The wild weekend of football led to many heated discussions regarding running the ball in the NFL. These conversations have been going on ever since the analytics community launched the “RBs Don’t Matter” campaign, a phrase that’s constantly misused and misunderstood. To me and most analysts, “RBs Don’t Matter” is an imperfect, oversimplified way of saying: 1) The running back position is more replaceable than other positions because it’s a position loaded with affordable talent, 2) rushing success can be explained by the number of defenders in the box and offensive line talent, not the running back himself most of the time, 3) passing offenses are more predictive of future success compared to rushing offenses, and 4) in general, passing is more efficient than running.
This column will take a deeper look at the fourth point: “In general, passing is more efficient than running”. I’ll give an overview of when NFL teams should pass or run the ball for every down and distance after analyzing @nflscrapR‘s 2019 play-by-play data. As you’ll see, I’m using expected points added (EPA) as evidence throughout the column because EPA predictive of future performance in the NFL. [[ad:athena]]
When To Pass and Run in the NFL
1st Down
How to read this chart:
The numbers on the bottom of the chart are the yards to go, so that yellow line at 0 is the first-down line just like you are used to seeing on television. The numbers on the left are expected points added (EPA). If you don’t know what that is, just know that we want to maximize EPA. The dots (red = pass, blue = run) are what we are comparing in this column. For example, Passing EPA (red dot) is at just under 0.2 on 1st-and-20, while Rushing EPA is at 0.11. That means passing is better than running for that specific down-and-distance.
1st-and-10: Almost Always Pass
The average expected points added (EPA) on passing plays on 1st-and-10 was 0.13 last season. That’s pretty damn good, especially in comparison to rushing plays (-0.05 EPA). In general, NFL teams should almost always pass on 1st-and-10. In my opinion, there’s not a single team that’s passing enough on 1st-and-10. This is the down and distance that teams are losing the most expected points over the course of a season by running.
1st-and-15 or 1st-and-20: Almost Always Pass
If there was an offensive penalty on 1st-and-10, then teams should most definitely pass on the next play. On 1st-and-15 and 1st-and-20, it’s way more efficient to pass than run. As you’ll see, that’s the case in all “blank-and-long” situations, except for when time remaining and score differential become major factors late in the fourth quarter.
1st-and-5: Pass or Run (Play to Strengths and Weaknesses)
Rushing EPA (-0.06) is slightly higher than Passing EPA (-0.09) on 1st-and-5, but the difference is so small. Coaches should make decisions based on their team’s talent and the opponent’s weaknesses. In my opinion, this would be a good situation to take a play-action deep shot.
2nd Down
2nd-and-Long: Almost Always Pass
Passing EPA (0.11) is way higher than rushing EPA (-0.04) on 2nd-and-10 plays. Nothing screams “I give up”, “I don’t trust my quarterback”, and “I’m afraid of winning” more than a 2nd-and-long run play. There’s nothing to suggest a surprising run works in these situations over time. It’s actually insane to me that some teams (NYJ, MIN, BAL, IND, NYG, PIT, SEA) run the ball over 35% of the time on 2nd-and-long. I mean, what the hell are the Jets doing by running the ball 52% of the time on 2nd-and-long.
2nd-and-Short: Pass or Run (Play to Strengths and Weaknesses)
On 2nd-and-2, 2nd-and-3, and 2nd-and-4, the difference between passing and rushing EPA is small. Probably small enough where coaches can actually make decisions with their own intuition and team’s strengths and weaknesses. You’ll rarely catch me complaining about a team running the ball on 2nd-and-kinda-short.
2nd-and-1: Pass (Use Play-Action)
2nd-and-1 is a different scenario. Teams ran the ball 76% of the time on 2nd-and-1, but EPA suggests a pass is the better play type. Passing EPA jumps to 0.25 here, while rushing EPA is down at -0.03. That’s a substantial difference. I’d guess that most defenses are ready to defend the run in this situation in an attempt to prevent a first down, which makes this situation an elite time for a play-action pass.
3rd Down
3rd-and-Long: Almost Always Pass
Teams rarely run on 3rd-and-long situations, so we don’t have to spend much time here. Of course, teams should be passing. Oh, by the way, teams should be throwing the ball 1-9 yards beyond the first-down marker on third downs. EPA and first-down rate are higher when the ball is thrown past the yellow line.
3rd-and-Short: Pass or Run (Play to Strengths and Weaknesses)
Passing and rushing EPA were pretty even on 3rd-and-2 and 3rd-and-3, a small enough gap that suggests coaches should make a play type decision based on their own team’s strengths and their opponent’s weaknesses. If a team does pass, I’d use play action and pass the ball beyond the first-down marker.
3rd-and-1: Usually Run
Rushing EPA (-0.05) is higher than Passing EPA (-0.17) in this short-yardage situation, so I’d lean on running the ball and would only use play-action if I opted for a surprise pass. Running the ball doesn’t mean bring in 800 linemen and 4,888 tight ends, however, because the goal is to limit the number of defenders in the box and offensive personnel determines defenders in the box. I’ll save the rest of my short-yardage rushing thoughts for a later column.
4th Down
The sample size is too small to draw conclusions for fourth downs from 2019 play-by-play data, but it should be similar to the third-down numbers we just went over; Pass on 4th-and-long and usually run on 4th-and-1.
2019 Team Rankings
Most Efficient Offenses on 1st-and-10
EPA suggests passing is more effective in this situation. This was true for 30-of-32 teams this year.
Rank | Team | Play Type | EPA | Play Type Rate | Notes |
1 | NO | pass | 0.44 | 55% | Way ahead of the league in EPA |
2 | DEN | pass | 0.33 | 38% | Didn’t pass nearly enough |
3 | OAK | pass | 0.32 | 40% | Didn’t pass nearly enough |
4 | KC | pass | 0.26 | 59% | KC should pass 90% of the time on 1st-and-10 |
5 | TEN | pass | 0.24 | 36% | Didn’t pass nearly enough |
6 | BAL | pass | 0.23 | 38% | Passing still had a higher EPA than rushing for BAL |
7 | SF | pass | 0.22 | 43% | Didn’t pass nearly enough |
8 | TB | pass | 0.22 | 46% | Didn’t pass nearly enough |
9 | SEA | pass | 0.21 | 47% | Didn’t pass nearly enough |
10 | ATL | pass | 0.20 | 53% | |
11 | LAC | pass | 0.19 | 50% | |
12 | LAR | pass | 0.16 | 53% | |
13 | BAL | run | 0.16 | 62% | The best rushing team was still more efficient passing |
14 | NYJ | pass | 0.15 | 55% | |
15 | GB | pass | 0.14 | 51% | |
16 | JAX | pass | 0.14 | 46% | |
17 | CHI | pass | 0.13 | 51% | |
18 | PHI | pass | 0.12 | 52% | |
19 | BUF | pass | 0.12 | 54% | |
20 | CAR | pass | 0.11 | 50% | |
21 | ARI | run | 0.10 | 51% | ARI was one of two teams that ran better |
22 | DET | pass | 0.09 | 42% | |
23 | CLE | pass | 0.09 | 49% | |
24 | CAR | run | 0.09 | 50% | |
25 | NE | pass | 0.08 | 52% | |
26 | DAL | pass | 0.07 | 45% | |
27 | CIN | pass | 0.06 | 49% | |
28 | PIT | pass | 0.05 | 45% | |
29 | OAK | run | 0.04 | 60% | |
30 | MIN | pass | 0.03 | 43% | |
31 | IND | pass | 0.02 | 46% | |
32 | GB | run | 0.01 | 49% | |
33 | CLE | run | 0.01 | 51% | |
34 | ARI | pass | 0.01 | 49% | ARI was worse passing compared to rushing |
35 | NYG | pass | 0.00 | 53% | |
36 | DAL | run | 0.00 | 55% | |
37 | DEN | run | -0.01 | 62% | |
38 | CIN | run | -0.01 | 51% | |
39 | SF | run | -0.01 | 57% | |
40 | BUF | run | -0.02 | 46% | |
41 | TEN | run | -0.02 | 64% | TEN was second in rushing rate on 1st-and-10 |
42 | PHI | run | -0.03 | 48% | |
43 | NE | run | -0.03 | 48% | |
44 | NYG | run | -0.04 | 47% | |
45 | NO | run | -0.04 | 45% | NO had the second-lowest run rate |
46 | HOU | run | -0.05 | 59% | HOU was one of two teams that ran better |
47 | IND | run | -0.06 | 54% | |
48 | WAS | pass | -0.06 | 32% | WAS had the lowest passing rate by far |
49 | SEA | run | -0.08 | 53% | |
50 | KC | run | -0.08 | 41% | KC had the lowest run rate |
51 | DET | run | -0.10 | 58% | |
52 | MIN | run | -0.10 | 57% | |
53 | JAX | run | -0.10 | 54% | |
54 | HOU | pass | -0.10 | 41% | HOU was worse passing compared to rushing |
55 | CHI | run | -0.11 | 49% | |
56 | LAC | run | -0.11 | 50% | Melvin + Ekeler combined to be one of the worst rushing offenses |
57 | PIT | run | -0.13 | 55% | |
58 | MIA | pass | -0.13 | 53% | |
59 | ATL | run | -0.14 | 47% | Needed to run less |
60 | WAS | run | -0.14 | 68% | WAS led in rushing rate on 1st-and-10 |
61 | NYJ | run | -0.15 | 45% | Needed to run less |
62 | TB | run | -0.16 | 54% | Needed to run less |
63 | LAR | run | -0.17 | 47% | Needed to run less |
64 | MIA | run | -0.17 | 47% | MIA’s rushing game was the worst unit in football |
Here’s how the final four remaining teams ranked this season on 1st-and-10 plays out of the 64 passing and running offenses:
- KC - Passing Offense (4th out of 64) and Rushing Offense (50th)
- TEN - Passing Offense (5th) and Rushing Offense (41st)
- SF - Passing Offense (7th) and Rushing Offense (39th)
- GB - Passing Offense (15th) and Rushing Offense (32nd)
It’s pretty clear that the Chiefs do not need any semblance of a quality rushing offense with Patrick Mahomes under center to be an elite offense. In fact, they’d probably be better if they passed the ball more than 90% of the time on non-short-yardage situations with Mahomes dealing. The other three teams have quality rushing attacks, but all three of their passing games were more effective on 1st-and-10s. Yes, that includes Derrick Henry and the Titans, who had the fifth-best passing offense in all of football.
Passing Rates on 2nd-and-5 or Longer
EPA suggests it’s best to pass these situations.
Rank | Team | Pass Rate | Pass EPA | Run EPA | Notes |
1 | NE | 79% | 0.04 | -0.17 | The NFL leader should pass 90% here |
2 | TB | 79% | 0.01 | -0.02 | |
3 | NO | 78% | 0.20 | 0.10 | |
4 | KC | 77% | 0.18 | -0.34 | KC had the worst Run EPA |
5 | LAC | 76% | 0.18 | -0.07 | |
6 | MIA | 75% | 0.13 | -0.10 | |
7 | LAR | 75% | 0.26 | -0.11 | LAR was 4th in Pass EPA |
8 | DAL | 74% | 0.34 | -0.18 | DAL led in Pass EPA |
9 | ATL | 74% | 0.02 | -0.13 | |
10 | HOU | 72% | 0.24 | -0.09 | HOU was awful on 1st down, good on 2nd |
11 | CHI | 70% | -0.04 | -0.19 | |
12 | CAR | 69% | -0.18 | -0.24 | CAR had the 3rd worst Run EPA |
13 | OAK | 69% | 0.19 | -0.04 | |
14 | DEN | 69% | 0.00 | -0.32 | DEN had the 2nd worst Run EPA |
15 | CIN | 68% | -0.10 | -0.20 | |
16 | CLE | 68% | -0.11 | 0.00 | Rare team to be better rushing |
17 | GB | 68% | 0.06 | 0.11 | GB was 3rd in Run EPA |
18 | NYG | 66% | -0.20 | -0.18 | Danny Dimes, Saquan struggled here |
19 | WAS | 66% | 0.08 | 0.06 | |
20 | ARI | 65% | 0.07 | -0.06 | |
21 | TEN | 65% | 0.30 | -0.11 | TEN was 2nd in Pass EPA |
22 | JAX | 65% | 0.07 | 0.09 | |
23 | PHI | 64% | 0.00 | 0.06 | |
24 | SF | 63% | -0.14 | -0.08 | Rare team to be better rushing |
25 | PIT | 63% | -0.08 | -0.14 | |
26 | SEA | 62% | 0.22 | -0.07 | |
27 | BUF | 60% | -0.11 | 0.14 | Rare team to be better rushing |
28 | DET | 60% | 0.10 | -0.08 | |
29 | MIN | 60% | 0.24 | 0.09 | |
30 | BAL | 57% | 0.27 | 0.21 | BAL was 3rd in Pass EPA, 1st in Run EPA |
31 | NYJ | 54% | 0.00 | -0.23 | Gase and Le’Veon were so bad |
32 | IND | 51% | -0.21 | 0.08 | IND was the worst in Pass EPA |
The Chiefs passed on 2nd-and-5 or longer 76% of the time but that number should be closer to 95% given their passing/rushing splits. Kansas City averaged +0.11 expected points added on passes and -0.34 EPA on runs, the worst in the entire NFL. Does that matter? No. Just let Mahomes deal and profit.
What about Tennessee? They have Derrick Henry. They should be running it, right? Well,,, probably not! The Titans had the second-best EPA on passes on 2nd-and-5 or longer (+0.30) and were average on runs (-0.11). This isn’t too say to completely go away from King Henry, but I wouldn’t take the ball out of Ryan Tannehill‘s hands in what-should-be passing situations, even when Henry is running like the best pure runner in the NFL. He should be saved for medium- or short-yardage and running-out-the-clock situations like these...
Rushing Rates on 3rd-and-1 and 4th-and-1
EPA suggests it’s best to run the ball in these situations.
Rank | Team | Rushing Rate | Rushing EPA |
1 | BUF | 100% | 0.21 |
2 | BAL | 94% | 0.42 |
3 | NO | 90% | 0.20 |
4 | NE | 89% | -0.11 |
5 | TEN | 87% | 0.91 |
6 | TB | 86% | 0.09 |
7 | KC | 85% | 0.17 |
8 | PHI | 83% | 0.26 |
9 | SEA | 79% | -0.41 |
10 | CLE | 78% | -0.66 |
11 | NYG | 78% | -0.35 |
12 | DET | 77% | 0.29 |
13 | OAK | 76% | 0.50 |
14 | CHI | 75% | -0.35 |
15 | LAC | 75% | 0.04 |
16 | SF | 74% | 0.24 |
17 | DAL | 74% | 0.14 |
18 | JAX | 73% | -0.05 |
19 | IND | 71% | 0.30 |
20 | MIN | 71% | 0.08 |
21 | PIT | 71% | -0.67 |
22 | CIN | 68% | -0.42 |
23 | ARI | 67% | -0.10 |
24 | CAR | 67% | -0.89 |
25 | GB | 67% | 0.17 |
26 | NYJ | 67% | 0.32 |
27 | WAS | 67% | 0.07 |
28 | MIA | 65% | 0.11 |
29 | ATL | 64% | -0.52 |
30 | HOU | 61% | 1.17 |
31 | LAR | 59% | -0.09 |
32 | DEN | 54% | 0.56 |
I don’t think it’s a coincidence that eight of the top nine offenses in rushing rate on 3rd-and-1 or 4th-and-1 made the playoffs this season. This is when teams should run the ball at a high rate, so it’s good to see some of the smartest teams get rewarded. The best teams, however, not only dial up rushing plays here, but they also dial up *good* rushing plays. All four of the remaining teams were really strong on short-yardage running plays during the 2019 regular season. The Titans averaged +0.91 expected points added thanks to King Henry, the Niners were at +0.24 EPA, the Packers checked in at +0.17 EPA, and even the Chiefs averaged +0.17.
Conclusion
Situations matter in the NFL. There are times when it’s inexcusable to not pass the ball (2nd-and-long, 3rd-and-long). There are times when its best to run the football (3rd-and-1). And there are even times when the analytics would agree that letting a coach decide to pass or run is best for the team (1st-and-5, 2nd-and-4, 3rd-and-2). It really is a game that requires a whole lot of context to figure out, but analytics can really help paint a picture of what is the best based on each unique situation. Unfortunately, it’s evident that analytics are not being used across the NFL right now when it comes to play-calling (For example, teams aren’t passing the ball nearly enough on 1st-and-10. 30-of-32 NFL teams were more effective passing the ball with a brand new set of chains, yet more than a handful of teams ran the ball more often than they passed on 1st-and-10). When teams really begin to optimize play calling with the helping hand of an analytics staff, they’ll have an even bigger advantage over the teams too stubborn to get with the program.
