Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2017 Play-Action Defense

Our look at play-action pass in 2017 flips to the defensive side of the ball. Carolina was historically good, Houston was historically bad, and a long-standing question about year-to-year correlation gets cleared up.

11 Jul 2018

2017 Run Offense by Number of Backs

by Bryan Knowles

As we continue our pre-Football Outsiders Almanac 2018 stat bonanza, we turn our attention to how offenses performed running with either one or two backs in the backfield. We're talking about I-formations, pro-sets and other relics of a bygone age.

As you might have gathered from our offensive personnel analysis, two-back sets aren't exactly popping up all over the place. Even when teams do put two running backs onto the field, they're splitting at least one of them out wide roughly 20 percent of the time. The percentage of running plays out of two-back sets has declined accordingly; it would be a massive tell if teams were always rushing fullbacks and H-backs onto the field every time they wanted to run the ball. The correlation between league-wide use of 11 personnel and two-back rushing percentage is -0.98, which is exactly what you would expect.

The same trend holds up on the team level, as well. Every team's percentage of total plays in two-back sets and running plays in two-back sets are within five percent of one another. Some teams may talk about "running the ball when the other team expects to you run," but no one wants to be the team flashing their obvious "this is a running play" sign on a regular basis.

We're harping on this because, to be clear, that's the only correlation two-back running has. There is essentially no correlation between a team's frequency of two-back running and their success when two-back running. You can measure it in the DVOA difference in running out of two-back or one-back sets, or you can measure it in the difference in yards per carry. Either way, the correlation coefficient is just about 0.1.

There are several possible explanations for this. For one, the passing game is more important than the running game, so if a team finds they pass better out of certain backfields, they may be willing to accept some subpar runs. If the overall efficiency climbs, then it's a trade-off worth making. Other teams may be stubbornly sticking with two-back sets in the thought that those are inherently better in the running game, and will allow them to set up the pass better. This isn't backed up by statistics or evidence or ... well, reality, but it is backed up by memories of the 1970s.

Whatever the reason, the important thing to remember is that usage is not correlated with success. Some of the teams that run most frequently out of two-back sets hover around the league average when actually doing so. Some of the teams that very rarely bring extra men into the backfield find success when they bother to do so.

For the purposes of this analysis, we're considering rushing attempts by any player who was lined up in the offensive backfield but was not at the quarterback position. Single-back formations mean one player other than the quarterback was lined up in the backfield and ran the ball; two-back or multiple-back formations mean more than one player other than the quarterback was in the backfield, and one of those players ran the ball. The numbers include direct snaps to a player who was lined up as a running back, but do not include scrambles, quarterback keepers, wide receiver or tight end end-arounds or reverses, or "Wildcat"-style plays. It only includes runs by a player lined up as a running back at the snap.

2017 Offense: Running with 1 or 2 RB in Formation (Not Personnel)
Offense Pct 2+ RB Rk 1-RB Rushes 2-RB Rushes Difference
SF 54% 1 162 4.4 -6.5% 14 192 4.0 -14.5% 17 0.4 8.0% 11
BAL 50% 2 210 4.9 2.7% 4 209 3.5 0.8% 7 1.4 1.9% 16
NE 49% 3 207 4.2 1.1% 6 196 4.6 19.3% 4 -0.4 -18.2% 27
BUF 40% 4 232 4.0 -19.3% 24 152 3.9 -17.7% 23 0.0 -1.7% 18
JAX 39% 5 280 3.7 -18.8% 22 182 4.3 -6.1% 11 -0.6 -12.6% 26
CAR 34% 6 221 3.7 -28.4% 31 112 3.6 -0.9% 8 0.0 -27.5% 31
TEN 33% 7 248 4.3 -4.3% 11 124 3.4 -13.8% 16 0.9 9.4% 10
CHI 33% 8 246 4.3 -3.9% 9 121 3.5 -24.5% 26 0.9 20.5% 6
ATL 32% 9 261 4.2 -12.2% 17 121 4.2 -10.4% 14 0.1 -1.7% 19
CLE 32% 10 199 4.3 -4.0% 10 92 3.9 -16.1% 21 0.4 12.1% 9
GB 31% 11 221 4.4 5.2% 2 97 3.4 -14.6% 18 1.0 19.8% 7
PIT 30% 12 276 4.0 2.6% 5 116 4.1 -4.7% 10 -0.1 7.3% 12
LAC 28% 13 270 4.3 -8.6% 15 106 2.9 -41.7% 30 1.4 33.1% 3
TB 27% 14 242 3.8 -19.8% 25 88 2.9 -17.1% 22 0.9 -2.7% 20
NYJ 27% 15 273 4.1 -21.6% 26 99 3.5 -34.6% 28 0.6 12.9% 8
DEN 26% 16 299 4.3 -5.7% 13 106 3.3 -36.1% 29 1.0 30.3% 4
OFFENSE Pct 2+ RB Rk 1-RB Rushes 2-RB Rushes Difference
CIN 26% 17 245 3.8 -12.2% 18 86 3.4 -10.9% 15 0.4 -1.3% 17
NO 25% 18 294 4.9 9.1% 1 99 5.6 20.7% 2 -0.7 -11.5% 25
IND 23% 19 289 3.6 -15.7% 20 88 3.8 -9.3% 13 -0.2 -6.4% 22
HOU 23% 20 303 3.8 -19.1% 23 89 3.8 -15.5% 19 0.0 -3.7% 21
DAL 21% 21 325 4.1 -3.0% 8 89 5.0 19.3% 3 -0.9 -22.3% 30
KC 20% 22 256 4.4 4.0% 3 65 5.1 1.1% 6 -0.6 2.9% 14
MIN 19% 23 358 4.2 -14.4% 19 85 3.2 -3.8% 9 1.0 -10.5% 24
NYG 15% 24 312 4.2 -16.3% 21 56 2.8 -22.1% 24 1.4 5.8% 13
WAS 12% 25 301 3.8 -21.7% 27 40 2.8 -23.6% 25 1.0 1.9% 15
SEA 11% 26 265 3.4 -28.8% 32 34 2.2 -57.0% 31 1.2 28.1% 5
LAR 9% 27 355 4.6 0.8% 7 34 2.5 -33.7% 27 2.1 34.6% 2
MIA 8% 28 306 3.9 -25.4% 29 26 6.5 23.0% 1 -2.7 -48.3% 32
DET 6% 29 304 3.3 -24.7% 28 21 3.3 -15.6% 20 0.0 -9.1% 23
OAK 5% 30 321 4.1 -4.7% 12 16 6.9 14.6% 5 -2.7 -19.3% 29
PHI 2% 31 387 4.6 -9.2% 16 8 0.3 -113.8% 32 4.4 104.6% 1
ARI 1% 32 351 3.4 -28.1% 30 5 1.4 -9.1% 12 2.0 -18.9% 28
NFL 25% 8,819 4.1 -11.3% 2,954 3.8 -10.2% 0.4 3.67%

Volume Comparisons

While the overall percentage of two-back runs stayed about the same in 2017, we did reach a new milestone. For the first time in our records, only one team ran out of two-back sets more often than one-back sets: the San Francisco 49ers.

San Francisco was dead-last in two-back runs in 2016; their rate increased by 51 percentage points in 2017. Chip Kelly and Kyle Shanahan are essentially speaking two different languages. Short of bringing in Jurgen Klinsmann, you won't find two coaches who have deeper fundamental disagreement on what proper football should look like.

At the same time, 54 percent is laughably low by recent 49ers standards, as Jim Harbaugh and Greg Roman's teams were never below 58 percent, and that was in their worst season. Kelly's all-shotgun, all-spread, all-single back offense feels like a crazy one-year bender in San Francisco, a city that wants you to remember that fullbacks are people, too. Just take a look at their past four playcallers, with their years in San Francisco highlighted:

Frequency of 2-RB Runs by 49ers Playcallers
Season Roman Chryst Kelly Shanahan NFL
2011 69% -- -- 50% 47%
2012 74% -- -- 55% 44%
2013 78% -- 2% 50% 40%
2014 58% -- 0% 44% 33%
2015 46%
47% 2% 42% 29%
2016 57% -- 3% 43% 25%
2017 -- -- -- 54% 25%

It's interesting to see how Shanahan has gone from slightly above average to the highest outlier as the league has shifted to more and more one-back sets. He didn't get more extreme; he just stayed still while the rest of the league ran away from him.

Shanahan wasn't the only person to bring two-back philosophy into a new home. The 49ers may have been the only team with a majority of two-back runs, but the Ravens were this close, with 210 one-back rushing attempts versus 209 two-back attempts. Baltimore brought in ex-49ers and Bills coordinator Greg Roman to be their senior offensive assistant, and Roman's two-back proclivities followed, even if Marty Mornhinweg was the chief playcaller. The Ravens jumped from 24 percent to 50 percent last season, the third-largest jump in the league. Between the 49ers and Ravens were the Jaguars, who leapt from 5 percent to 34 percent in Nathaniel Hackett's first full season as coordinator.

From 2013 to 2016, four teams ran with two backs at least 40 percent of the time: Atlanta, Buffalo, Carolina, and Green Bay. None of the four hit that benchmark in 2017. Shanahan's departure knocked Atlanta out of that list. Carolina lost Mike Tolbert to Buffalo, and they fell out as well. Green Bay fell under the 40 percent threshold despite a healthy Aaron Ripkowski; we'll get back to them soon enough. Buffalo was just four carries away from hitting 40 percent, with the addition of Tolbert not enough to offset the loss of Roman. That means your current leader for most consecutive seasons with more than 40 percent two-back running is New England, who have now reached that mark a whopping two seasons in a row. Every streak has to start somewhere.

The Bills' drop-off of 17 percentage points was the third-largest in the league last season. Minnesota was second, dropping 19 points in Pat Shurmur's first full season as coordinator. In first place, though, were the Los Angeles Rams. Jeff Fisher's teams loved two-back running; they lead the league in both 2014 and 2015, and were the last team to top both 60 percent (in 2015) and 65 percent (in 2014, under new Seahawks offensive coordinator Brian Schottenheimer). They had cut back somewhat in 2016, but the arrival of Sean McVay hastened the end. Two seasons ago, the Rams had 221 carries out of two-back sets; that fell to just 34 last season. Their 21-percentage point drop led the league.

Of course, 34 is a nearly a whole game's worth of carries. That's a metric ton compared to Arizona and Philadelphia, each of whom finished with single-digit carries out of two-back sets. This is an important milestone, as they become the first non-Chip Kelly teams to have single-digit seasons. Arizona just missed out on the record—the 2014 Eagles had just two carries, both in one game—but still becomes the first team outside of Philadelphia to basically remove the two-back running game from its playbook entirely.

Philadelphia continues to hold on to their streak; they are the only team to have had less than 10 percent of their carries in two-back sets in every season stretching back to 2013. Yes, that means the Super Bowl was between the team which uses the most two-back rushes in recent history versus the team which uses the least. If that's not anecdotal evidence that neither strategy is inherently right or wrong, I don't know what is.

Also, the NFC West has the team with the most two-back carries and the team with the fewest two-back carries. They're also welcoming the last offensive coordinator to top 65 percent in Seattle, just after a new coach eradicated the last bits of that strategy from his old team in Los Angeles. The NFC West is weird, you guys.

Efficiency Comparison

One-back and two-back sets were roughly equally successful in running the ball. The average team had only a 3.7% spread in DVOA between one-back and two-back sets, which may be more attributed to which teams focused on which strategy, rather than an inherent bias towards one or the other being more effective. For what it's worth, however, two-back sets have now been the more effective rushing formation two years in a row after five straight years of one-back leads. That is technically a trend, even if two-back's DVOA actually dropped from last season. Perhaps this is an effect of lighter, faster defenses becoming more vulnerable to two-back sets. Defenses are being forced to deal with three wideouts in base packages, and are sacrificing some raw strength and size for the speed necessary to keep up. Or maybe it's just a small sample size fluke or random variation; the difference just isn't large enough to say anything for certain.

As pass plays are more efficient than run plays, it's more difficult for teams to have positive running DVOAs in any split. Congratulations are in order, then, for the Patriots, Ravens, Chiefs, and Saints, the only four teams to have positive rushing DVOA out of both one-back and two-back sets. The Patriots and Ravens both achieved that with more than 150 carries in both formations, making it all the more impressive—no small-sample size shenanigans here.

All four of those teams finished in the top seven in rushing DVOA, so their success in both formations isn't too unexpected. They were joined in that top seven by Dallas, Green Bay, and Pittsburgh, and the situations on those teams were slightly more interesting.

Both Green Bay and Dallas were in the top 12 in using two-back formations to run, and yet both were more effective when running out of single-back formations. The Packers are most notable here, ranking 18th in the league in two-back rushes but second in one-back sets. They also averaged more than a yard extra per carry out of one-back sets. Remember when we said that Green Bay fell out of the 40-percent two-back club earlier? This is a good thing; they have been better running out of one-back sets for two years in a row, and it may be finally sinking in that they don't need to continue to sacrifice plays to the ghost of John Kuhn.

Dallas had the opposite problem. They actually had a negative running DVOA out of one-back sets, but the third-highest DVOA when operating out of a full backfield. It should be noted that they only ran 89 two-back runs, so it's a smaller sample size, and that their -3.0% DVOA when running out of single-back sets was still eighth-best in the league. Still, that was a major differential for the second-best rushing offense in the league. In general, Dallas was better when they went with more compact formations and avoided going three-wide; they were one of only eight teams to have a lower DVOA in 11 personnel than in other groups. The Cowboys' best formations had Jason Witten as an inline tight end and James Hanna as an H-back, helping open holes for Ezekiel Elliott, and this was true both in the running and passing games. Normally, this is where I would say the Cowboys should use more two-tight end sets in 2018, but the retirement of Witten and the reshuffling of Dallas' receiver corps makes that an iffier proposition.

Most of the other major differences between one- and two-back sets come from small sample sizes. The Eagles and Rams were did significantly worse while running out of two-back sets, but they were both running out of single-back 90 percent of the time anyway, so there's not much room for better strategy there. Miami, Oakland, and Arizona were all significantly better running out of two-back sets, but with a combined 47 attempts, that's mostly a small-sample size issue.

Four of the 12 teams with at least 100 carries in both formations were significantly better in one formation as opposed to the other. The Chargers, Broncos, and Bears were all significantly better running out of one-back sets, each going from the top half of the league to the bottom seven when shifting from one to two backs. Los Angeles and Chicago were also both in the top dozen teams in two-back usage despite that differential; teams like this really blow up any correlation between success and usage rate.

The Panthers were the one team to see at least a 25.0% increase when running with two backs in the backfield, rising from 31st in the league to eighth in their heavier sets. This is new—they were significantly worse in two-back formations in 2015, and close to even in 2016. You would think that the loss of Mike Tolbert would have hurt the Panthers, but quite the opposite! Perhaps Ed Dickson lining up as an H-back is less of an obvious run tell than bringing in a fullback; the Panthers threw the ball roughly 43 percent of the time with Dickson as an H-back in 2017, compared to 26 percent with Tolbert as the lone fullback in 2016.

In our next column, we'll take a look at this from the defensive side of the ball to see which teams struggled to handle extra blockers and which teams begged you to clog up the middle of the field with fullbacks and H-backs.

Posted by: Bryan Knowles on 11 Jul 2018

7 comments, Last at 18 Jul 2018, 3:27pm by CaffeineMan


by The Ninjalectual :: Thu, 07/12/2018 - 9:42pm

Well-written piece!

With the Panthers, did they not run from a Stewart/McCaffery backfield often?

by Bryan Knowles :: Fri, 07/13/2018 - 6:16am

Not as often as you'd expect.

I don't have the exact numbers here -- formation data and personnel data are two different things and, annoyingly, in two different formats -- but it looks like about 52 percent of Carolina's two-back runs came with two actual running backs, and 48 percent came with a running back and a tight end in the backfield.

McCaffrey had some weird splits in 2017 as Carolina really struggled to use him properly -- or, rather, there were some things they did that really worked and some things they did that really DID NOT WORK, and they had trouble distinguishing between the two. We go into that a lot in FOA 2018 (cheap plug!)

by Bright Blue Shorts :: Mon, 07/16/2018 - 4:53am

I very much liked "Short of bringing in Jurgen Klinsmann, you won't find two coaches who have deeper fundamental disagreement on what proper football should look like."

by CaffeineMan :: Mon, 07/16/2018 - 4:04pm

Help a very casual soccer fan out? I knew Klinsmann was head coach of the USMNT and he was fired, but that's it.

by Bright Blue Shorts :: Mon, 07/16/2018 - 6:19pm

"Chip Kelly and Kyle Shanahan are essentially speaking two different languages. Short of bringing in Jurgen Klinsmann, you won't find two coaches who have deeper fundamental disagreement on what proper football should look like."

If you already know Klinsmann is a soccer coach then I can't explain it any better.

by Bryan Knowles :: Mon, 07/16/2018 - 7:30pm

The joke is that soccer is also called football, so what a soccer coach and an NFL coach would think 'proper football' is would be very different.

by CaffeineMan :: Wed, 07/18/2018 - 3:27pm

Ah, I was overlooking the obvious. I was assuming it was a shot at Klinsmann's performance as head coach of the USMNT.