Writers of Pro Football Prospectus 2008

Most Recent FO Features

McKinneyBen15.jpg

» 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.

19 Jun 2018

2017 Offensive Personnel Analysis

by Bryan Knowles

Every year, going back to 2010, Football Outsiders has published some version of this offensive personnel analysis. Every year, we begin these articles in roughly the same way: talking about the dramatic rise of three-wideout sets.

For six consecutive years, we have documented the rise of the 11 personnel grouping (one running back, one tight end, and three wide receivers). From being used just under 40 percent of the time in 2010 to topping 60 percent last year, it has seen its popularity and success only grow. Every single year, it has been used slightly more often than the year before. Every year since 2012, it has been more effective than any other common formation. And every year, there's an argument to be made that teams aren't using it enough; that if it's producing a better DVOA than any other formation, teams should be using it more and more. There seemed to be no end in sight for 11's dominance.

You see what we're setting up here, right?

For the first time since we've started recording personnel groupings, 11 personnel usage dropped last season, falling from 60.4 percent of plays to 59.3 percent. DVOA for 11 personnel groupings also fell, from 7.5% to 7.0%. Neither of these dropoffs are overly dramatic; 2017 still saw the second-highest percentage of 11 personnel we've ever recorded. It's just that whenever you see a half-decade trend reversing itself, that's worth stopping to take note.

It's also worth noting that this wasn't a result of one or two teams deciding to buck the trend and single-handedly knocking a percentage point off the usage rate. For the second straight year, every team in the NFL used 11 personnel on at least 40 percent of their snaps; it's not one or two outliers tanking the group down. Fifteen teams saw their 11 personnel usage drop from 2016 to 2017; nine of them saw it drop by at least five percentage points, and three saw it fall off by at least 20 points. While not exactly a league-wide phenomenon, this also wasn't just a case of one or two teams running out of receivers or suddenly changing philosophy.

Before we get into things any further, a quick note: this is personnel data, not formation data. When Christian McCaffrey goes out wide, he's still counted as a running back. When Randall Cobb lines up in the backfield, he still counts as a wide receiver. We're using the standard numerical system where the first digit is the number of backs, and the second digit is the number of tight ends. We count formations with six or more offensive linemen separately, rather than counting the offensive linemen as tight ends. A formation with two backs, one tight end, one wideout, and six offensive linemen is marked as "621" and not "22."


Offensive Personnel Groupings
Personnel 2016 Pct 2017 Pct Difference
2017 DVOA
11 60.4% 59.3% -1.1%
7.0%
12 16.9% 18.8% +1.9% 2.9%
21 8.0% 7.7% +0.7% 5.4%
13 2.8% 4.0% +1.2% -15.8%
6+ OL 5.4% 3.5% -1.9% -3.8%
22 1.8% 3.1% +1.3% -5.5%
10 2.6% 1.2% -1.4% 8.1%
Misc 3.1% 2.8% +0.3% -6.0%

11 remains the predominant personnel grouping, even with the slight downturn in 2017.

Once again, every team used 11 personnel as their primary offensive setup. Twenty-four teams used it more than half the time, which is down a bit from last season, but still fairly prevalent.

The real question is what this slight decline actually means. Is it a one-year blip, caused by specific 2017 circumstances, and will the ascension start anew in 2018? Is it a sign that we've reached a level of equilibrium, and that using 11 personnel on more than 60 percent of snaps has significant diminishing returns? Or is it the beginning of an actual decline, with two- and three-tight end sets becoming more and more popular?

Obviously, we won't be able to answer that question until we see what happens next season, but looking at the teams that saw their 11 usage drop the most will give us a little bit of context to try to answer that question.

Some of it can be explained with injuries. The Giants were, by a wide margin, the team that saw the biggest dropoff. They went from a league-record 92 percent of 11 personnel plays in 2016 all the way down to 61 percent last season. Now, this is just a hunch, but I suspect that three-wideout sets look a lot more enticing with Odell Beckham, Sterling Shepard, Brandon Marshall, and Dwayne Harris are all healthy. When you're down to your fifth, sixth, and seventh receivers, you may want to consider seeing what your tight ends can do, especially when one was a first-round pick.

Dropoffs on other teams can be explained by coaching changes. Comparing Chip Kelly's 49ers to Kyle Shanahan's is like watching two different sports. The Kelly 49ers came out in 11 personnel 76 percent of the time; Shanahan's version used it just 48 percent of the time, as he came from Atlanta with weird, byzantine concepts like "having a fullback." Offensive coordinator changes in Jacksonville and Baltimore may also go a way to explaining their near-20-point dropoffs; it was the first full seasons for Nathaniel Hackett and Marty Mornhinweg, and their offensive identities are quite different from those of their predecessors.

The other team to see a big dropoff was the Bears, falling from 65 percent to 46 percent. Cam Meredith tore his ACL in August. Coupled that with a desire to protect a rookie quarterback, that meant using extra tight ends for protection made a lot of sense on paper. There's something else going on there, though, that may actually begin to explain some of this -- or, at least, highlight a growing trend.

Tarik Cohen was targeted 35 times out wide or in the slot. He still counts as a running back, so he allowed the Bears to have a three-wide look without actually using three wideouts. Alvin Kamara in New Orleans and Christian McCaffrey in Carolina offer the same sort of versatility, regularly splitting out wide for teams that used 11 personnel less than half the time. The 49ers, Patriots, and Ravens also all used 11 personnel less than half the time; they had the second-, third-, and fourth-most running back pass targets. When you have a running back who can catch the ball, be it split out wide or not, there's less need to actually get extra receivers out onto the field.

In 2016, the Browns led the league with 25 percent of their targets going to running backs. Last year, nine different teams passed that mark, including six of the eight teams to use 11 personnel the least. The exceptions were Tennessee, who got major receiving contributions out of their tight ends, and Atlanta, who seemed to avoid 11 personnel out of residual Shanahanniness more than anything else. Some of this is correlation versus causation -- do teams throw to running backs more because they're not in 11 personnel, or do teams avoid 11 personnel because they can throw to running backs more? -- but as backs who can split out wide rise in popularity, we may see 11 personnel begin to recede more and more.

You would expect, in a rational and sane world, that the teams that did the best when using 11 personnel -- or, at least, the best compared to the rest of their offense -- would be the teams that used it the most. The NFL is neither a rational nor sane world.


DVOA in 11 Personnel
Team 11 Pct Rk Yds 11 DVOA Rk Non-11 DVOA Rk Diff Rk
LAR 81.4% 1 6.3 25.5% 6 -21.2% 30 46.7% 1
MIA 75.4% 2 5.2 -11.3% 28 0.2% 14 -11.5% 28
DET 75.0% 3 5.8 10.3% 14 -3.3% 19 13.6% 14
OAK 73.1% 4 5.9 13.2% 12 0.6% 12 12.6% 15
CIN 70.8% 5 5.2 -1.5% 22 -8.2% 21 6.7% 20
PIT 70.1% 6 6.4 28.6% 5 3.4% 11 25.2% 5
PHI 66.6% 7 6.1 19.7% 7 5.8% 10 13.9% 13
SEA 66.3% 8 5.8 18.0% 9 -11.0% 24 29.0% 3
DEN 66.1% 9 5.1 -9.2% 26 -24.8% 32 15.6% 12
WAS 65.6% 10 5.4 1.0% 20 0.3% 13 0.7% 24
IND 64.8% 11 4.9 -9.8% 27 -19.8% 29 10.0% 16
GB 64.5% 12 5.3 11.2% 13 -10.3% 23 21.4% 6
CLE 63.7% 13 5.2 -19.8% 31 -13.6% 25 -6.2% 27
TB 62.1% 14 6.2 19.5% 8 -1.8% 18 21.3% 7
NYJ 62.1% 15 5.5 3.9% 18 -21.7% 31 25.6% 4
NYG 61.5% 16 5.2 -2.8% 23 -9.5% 22 6.7% 19
Team 11 Pct Rk Yds 11 DVOA Rk Non-11 DVOA Rk Diff Rk
DAL 61.3% 17 5.5 6.5% 16 11.1% 5 -4.6% 26
LAC 56.8% 18 6.7 30.9% 3 -1.3% 17 32.2% 2
MIN 56.3% 19 6.1 16.8% 10 13.2% 4 3.6% 23
BUF 55.5% 20 5.3 0.0% 21 -17.9% 28 17.9% 10
JAC 53.8% 21 5.7 -4.4% 24 9.5% 7 -13.9% 29
KC 52.4% 22 6.5 31.7% 2 10.6% 6 21.1% 8
ARI 52.2% 23 4.8 -21.5% 32 -7.0% 20 -14.4% 30
HOU 52.0% 24 6.1 2.4% 19 -13.7% 26 16.2% 11
CAR 49.8% 25 5.6 5.1% 17 -0.5% 16 5.6% 21
ATL 49.7% 26 6.1 14.9% 11 9.4% 8 5.6% 22
NO 49.1% 27 6.7 34.3% 1 14.2% 3 20.2% 9
SF 47.9% 28 5.4 6.9% 15 -0.2% 15 7.2% 18
CHI 46.1% 29 5.3 -6.1% 25 -13.9% 27 7.9% 17
TEN 45.1% 30 4.6 -19.6% 30 18.2% 2 -37.8% 32
NE 43.9% 31 6.4 28.8% 4 31.8% 1 -3.0% 25
BAL 39.5% 32 4.2 -16.9% 29 6.5% 9 -23.5% 31

In a world in which having three receivers on the field was banned, the Titans would be an offensive powerhouse. They go from 18th in offensive DVOA to second if you ignore all 11 personnel situations, and their DVOA only gets higher as you ignore four- and five-wide sets. As Rishard Matthews is their only real proven wideout, Delanie Walker is one of the best pass-catching tight ends in the league and Jonnu Smith has intriguing upside, expect that to continue going forward. The Titans wisely were 30th in 11 personnel usage, and honestly could afford to drop a little bit more going forward.

The Rams, meanwhile, should be pushing the NFL to outlaw two-tight end sets. No team in football used 11 personnel more than the Rams, and no team saw a bigger boost to their fortune when using it than Los Angeles. The 81 percent of the time they trotted out three wideouts is nothing, in the grand scheme of things -- the Giants of 2014 and 2015 showed you could push 90 percent. Come on, Sean McVay -- here's one more offensive record you can bring down!

Just below the Rams in 11 personnel usage was Miami, and that should not have been the case. The Dolphins were much, much more effective when they brought in a second tight end rather than using three receivers. The goal was to get Jarvis Landry, DeVante Parker, and Kenny Stills on the field at the same time as often as possible, with all three playing about 90 percent of snaps when healthy. In fact, no team in football had more passes out of 11 personnel than the Dolphins did. And yet, their passing DVOA in 11 personnel was -6.5%; in all other personnel groupings, it was 31.6%.

One team that should be looking to use 11 personnel more in 2018 is the Chargers, who had the third-highest DVOA out of 11 personnel but used it just 56.8 percent of the time. With Hunter Henry out for the season, both Mike Williams and Tyrell Williams should find it easier to see the field next to Keenan Allen.

The increase of 11 formation has meant something of a homogenization of personnel groupings around the league, but there are still some teams out there doing different things. Here are some of the remaining bits and bobs from the personnel data:

  • Baltimore led the NFL in both 12 and 22 personnel usage. All in all, they had 495 snaps with two tight ends on the field; Minnesota and Kansas City were the only other teams even over 300. Throw in the odd three-tight end set near the goal line, and the Ravens ended up with multiple tight ends on the field 55 percent of the time; no one else even hit 40 percent. And now, they've added Hayden Hurst and Mark Andrews in the draft to bolster the position that much more. We know the Ravens have talked about experimenting with a two-quarterback set with Joe Flacco and Lamar Jackson on the field together; what we didn't know is if it'll be Flacco, Jackson, and four tight ends. The ultimate big formation!
  • We are required by law to have one Jimmy Garoppolo tidbit in all of these offseason stat analysis pages. For the third straight year, Kyle Shanahan's teams led the league in 21 personnel as he brought his system to the 49ers. The 49ers ran 370 plays with two running backs on the field last season. Under Chip Kelly, that number was 15, more than half of them in one week due to injuries. The 49ers saw their 21 personnel jump from 26 percent when Brian Hoyer or C.J. Beathard was under center to 42 percent with Garoppolo. This is almost entirely because the 49ers were actually competitive when they had Garoppolo under center; the 49ers ran 21 personnel in 35 percent of their snaps in one-score games regardless of who their quarterback was.
  • Arizona decided to skip 11 personnel and jump straight to 10. They went four-wide on 14 percent of their snaps; no one else even hit three percent. The loss of David Johnson forced them to scramble around early in the season with uncharacteristic formations; their use of 10 personnel dropped off in the second half of the season. It's actually too bad they didn't keep it up -- it was their most successful formation in 2017, with a 12.3% DVOA.
  • All 32 teams had at least 400 snaps in 11 personnel. All teams had at least 100 snaps in 12 personnel. These are the two universal groupings; everyone has some form or another of them in their playbook. They are technically joined by two other formations, though those are not precisely a vital part of every team's strategy. Everyone ran a version of 21 personnel, though seven teams -- Arizona, Indianapolis, the Rams, Miami, Oakland, Philadelphia, and Washington -- ran fewer than five such snaps all season long. Everyone ran 13 as well, though Pittsburgh and Houston each had fewer than five snaps of their own.
  • The first "missing" formation is 22. While it was used on 3.1 percent of snaps in 2017, five teams -- Houston, Jacksonville, the Rams, Philadelphia and Washington -- didn't use it at all. When we first started keeping track of personnel data in 2010, 22 was one of the Big Four formations; a clear fourth place but still used 7.2 percent of the time. Now, it isn't even in some team's playbooks.

Posted by: Bryan Knowles on 19 Jun 2018

11 comments, Last at 28 Jun 2018, 6:57am by Bright Blue Shorts

Comments

1
by Bryan Knowles :: Tue, 06/19/2018 - 2:24pm

Some more stuff from the data that wasn't good enough to go into the article!

*Ignoring sample size limits, the worst formation in football in 2017 was 32. Denver ran exactly two plays out of 32, back-to-back, against Oakland on the 1-yard line. The first saw Devontae Booker plow into the line for no gain. They second saw Paxton Lynch's pass bounce off three people, ending up in the hands of Navorro Bowman for Oakland's first interception of the season. That's a DVOA of -350.1%. Burn those plays!

*Sliding up the "minimum plays required" list, the worst personnel grouping was 03 (9 plays, -150.7% DVOA); Kansas City tried that a handful of times and it worked exactly once, when Alex Smith was forced to improvise and scramble for his life.

*Then it's 00 (23 plays, -60.0% DVOA); this was usually Green Bay with Randall Cobb lined up in the backfield rather than a true five-wide set.

*After that it's 610 (54 plays, -28.2%), half of which belong to Houston. Weep for that offensive line.

*Capping us off was 20 (281 plays, -20.5% DVOA), which was generally just a four-wide formation with someone like Alvin Kamara or Tevin Coleman serving as a receiver.

*The best formation in football, once again ignoring sample size limits, was 14. This is exactly one play – Doug Martin's one-yard plunge against New England back in Week 5. This was a true four-tight end play; no one split out wide or in the backfield as a fullback or anything, just two tight ends in either side of the line. It was the only four-tight end play of the year. It was worth 124.0% DVOA, and then was promptly never used again.

*Going up the minimum plays required list, the next best personnel grouping was 23 (95 plays, 16.2% DVOA). This was almost exclusively used tight-end happy teams (Baltimore, New England, Minnesota) on the goal line or in similar power situations, but San Francisco actually pulled of the rare 23 empty backfield formation in Week 15.

Then you have 622 (135 plays, 13.2% DVOA). Having six linemen and no receivers screams "run", which may be why nearly a quarter of 622 plays were passes. Misdirection!

Finishing our tour of the best personnel groupings no one bothered to use is the 01 (240 plays, 10.7% DVOA). The Rams and Packers both could be counted on to pull this one out two or three times a game. The Rams version used Tavon Austin as a running back, while the Packers generally actually went empty.

2
by justanothersteve :: Tue, 06/19/2018 - 3:15pm

Nice supplementary info. Love the data showing the Packers need to burn those Randall Cobb in the backfield plays. It was nice as a gimmick for a couple games. Everyone else has figured out how to defend it but unfortunately that hasn't registered in McCarthy's imagination-starved brain.

3
by Aaron Brooks Go... :: Tue, 06/19/2018 - 4:05pm

Is an empty 23 a set where the RBs are really FBs and are lined up like TEs?

5
by Bryan Knowles :: Tue, 06/19/2018 - 4:34pm

There's a couple ways to do it.

When the 49ers ran it, they had two tight ends lined up on either side, a third tight end (Kittle) in the slot, and the Breida and Hyde split out wide; they were in a hurry-up situation and didn't bother to scramble any receivers onto the field for their next snap.

The Vikings also ran had one empty back 23 set, but it was on the goal line. They lined up in a pretty standard heavy I formation, and then shifted to split a tight end and a running back out wide (Rudolph and McKinnon), leaving two tight ends and a fullback (Ham) lined in tight. The pre-snap motion (plus a fake end around) baffled Washington, and Morgan had an easy uncovered touchdown in the flat.

https://twitter.com/NicholasJOlson/status/985927359667765249

4
by Aaron Brooks Go... :: Tue, 06/19/2018 - 4:06pm

The success of the 10 reminds me of the Davis/Levy Lions, who in 1990, didn't even have a TE or FB on their roster.
https://www.pro-football-reference.com/teams/det/1990.htm

Essentially every play was either a 10 or a 00.

9
by The Ninjalectual :: Wed, 06/27/2018 - 8:03pm

That's crazy roster management! Madden wouldn't even let you do this

10
by ssereb :: Wed, 06/27/2018 - 9:37pm

They didn't even have a real backup to Barry Sanders, since it looks like James Wilder got 12 touches all season. Somehow this offense scored the fifth-most points in the NFL that year.

11
by Bright Blue Shorts :: Thu, 06/28/2018 - 6:57am

The problem is they conceded the 3rd most! Minus-40 differential.

The 1-15 Patriots conceded the most (446 vs 181 scored). Distant memories of when their franchise was truly awful.

6
by JimZipCode :: Wed, 06/20/2018 - 10:36am

In the Ravens case, the decline in 3WR sets may not all have been strategic intention. After Maclin got hurt and Perriman started tipping passes to defenders, the Ravens struggled to put 3 good WRs on the field. (Hell, they struggled to put ONE good one on the field.) This offseason's dramatic pass-catcher purge in Baltimore is unambiguous evidence of how bad they think the problem was.

(Their TE corps wasn't all that great either.)

I'd expect to see a lot more 11 personnel from the Ravens this year. Their top offensive personnel grouping is clearly:

Alex Collins or Kenneth Dixon
Hayden Hurst
Crabtree - John Brown - Snead

If those guys stay healthy, Baltimore should align more closely with league norms. The wildcard might be Greg Roman's exotic run packages — we might see more 622 here than from most teams.

7
by jtr :: Wed, 06/20/2018 - 11:31am

Greg Roman has also historically really liked fullbacks. His 49ers offenses were regularly waaay last in 11 personnel usage, largely because he loved fullback/H-back Bruce Miller so much. I would be surprised to see one of his teams in the top half of the league in 11 personnel.

On a sidenote, the 49ers had a weird whiplash back and forth in personnel preference. Under Harbaugh/Roman, they were one of the only teams to still run a 2-back offense. Then Chip Kelly of course almost completely purged the fullback from the offense. Now under Shannahan, they're once again the last bastions of I-formation football.

8
by ssereb :: Wed, 06/20/2018 - 5:22pm

Patrick Ricard played 13.8% of the Ravens' offensive snaps last season (he also played a little bit of nose tackle). Given that that was with a decimated offensive line, I think that number might go down some, though with Roman's tendencies, it probably won't hit zero.