probably knows that passing in football is generally more efficient than rushing. But how much passing is really enough?

To answer this, we first need some prior knowledge of how pass/run balance works in NFL. For example, Sean Clement found that previous run attempts have no effect on a team's running success rate; hence, theoretically speaking, no matter how many runs a team has, its efficiency is still the same (and lower than that of passing). But one of the reasons why teams run the ball is to set up play-action, which utilizes the threat of running and is more efficient than normal passes. Therefore, the more precise question to ask is: is the value of play-action good enough for teams to mix runs into their play calling?

Ben Baldwin has written multiple articles on play-action and related findings (I, II, III, IV). To summarize his finding to a few points:

1. Play-action success doesn't depend on past success running the ball, nor on the number of previous runs.
2. Inversely, rush success doesn't depend on either past play-action success or the number of play-action passes.
3. Play-action is more effective in a run-heavy situation.

But still, no teams would call 100% pass plays in real competition to see if passing is still better than rushing. To tackle this problem, I've built a classifier to predict chance of passing on a given play based on offensive formation, score margin, time remaining, down, etc., and observe how expected points added (EPA) of play-action passes, non-play-action passes, and rushing plays change for different pass probabilities. As someone said, "In football, everything interesting is probably selection bias," so there are few things to note before drawing any conclusion.

  • Teams that have a worse passing game (and to a lesser extent running game) would fall behind more and thus run more plays in high-pass probability scenarios. Therefore the classifier has to be controlled for season offensive and defensive DVOA as well.
  • Although passing is more efficient than rushing most of the time, there are still cases where rushing is better -- i.e., short-yardage situations on third and fourth down. To limit the scope of our research, it's better to consider a fixed down-and-distance situation, and first-and-10 would be most convenient here.
  • The final two minutes of each half have been discarded.

With those qualifiers in mind, here is the result of EPA on first-and-10 assuming an average offense and defense for both passing and rushing game (EPA data courtesy of NFLscrapR; DVOA data courtesy of Football Outsiders):

EPA Pass Prob

(Ed. Note: The original version of this graph was slightly off; this is a new, more accurate graphic.)

The graph shows that rushing and non-play-action pass efficiency is mostly constant for any pass probability, while play-action pass efficiency decreases when pass probability is high. As a result, it's likely better off to pass most of the time on first down, but let's examine play-action some more.

Different teams obviously have different tendencies to call play-action passes instead of normal passes. For example, here are the play-action rates of the 2017 Steelers and 2018 Rams, two teams that called play-action passes at drastically different rates:

By using the EPA of different play types and their frequencies, we can calculate how the projected EPA values change across pass probability by summing the statistical expected value. Here, two teams -- one with low play-action usage (corresponding to 17 PIT) and one with high play-action usage (corresponding to 18 LAR) -- are plotted to demonstrate how play-action percentages affect projected EPA value for a team, assuming an average offense and defense:

The result shows that while higher play-action rate will lead to higher EPA, since it's less effective when pass probability is low, overall EPA is still at its peak when teams pass 100% of the time.

But does that mean teams should blindly pass 100% of the time on first down? The answer is simple: "it depends." Consider two teams in 2018 as a case study.

Patrick Mahomes was crowned as 2018 MVP with 50 touchdown passes, and his passing numbers were of course elite. In 2018, the Chiefs had a 62.6% unadjusted VOA on non-play-action passes but "just" 49.6% on play-action passes, while having a 12.2% unadjusted rushing VOA. As a result, we would expect their overall offense efficiency would be the highest when passing at a high volume:

Conversely, Jared Goff is obviously not as talented a quarterback compared to Mahomes, but thanks to Sean McVay's scheme and strong defense his team was able to get to the Super Bowl in 2018. The 2018 Rams had a 42.5% unadjusted VOA on play-action passes but only 27.4% on non-play-action passes. Here is the projected EPA versus pass probability:

Unlike what we saw with the Chiefs, the EPA for the Rams peaks when pass probability is close to 50%.

When designing run/pass balance, player ability should be heavily considered as well. For a rule of thumb, when non-play-action pass efficiency is higher than the average of play-action passes and rushing plays, teams should look to pass at much as possible on first downs. Otherwise, if the non-play-action plays are less efficient than that average, teams should look to play-action half of the time in order to "establish the play-action."

There are a few other factors teams need to consider as well:

1. Dropbacks increase a quarterback's chance of injury more than handing the ball off, and teams may not want to risk too many in the regular season.
2. There is no magic way to determine a quarterback's quality besides on-field snaps; therefore, at least in the first few weeks of a season, teams need to try different things to determine the best strategy.
3. Player emotion is also an important thing to consider. For example, some players, such as Ben Roethlisberger, don't feel as comfortable executing a play-action pass from under center; it's the major reason why the Pittsburgh Steelers have a very low play-action percentage with him as the quarterback.

To conclude, there are two type of optimal run/pass ratios in NFL (except for the case where your quarterback is so bad that rushing will be more efficient): if your quarterback is good at play-action but not that good on non-play-action passes, teams should run about 50% of the time in order to "establish the play-action." Otherwise, teams should theoretically pass 100% of the time.

Although it seems that it's the takeaway of the article, I am not suggesting that passing 100% of the time is a viable strategy. Instead, I'm suggesting that the correct passing equilibrium is still far away from current run/pass ratios.

Lau Sze Yui studies sports analytics in his spare time and can be found on Twitter @903124S.

Comments

52 comments, Last at 01 Apr 2020, 11:24am

1 Sometime when I am free in…

Sometime when I am free in the next few months, I plan to do an analysis of running the ball with scrambles removed to see how run efficiency has performed over time. Just a hunch, but I suspect its never been a worse time to run the ball in the NFL unless your qb is a scramble specialist.

In that vein, even if your qb is bad, I suspect its still better to pass near the 100 percent mark, at least in today's nfl. Passing is so far ahead of running on average that unless circumstances dictate, you should just keep passing. Perhaps a rule change to discourage so much flagrant passing may be in order

6 Something like that coming

A look at how running efficiency has developed over the last 30 years should be coming at some point soon on this website. We'll see how much work I do to try to remove scrambles from it.

7 In thinking about, it gets…

In thinking about, it gets quite complicated. Running in general is also a lot more selective than it used to be. Lot more teams probably run in favorable down and distances in today's nfl. I also wonder how much give up runs on third and forever have evolved compared to the past, in an era where interceptions were much more tolerant.

14 Passing is so far ahead of…

Passing is so far ahead of running on average that unless circumstances dictate, you should just keep passing.

The more correct way to state this is that the rules have been bent to the point that only passing makes sense. Without entirely realizing it, we turned the NFL into the CFL.

\It's interesting that in the 1985 analysis, rushing actually makes sense

18 I think this is one take…

I think this is one take. The other take is that rushing has also declined in value over time. Team's are weaker at running the ball than they have ever been. My first time seriously following the league was 2003.

14 running backs had over 1200 yards rushing.

9 running backs with over 1400 yards.

4 running backs with over 1600 yards.

Running backs were so prominent that LT with over 1600 yards rushing was not a probowler. 

 

Now look at 2019:

5 running backs had over 1200 yards(Lamar Jackson was the 6th leading rusher)

2 running backs had over 1400 yards.

0 running backs were over 1600

 

The story is even worse when we look at 2018.

 

Something has fundamentally changed and we're not talking about an era that was 20 years ago. Rule changes explain some of it, sure, but something has made running just not as effective as it used to be. 

27 2003 was at/near the peak of…

2003 was at/near the peak of the workhorse back era.

13 RBs averaged 20+ rushes per game that year:  in 2019 it was ONE (Derrick Henry).  9 RBs in 2003 had 320+ carries; this year it was ZERO (Henry missed a game; only he and Elliott topped even 300 rushes).

Running by RBs is down from 2003:  there were 24.9 RB carries/game that year, compared to 21.9 this past season - a drop of 12%.  But carries by teams' lead backs were down more:  16.2 to 13.2, a 19% decline.

The game has evolved in at least two ways in 16 years:  fewer runs by RBs overall; less use of workhorse backs who dominate each team's carries.  It's the latter trend that caused the changes in rushing yard totals that you cite.  But using 2003 as the start point of the analysis is slightly misleading.

I should update this study.

https://www.footballoutsiders.com/ramblings/2009/rb-usage-part-i

https://www.footballoutsiders.com/ramblings/2009/rb-usage-part-ii

28 I didn't mean to…

I didn't mean to intentionally take 2003 to skew the numbers. It was my honest to god first year really following the league. I had no idea it would be a watershed moment.

2 Great research! Looking at…

Great research!

Looking at the middle section of the first graph, from 38%-62% pass probability, there's a monotonic pattern where, as pass probability goes up, the success of PA goes down a lot, the success of non-PA passes go down some, and the success of run plays goes up a little. That pattern makes a lot of sense - plays become more effective when they're more surprising.

Intuitively, we might expect the same monotonic trends across the whole range, but that's not what the graph shows. Most relevantly, the success of non-PA passes starts increasing rather than decreasing around the 85% pass probability mark. That increase in the 85-100% range looks to be what is driving the two-peaked pattern in the rest of the graphs. So I'd love to take a closer look at what sort of plays get classified in the 95+% pass probability range to see if we can understand why passing is so successful there.

One possibility that comes to mind is that a lot of these might be plays by teams that are trailing by a lot, where the defense is playing soft in order to prevent a quick score. Those plays might have good EPA, but not generalize to a pass-happy approach throughout the game.

Another possibility is that those might be teams that come out in empty backfield sets, or otherwise obvious passing formations. And there might be a selection effect there, where teams with a good passing offense use those formations more often than teams with a bad passing offense.

It also could be random noise, if the sample size there is small enough.

Seems like your data set should provide ways of investigating these possibilities, in addition to just looking at that subset of plays to try to see what's going on. Those are the next things that I'd want to look at after seeing this analysis.

5 100% pass probability

Maybe this is an answer--team has extremely high pass probability (think 1st & 10, 4th Q, down 14). Team passes, picks up ~20 yards and moves from midfield to the opponents' 30. No defense should want to give that up (at any point in the game), and it would obviously generate good EPA. 

However, and this part of my post is directed at the author--how often is that the outcome of the play? Unless I am misinterpreting the graphs (possible), this is not taken into account.

Let's assume our team has a 90/10 pass/run ratio, to account for some short-yardage rushing plays that are more likely to succeed than passes. Once a defense knows that we are going to play like this, they have a much better chance to stop our offense because they know our tendencies. Obviously, this is why play-action passes, draws, reverses, etc. are used. On the other hand, we have all seen times where a run/QB sneak/Hail Mary/quick pass out-of-bounds to set up a shorter, end-of-half FG/etc. is successful, even when it was obvious what the play call was/should have been.

The question that many an OC tries to answer is--where is that equilibrium? Realistically, running when the defense expects a pass, and passing when they are expecting a run, is a good formula for success, except in the most extreme scenarios. That is why we constantly hear about "staying on schedule," and indeed DVOA accounts for that. Also, "staying on schedule" allows for teams to be more unpredictable, and generally leads to longer, more effective drives and more points. Anyone posting here knows these basic ideas.

So--where is the equilibrium? IMO, it boils down to: 1--your personnel (which the author mentions)--basically, Mahomes is better than Haskins, the Chiefs should pass more. 2--game situation (which the author wisely accounts for) 3--where are your match-up advantages or disadvantages? 4--in specific game situations, can your offense execute if the defense knows what is coming? 

Basically, there is no right answer. In some situations, anything but a pass would be foolish; in other situations, anything other than a run would be. Some situations are great for play-action, others are not. Some match-ups may favor running more b/c of injuries, team strengths, etc.--some would favor passing for the same reasons. 

For another guest article, I would love to see teams' record based upon the run/pass ratio BEFORE the last half of the 4th Q (or last 5 minutes), to eliminate clock concerns. I would suspect that somewhere around 60/40 is the best spot.

16 Thanks for your comment…

Thanks for your comment. Excluding plays from last few minutes don't affect the result too much since in model empty formation is considered as 100% pass play as well.

21 Excluding plays

My only idea for excluding plays before the last 5 minutes is to view pass/run ratio on both 1st & 10 plays, and also all plays in general, that results in WINNING vs. LOSING. Obviously, using EPA or WPA is an approximation of this.

There will be some bad weather games that should probably be excluded from the sample; but my curiosity is what ratio would be ideal based on game results? I am sure there will be some games where team B comes back in the last few minutes to beat team A because of recovered onside kicks, all-time blunder plays (Herm Edwards TD return of a fumble against the Giants), etc., where their pass/run ratio was outside of a standard deviation from the mean. But, if we included those last few minutes, the pass/run ratio of some of those winning teams would probably move back inside that 1-standard-deviation band, and others would move out--probably a net wash.

IMO--straying too far away from 60/40 pass/run ratio, in today's NFL, is unlikely to lead to victory. If you are passing too much on 1st down, you are probably playing catch up; if you are running too much, you will probably need to catch up. I am pretty sure that FO has proven that the best teams pass to get ahead, and then run to stay ahead (i.e., run the clock out). Having said all this, I would sure love to see the numbers that show that ratio.

EDIT: Just saw that the first graph has been edited. It seems to show that the highest EPA for passing plays, both with and without play action, peaks at around 37.5% pass probability; and rushing EPA seems to peak at about 65% pass probability. In other words, it seems that the author has proved that the best plays are passing when it looks like a run, and running when it looks like a pass. Not only that, but play action passes when it looks like a run work better than a pass play without a fake. I would suspect that's why RPO's are becoming more popular--do the opposite of what the defense is prepared to take away.

 

22 Why do you feel going to be…

In reply to by Joseph

Why do you feel going away from a 60-40 split would be suboptimal in general?

The theory I keep reading is teams adjust play calling and formations so if you skew too heavily in one direction ,you are not going to take advantage of the cross-match. 

 

but every study trying to correlate pass efficiency and run efficiency has failed to show up significantly in the data. Play action is almost independent of running effectiveness. It seems the overwhelming drivers are down and distance and of course weather. 

 

My argument has always been less about how efficient passing offenses are and just how inefficient running offenses have become. Show me a team that has an ineffective passing game and I'll see the same team show up with an even worse running game.

26 My argument has always been…

My argument has always been less about how efficient passing offenses are and just how inefficient running offenses have become. Show me a team that has an ineffective passing game and I'll see the same team show up with an even worse running game.

Part of the above analysis is that if you have one, you tend to have the other.

At the extreme end, consider option teams in the NCAA. They pass rarely, but really efficiently, because the defense is committing everything to stopping the run.

The Vick Falcons are a counter-example, though. In 2004, they were the 3rd-most efficient rushing team (and the most prolific in terms of yardage, also lead league in yards/attempt), but the 31st-ranked passing offense (2nd fewest attempts, 3rd fewest yards, because the Steelers, Ravens, and Bears still existed). You will see this distribution with teams that rely on running QBs or option offenses (ATL) or who have really good offensive lines (Pittsburgh; 5th in rushing DVOA).

30 60/40 split

Just my opinion, but I think that the 3 graphs here that show a pass is most effective at the 37th percentile, and a run at about the 60th percentile, seems like that should be the number. The math may point to a 62/38, or 65/35, but still pretty close to the 60/40 split.

The extremes are obviously ridiculous--but in general, NFL defenders are too good for an offense to repeatedly indicate, "we are going to pass/run" without the defense stopping it. Down & distance, score, time remaining, etc. dictate the pass or run in certain instances; personnel, formation, and down & distance also can be reliable indicators on certain plays. But when you are dealing with 1st and 10, outside of late-game scenarios, an offense should be as "unpredictable" as possible: mathematically, that means 50/50. However, given what we know about passes being better than runs in general to maximize winning, tilts the odds toward passing. Too much passing, you give up the unpredictability. Toward the end of the game, running is better for the winning team, because of running out the clock. 

Thus, idealistically, it should probably be 67/33 for the first three quarters to try to build a lead, and more like 50/50 in the 4th (percentages approximate), as your last few 1st & 10's will hopefully be runs to bleed the clock.

 

Regarding efficient offenses, I agree that it is rare to have a team great at running and horrible at passing, and even vice-versa. But, I think this is also a product of personnel. Think about the 99 Rams--to try to stop their passing, you needed more DB's. That just meant that Marshall Faulk could run all over you. Also, the Pats a decade ago--with Gronk and Hernandez on the field, they could attack base defenses with the pass, and dime defenses with the run. In other words, the best offenses are good at a lot of things, partly because they have personnel that attack defenses in multiple ways without substituting. Inefficient offenses have either sub-standard personnel (likely), or also have "tells" that make them easier to defend. (Think about the speedy gadget player who only plays about 5-10 snaps per game.)

Slight tangent on this argument--Taysom Hill was tendered at a first-round level, because he has become a reasonably effective runner and receiver, who can still play QB at a back-up level. Because Sean Payton has expanded his role to line up practically everywhere except O-line, Hill coming into the game does not mean QB option. While he is pretty good at that, he can also line up in the slot and run a pass route. Because he also is the regular up-back on the punt team, a fake can be executed without any change in personnel. In other words, in my opinion, his biggest value to the Saints is the unpredictability he brings to the offense. 

34 My one counter to the above…

In reply to by Joseph

My one counter to the above on predictability, teams have introduced unpredictability with passing, it just takes the form of screens, quick hitches, flea flickers etc etc. The innovation in passing has been squeezing more mileage out of short passes. I think the run vs pass distribution has a static notion of what passing involves.

I don't have a sense of this is, just a hunch. it's much easier to creatively gain yards through the pass than it is through the run( unless your QB is mobile). 

35 Precisely. The entire run…

Precisely. The entire run/pass dichotomy is somewhat less than accurate. Now, do some teams try to execute what would classically be labeled a run more frequently than what would be optimal? Sure, it is very likely! Football coaches, like all people, are prone to poor risk/reward calculations! However, determining exactly when that is happening is far more difficult than what even the most advanced analysis typically lets on. 

11 I've thought about this as…

I've thought about this as well and look at plays with empty formation (hence almost 100% of dropback except jet motion), and EPA per those play are still similar. Hypothetically it would be a better to separate pass from trailing vs pass from empty formation, but since there is only a small difference it didn't make into the article

4 3rd and 15 is a fantastic time to run the ball. Why?

Because the defense is defending the pass. They won't let any passes reach the down marker, but they AFFORD to be porous on the line, because they'll catch a RB before he gets to the down marker. They don't have that luxury on regular downs when a team might be expected to normally run the ball because they HAVE to always respect the run. Look what would happen if they didn't. Huge chunks of yards every play if they played as if the QB was always going to pass.

So what if there never even was a RB? What if the QB HAD to throw on every play? they'd tee off on him every play. They'd play a different style of defense completely. And that's what would happen if teams abandoned the run.

Which is why these studies on how useful the run is completely miss the point.

Play action, not play action, makes no difference. It isn't about EPA, it's about how many MILES a RB would get if the defense didn't account for him period. He HAS to run the ball, so that they HAVE to account for him.

You don't have any data showing how poorly a QB would fare if there were no RB because there is no such data.

31 Belichik and Brady spent a…

Belichik and Brady spent a good chunk of the past decade happily piling up super easy victories, by running the ball down a defense's throat, whenever some analytical mastermind decided that only the Pats passing game should be accounted for. You'd think more people would have noticed.

36 Yes, the Patriots are…

Yes, the Patriots are masters at finding what defenses are giving them, exploiting matchups, manipulating personnel, etc. They've won a few games over the years where they've managed to overwhelm their opponent in the run game. But going back to 2007, when they ushered in the era of the shotgun as the base formation, they are probably the team that has exploited the benefits of a pass-first philosophy more than any other.

This is a team that almost always plays from ahead, and yet has (with Brady in the lineup) averaged 37 pass attempts per game and finished in the top 10 in pass attempts nearly every season for a decade. A decade where they've won 12.5 games per year with an average point differential of +167. There aren't many Tom Bradys out there, but NE is like Exhibit A of what you can accomplish by pushing the boundaries of passing.

37 See, it's a mistake to even…

See, it's a mistake to even call their philosophy "pass first". Their philosophy is "exploit the most vulnerable match up, adjusted to yardage needed in that specific point of the game, to deem the play a success, first". That's before we even get to the somewhat false delineation, or certainly unnuanced division,  between what is labeled a run, and what is labeled a pass. Now, it is true, that all other things being equal, defending what is typically thought of as a pass requires a greater amount and diversity of talent distributed among 11 players, than what is typically thought of as a run, which means on average passing will be more efficient. But defensive coordinators know this! Which means offensive coordinators have to account for that knowledge.

I don't mean to be too critical. I actually like the piece. I just think that in reading it, too much of this game's mind-boggling complexity can be overlooked.

38 The problem is, the pendulum…

The problem is, the pendulum never swung back. Ok, so defenses have had nearly 20 years to adapt to the changing environment, yet the efficiency difference between the run and the pass has only grown. The Patriots, for all of their ruthless exploitation, have been a pass first team ever since 2006. They run very efficiently, but they've never stuck to a philosophy that someone like the Titans employed during the postseason. I can't remember the last great offense that went exclusively through the run. The 2012 Vikings offense was still bad in the aggregate. 

Going back to my original statement, there's been a decline in standard running for years now and I wonder if this is related to the offensive line or the offense in general. As the pass continues to dominate, defenses adjust, but so do the players on the offense. Linemen are expected to be better pass blockers at the expense of run blockers. Tight ends are more built like hybrid receivers and the receivers themselves become worse blockers. Running backs now are prized for their pass blocking and receiving skills and it trickles down across the board. That is my theory. 

40 To be more accurate, every…

To be more accurate, every time the pendulum appears as if it will start back, the rules are altered to prevent it from happening, by making passing easier. To the point that working on being good at running the ball becomes very much a secondary goal.

39 I understand the point you…

I understand the point you're making. It's certainly an oversimplification to think of run/pass in binary terms like choosing one or the other is the only decision you make when you call a play (I know that bothers me), or that all runs and all passes are alike.

If I had to sum it up: a lot of NFL coaches have no problem acknowledging that passing plays are more efficient on average than running plays, but seem to think that skewing their offense toward a greater emphasis on the passing game is going to come back to hurt them for reasons that we consistently find don't affect passing or rushing success - things like the need to achieve a "good balance," to establish the run, to set up play action, to avoid 2nd and 10 like it's a dramatically worse situation than 2nd and 7, etc. I think there's still plenty of room for NFL teams to re-calibrate their thinking. They're nearly all more capable of attacking offenses through the air, or dictating terms to defenses through the passing game, than they realize they are. And the costs or trade-offs they've internalized about calling more pass plays may not be as bad as they think, or may not even hold up at all.

41 This is a team that almost…

This is a team that almost always plays from ahead, and yet has (with Brady in the lineup) averaged 37 pass attempts per game and finished in the top 10 in pass attempts nearly every season for a decade.

It's interesting how many plays New England runs. Since 2007, they've led the league in plays run more times than they've finished outside the top-10. Yes, they pass a lot. They also run a lot. They usually slot in towards the bottom half of the top-10, finishing 2nd or 3rd three times.

I suspect this is a reason DVOA loves them (and slightly less, the Eagles, who seem to operate similarly).

17 common sense talk is frowned…

common sense talk is frowned upon here , sir. itn here and other itnernt places, we beluieve it is possible for running backs to be eradicated. every offensive play could and showuld have a quarterback, two tackles, two guards, a cmneter, and then a combination fo five wide receivers and tight ends. nobody in backfield except quarterbakc. 

23 Here's the thing about the…

Here's the thing about the point you make: so long as running the ball is legal, RB on the field or not, how much would a team have to pass on 1st and 10 before defenses completely give up accounting for the run? As long as a team *can* run the ball if it chooses to, that limit might be very, very high!

If a team called passing plays 90% of the time on 1st and 10, but still ran the ball in other situations and demonstrated that they have running plays in their playbook using the same formations they use for pass plays - this can include empty formations - could defenses really give up defending the run? Schematically, I don't think they could, for the reasons you mention. But that wouldn't actually require offenses to run the ball at any specific rate on 1st and 10, like 40% or 50%. I think they just have to show that they *can* run the ball, and I think this is borne out pretty well by the fact that the great work done here and by other analysts in recent years has repeatedly shown that there's no relationship between rushing rate and passing success within the ranges that NFL teams currently operate in.

And if defenses do start dramatically adjusting to an extreme passing tendency, like lining up with 4 guys in the box or training their LBs to take 3 steps back at the snap, no sane analyst is going to tell an offense not to try to take advantage of that by running straight ahead. Until defenses start doing that though, it sure looks like there's still plenty of room for offenses to continue to increase passing rates in these neutral types of situations.

29 I suspect this is correct. …

I suspect this is correct.  The threat of the run is sufficient.  As long as defences deploy balanced tactics, offenses should pass, not run.  Whether they put an RB in the backfield or not depends on what they need to do to keep the defence from selling out against the pass.  But they should never use that RB until the defence changes personnel or alignment sufficiently to improve the probability of run success.  Of course there'll be constant cat-and-mouse games by the defence about what they're doing, but until defenses are routinely pulling numbers out of the box, the pass rate should likely be 95% plus for traditional offenses.

The flip side of that is that a non-traditional offense such as 2019 BAL has already adjusted personnel to take advantage of current "standard" defensive personnel.  It's likely that they can and should run much more often than a traditional offense until they pull numbers into the box.

9 Do not forget the red zone

I think this needs to be looked at with short fields as well. As a Seahawk fan, my opinion of pass every time is not high(hand off to Lynch and we have two SB wins). Seriously, has there been any work on the red zone or, even better, on short fields based on five yard increments?

19 Game Theory

It looks like most of the good "veteran comments/?s" surround the "prisoner's dilemma" aspect to all this.

Similar to that semi recent article a guy did about run success prediction based on formation, and how it doesn't really represent reality, since if you line up (in say) 23 personnel, the defense is likely to put 8 in the box (cat and mouse)...ie you can't just look at the offensive personnel, but instead, somehow try to simultaneously control for the defense's reaction. (I know you controlled for a lot, but not sure if I saw mention of something that would DIRECTLY address this).

Either way, great work...really well done.

20 I had a similar reaction. It…

In reply to by RyanNewman20

I had a similar reaction. It reminds me quite a bit of this article about sacrifice bunting in baseball: https://blogs.fangraphs.com/were-the-yankee-sac-bunts-in-the-8th-inning-correct/

32 This is where advanced…

This is where advanced football analysis so frequently fails. The best cure I can think of is to go back and chart the Pats offense over the past decade, paying very close attention to personnel and presnap alignments.

42 I don’t think it “fails” so…

I don’t think it “fails” so much as it chooses to make assumptions about (or completely ignore) the variables that can’t be measured or, in some cases, even defined. Analytics as a field is primarily interested in measuring the outcomes that actually happen and extrapolating from there.

Having said that, a more process-oriented approach to football efficiency would definitely make for an interesting article. 

33 To follow on, the question…

To follow on, the question is never "Should the offense pass or run, and if it passes, should they pretend to run first". The question is always "What individual players are on the field, with their particular skill sets, and how can we get the offensive players aligned in a fashion which will force the individual defenders to try to do something they aren't good at, relative to other skills, while our offensive players are doing things they are good at, relative to other skills.That means we need offensive players who are more multi-skilled than the norm".

43 I would spin this in a…

I would spin this in a different direction. Let's say you are Bryan Flores pre free agency. You are starting from as close to scratch as it gets. What type of team should you try to build? How might you go about molding your team? 

I think those questions are interesting from a defensive perspective because you can go in many directions, but I'd argue its pretty basic answer on the offensive side at least in this respect. Given the efficiency difference, the conversation should be, "We need to pass the ball well". That has to be every gm's defacto response to team building. 

The run game has sadly become the like the post game in the NBA. 

 

44 You can have counter-factual…

You can have counter-factual success.

The Rams made a SB with Jared Goff. The Royals reached two WS and won one based on signing players who were fast in an age of power. The Lakers had the best record in the NBA in part by playing three PFs.

46 "The Rams made a SB with…

"The Rams made a SB with Jared Goff."

 

That looks laughable right now, but they had the number 5 pass offense that was higher in efficiency than their number 1 rush offense.

 

"The Lakers had the best record in the NBA in part by playing three PFs." 

 

Assuming you are referring to the current Lakers, they in fact have the 2nd best record in the nba and that was achieved because they have 2 of the top 6 or 7 players in the nba and one of those players is a pantheon level player. 

45 Well, by definition, without…

Well, by definition, without above average quarterbacking, it is impossible to have a multidimensional offense that is a threat to score from anywhere on the field, so you start there. The problem with the 2012 Vikings offense was not that they were paying too much money to Adrian Peterson. The problem was that they had drafted Christian Ponder with a high 1st round pick to play qb in 2011, instead of Russell Wilson with a 1st or 2nd round pick in 2012, and I will maintain that it was very knowable in 2011 that Christian Ponder was not worth a high 1st round pick, and that Russell Wilson was a very strong value anywhere after the 1st round (at the time I was saying he was a 1st round value).

 

47 My point about the vikings…

My point about the vikings illustrates exactly why the NFL has moved on from running game as a successful engine of an nfl offense. If your passing game is great and your run game is absolutely horrendous, you can still have a good offense. If the reverse is true, its most likely to be at best average and more realistically below average. That is today's nfl unfortunately. 

48 Well, sure, but that's a…

Well, sure, but that's a long way from the typical arguments put forth, like "running backs don't matter", "never pay veteran running backs a lot", or "never draft a running back in the 1st round".

50 I think the arguments…

I think the arguments against running backs are not absolute statements but more like how we should view the position with regards to tradeoffs in efficiency and the salary cap.

As we all know, everything in the nfl matters to an extent, but long snappers and full backs fall at the bottom of the totem pole for a reason. 

In thinking about all of the teams that paid big prices for running backs, all of them had or seem to be having buyer's remorse. Dallas, the Rams, the eagles with Murray, etc etc. And I maintain that drafting Barkley was a mistake for the Giants, but that's a different story. In cold brutal terms, I'd probably keep franchising the running back until became unpalatable to do so. It doesn't make sense to pay veteran backs big money anymore.

 

51 I didn't like the Rams or…

I didn't like the Rams or Cowboys signings because the players involved simply aren't explosive enough. The Barkley draft wasn't great because the rest of the roster was too crappy. You really want to get value on a 1st contract with an rb drafted in the first round, before the pounding takes effect, which means a team ought to have some talent in place. I tend to agree that franchising great rbs, when the rookie deal expires, makes the most sense, in brutally utilitarian terms. 5 years on the rookie contract, 1 franchise tag year, and then adios, as the 28th- 30th birthday appears on the horizon. I really wish the new CBA had addressed this.