Stat Analysis
Advanced analytics on player and team performance

Defensive Personnel Analysis 2019

Seattle Seahawks LB Mychal Kendricks
Photo: USA Today Sports Images

Earlier this week, we looked at which personnel groupings were most popular on offense in 2019. Now, it's time to flip that around and look at how defenses have responded.

It's not new that the term "base" defense no longer accurately describes modern defensive design; the days of teams trotting out seven men in the box down after down after down are, at the moment at least, as anachronistic as a team coming out in the old seven-diamond would be. Base defense did make a slight resurgence this past season, but it's still used on under 30% of all defensive snaps; it has become a situational front used against certain matchups, rather than the foundation of defensive strategy.

This, of course, is a reaction to 11 personnel taking over the league; it's simply hard to match up against three wide receivers with just four defensive backs. The slight resurgence of two-back or two-tight end sets made base slightly more palatable last season, but even then, that second tight end or running back is more likely to be a passing threat now than they were in the 1980s, so it's frequently easier just to leave that nickel corner out there and call it a day. In fact, teams saw more use of dime defense in 2019, as getting more and more defensive backs onto the field remains the agreed-upon strategy to handle modern NFL offenses.

Defensive Personnel Groupings, 2019
Personnel 2018 Pct 2019 Pct Difference 2019 DVOA
Nickel 60.5% 55.9% -4.6% 1.5%
Base 25.0% 26.7% +1.7% -3.9%
Dime+ 13.3% 16.0% +2.7% 3.0%
Goal line 0.8% 0.8% +0.0% -1.9%
Big 0.3% 0.6% +0.3% -19.5%

A couple quick notes before we continue:

  • For the record, 54% of base snaps were in 3-4 and 44% were in 4-3; that's almost an exact mirror of their totals from 2018. That doesn't add up to 100%; there were 133 snaps of 2-5 (most frequently from the Jets), 65 snaps of 5-2 (often from San Francisco or Tampa Bay), six snaps of 1-6, and even two of 0-7 (Chicago had a real "how do I hold all these linebackers?" problem at times). However, as we note in the book, it can be tough to differentiate a defensive end from a linebacker in modern defenses.
  • While we group 4-3 and 3-4 defenses together at this point, the majority of teams do still use one or the other more or less exclusively. 13 teams were primarily 3-4 units, including, surprisingly, New England -- they had used a fairly even mix of 3-4 and 4-3 in 2018, but ran just three snaps of 4-3 last season against 133 snaps of 3-4. That might be the Jamie Collins effect there, so we'll see what happens in 2020 now that Collins is in Detroit. Nine teams were primarily 4-3 units, led by Indianapolis, although the Colts did see an increase in their 3-4 snaps -- specifically, going from zero to three. That leaves ten teams which saw neither front hit 75% of their base snaps -- the Falcons, Panthers, Bears, Jaguars, Dolphins, Jets, Buccaneers, Lions, Chiefs, and Saints. That's an increase from 2019; we'll stick a pin in that and see if that's a trend rather than random noise.
  • "Dime+" includes any package with more than five defensive backs. That includes all your dime packages, as well as the 299 snaps in quarter, 22 snaps with eight defensive backs on the field (… half-dollar?), and even four snaps with nine defensive backs (… the Sacagawea package?). If you're curious as to what a play with nine defensive backs looks like in practice, it's the pinnacle of optimum play:
  • "Big" defenses are 4-4 or 3-5 lineups, while "goal line" includes all other personnel groups with fewer than four defensive backs. More than half of those defenses were used on the 1-yard line, but that wasn't a literal necessity; the Lions even busted out a "goal line" look on their opponent's 13-yard line as they sold out to stop the run.

Last season, defenses did have a better DVOA in base than they did in any of these new-fangled multi-defensive back formations, and that's fairly typical. That does not, however, mean that base defense is better and every defensive coordinator in the league is getting things wrong. It's a matter of matchups more than anything else. Base defense continues to vanish against 11 personnel, the most effective offensive personnel group. There were about 1,500 snaps of base defenses against 11 personnel in 2017, 1,000 in 2018, and just 660 in 2019, mostly due to one team (we'll get to that). The median team played just six snaps of base defense against three-wide sets.

Teams do actually have a better DVOA against 11 personnel in base than in nickel or dime, but that's almost entirely due to the offense's run/pass ratios; defenses use base personnel when they think the opposing offense is going to run. So while nickel and dime+ defenses ended up facing pass attempts on 72% of their snaps against 11 personnel, base defenses faced passes only 59% of the time -- and those passes were, on average, a full yard shorter than the ones faced by nickel packages. That right there explains most of base's relative success against 11 personnel; they faced easier situations. Much of the rest of it can be explained by selection bias; if your team can't match up base against 11, you're simply going to run nickel defense -- or you're going to get fired.

Defensive Personnel DVOA Breakdown, 2019
Defense Base Nickel Dime+
Split % Used DVOA % Used DVOA % Used DVOA
11 personnel 3.5% -0.9% 73.7% 2.3% 22.7% 2.3%
12 personnel 57.4% -5.1% 37.7% -2.2% 4.4% 12.6%
21 personnel 68.6% -3.5% 27.1% 4.4% 3.7% 0.4%
All pass 18.0% 10.7% 59.1% 7.6% 22.3% 3.3%
All run 41.1% -13.8% 50.6% -8.7% 5.7% 1.7%
Overall 26.7% -3.9% 55.9% 1.5% 16.0% 3.0%

Now that we've talked about the league-wide trends, we can talk about team-specific oddities. That requires the annual big honkin' table, which we have here.

Defensive Personnel Frequency, 2019
Team Base Rk Nickel Rk Dime+ Rk
ARI 38% 2 57% 17 2% 26
ATL 24% 20 71% 6 2% 28
BAL 9% 32 46% 26 41% 3
BUF 22% 21 77% 3 0% 32
CAR 32% 6 66% 9 1% 30
CHI 31% 9 62% 14 6% 20
CIN 29% 13 49% 22 20% 10
CLE 12% 31 85% 1 2% 25
DAL 32% 7 63% 13 4% 22
DEN 27% 18 65% 10 8% 18
DET 19% 27 46% 25 34% 5
GB 19% 25 28% 30 51% 1
HOU 29% 14 51% 19 19% 13
IND 33% 5 47% 24 20% 11
JAX 30% 10 60% 15 9% 17
KC 27% 17 38% 28 34% 6
LAC 29% 11 51% 20 19% 12
LAR 34% 3 23% 32 42% 2
MIA 32% 8 36% 29 31% 7
MIN 26% 19 72% 5 1% 31
NE 15% 30 42% 27 41% 4
NO 16% 29 64% 11 19% 14
NYG 18% 28 63% 12 18% 16
NYJ 19% 24 77% 2 3% 24
OAK 19% 26 74% 4 6% 21
PHI 20% 22 48% 23 26% 8
PIT 29% 15 51% 21 18% 15
SEA 69% 1 27% 31 3% 23
SF 28% 16 70% 7 1% 29
TB 29% 12 66% 8 2% 27
TEN 20% 23 54% 18 26% 9
WAS 34% 4 59% 16 6% 19
AVG 27% -- 56% -- 16% --

Alright, I said we'd come back to the one team who stubbornly played base against 11 personnel, so hi, Seattle! The Seahawks had 217 snaps in base against 11 personnel; no other team had more than 60. Seattle played 69% of their snaps in base formation, which is mind-boggling. Two years ago, we mocked Cleveland relentlessly for going in base 66% of the time in their 0-16 season; they were the only team since 2014 to use base over 60% of the time. Now they have company; a defense rushing out a front seven at a rate we have not seen since the three-receiver revolution took over the NFL. When we talked about the Browns, we laughed and moved on; they were terrible, so of course their terrible defensive strategy was also terrible. But the Seahawks were a playoff team last season; they were an inch away from winning the NFC West. As such, this one deserves a bit of a closer look.

First and foremost, Seattle was correct in shying away from their nickel package. They had a -3.4% DVOA in base personnel, compared to a 30.4% DVOA in nickel. That puts them squarely in the middle of the pack in base DVOA but dead last with five defensive backs on the field. It is rare for a defense to be that much worse in one particular personnel group than another. Take Miami, for instance -- they had the second-worst DVOA in nickel at 27.4%, but their overall DVOA was 21.8%; no real change there. The Seahawks had a 2.6% overall defensive DVOA, and yet were Dolphins-esque with that fifth defensive back out there. Opposing offenses had a 41.4% DVOA against Seattle when they were in nickel -- that's adjusted for how good the Seattle defense was overall, and the Seahawks were giving up a full yard per play more than they were in base. No other defense allowed their opponents to have an offensive DVOA of over 15.7% when they were in nickel defense. For the Seahawks, switching from base to nickel was the equivalent of swapping the opposing offense from last year's Colts for the 2010 Patriots. More like the Legion of Bust, am I right?

For comparison, the 2017 Browns went from a -0.2% DVOA in base to a 12.3% DVOA in nickel -- they, too, were better in base, but not to the extent that Seattle was. And the Browns were significantly worse against the pass in base defenses than they were in nickel, meaning that they weren't doing a good job of matching personnel to situation. Against the pass, the Seahawks had a 0.3% DVOA in base and a 31.2% DVOA in nickel. Their extremely high level of base defense isn't because of poor game planning or adherence to a terrible philosophy; it's because their nickel package was terrible. They only went to it when they were absolutely sure that they were facing a pass. Their nickel sets faced only 19 runs all season long. It should be noted that the nickel package only marginally improved to 24.9% when Quandre Diggs arrived midseason. The issue was that their planned third corner, Jamar Taylor, was a disaster; the Seahawks were just better off keeping K.J. Wright and Mychael Kendricks on the field. Pete Carroll has always liked base more than most of his contemporaries, but never to this extent, and never with this big of a split in outcomes. This was a change caused by a specific personnel situation, and should be temporary. After all, the Seahawks brought in Quinton Dunbar to lock up that slot corner spot and … oh. Well, maybe someone will rise to the occasion this season.

Seattle's not the only team which had to do some strategic reshuffling due to personnel issues. The Chargers went from playing dime+ 64% of the time in 2018 to just 19% last season, due in part to Derwin James' injury and also to actually having healthy linebackers for once. Other teams saw their defensive philosophies shift wildly thanks to the rotating chairs of defensive coordinators last season. The 2018 Cardinals led the league in nickel usage at 84%; Steve Wilks brought that with him to Cleveland which went from 66% to 85%, most in the league. Wilks was replaced in Arizona by Vance Joseph; his 2018 Broncos had led the league in base usage at 45% and so Arizona saw its base defense jump from a dead-last 10% in 2018 to 38% in 2019, first in the non-Seahawks division. And with Joseph gone, the Broncos brought in Vic Fangio -- Fangio's Bears had a 76% nickel usage in 2018, and Denver's nickel usage jumped from 27% to 65% last season. It makes you long for the stability in Buffalo, which went from 21%/77%/1% in 2018 to 22%/77%/0% in 2019; they were the only team in 2019 to run zero plays with more than five defensive backs on the field.

Most teams' frequencies went nickel-base-dime; that was the order for 22 teams last season. You can split them into subgroups, of course. Teams such as the Bills, Browns, and Jets basically lived in nickel and only brought in an extra front seven player in absolutely sure running situations, while teams such as the Dolphins, Colts, and Bengals flipped from nickel to base to dime all willy-nilly. Of these 22 teams, the Browns are probably the most interesting; even in the modern NFL, sticking in nickel 85% of the time really jumps out at you. Was it effective? Yes. Ish. The Browns ranked 17th in nickel DVOA at 1.6%; in all other personnel groupings, they were dead-last at 29.0%. I'm not sure the "we're No. 17!" foam fingers are going to sell particularly well, mind you, and the Browns had been fifth in non-nickel DVOA in 2018; you could make the argument that Wilks fixed the nickel defense at the expense of everything else.

The second-most common pattern was nickel-dime-base, with teams pretty much eschewing an extra linebacker for a third safety most of the time. Some of this can be directly traced back to Bill Belichick: you have his Patriots, defenses led by his former assistants Matt Patricia (Lions) and Jim Schwartz (Eagles), and a defense led by his former player Mike Vrabel (Titans). Ex-Patriots coach Brian Flores' Dolphins nearly joined the group as well and probably will in 2020. The Ravens, Chiefs, and Saints round out this crowd, giving us six defenses that finished in the upper half of the league and then also the Lions. Some of the heavy defensive back usage may be based on the fact that a lot of these teams were winning a lot, so their opponents were passing a lot, so more defensive backs were needed, but this wasn't just about teams bleeding out leads with extra backs. The Patriots and Ravens, especially, made three-safety packages a huge part of their look and it paid massive dividends. It makes sense -- you don't want a third or fourth linebacker on the field because it's no longer the 1980s, but you don't want to lose that physicality in the box entirely, so you stick an extra safety out there to improve your pass defense without harming your run defense entirely. Maybe in ten years' time, we'll be listing third corners and third safeties as starters and pining for the good ol' days where you might see a front six.

The Packers took this philosophy to the extreme last season, as they were the only team to go dime-nickel-base. They only ran 201 plays without an extra defensive back, and their 510 snaps of dime+ were 77 more than the second-place finisher. Without a true three-down linebacker to play next to Blake Martinez, the Packers saw their dime+ percentage rise from 41% in 2018 to 51% last year; coupled with the Chargers' drop-off, that gives them the dime-usage crown. I'm not sure they would have passed the 50% mark if they had had more options at linebacker, but Mike Pettine does like the versatility in coverage you get from three safeties. You do wonder if that over-reliance on dime schemes ended up hurting the Packers in the playoffs when San Francisco ran over them, though I'd caution against using one bad game as a massively significant datapoint. After all, the Packers were better against the run then the Seahawks, who we already established lived in base personnel.

That covers 31 teams, leaving just the Rams, the only dime-base-nickel team in the league last season. In 2018, the Chargers were the only DBN team in the league, so the title has just moved across town. But the Chargers became more and more of a dime-first defense as the year went on in 2018, as Derwin James blossomed and the linebackers joined the IR. The Rams went on the opposite journey, running 50 snaps of nickel in the first half of the year and 193 the rest of the way -- or, to put it another way, going from 32nd to 24th in nickel frequency. That means they did a complete flipflop, ending the year as a nickel-base-dime team, albeit with a fairly even split between the three -- 33%/37%/29%. I expect that's at least in part due to John Johnson's injury, though it will be interesting to see how the Rams shape up with a healthy lineup in 2020.

Comments

18 comments, Last at 23 Jul 2020, 8:04pm

1 How much of this is governed…

How much of this is governed by what their opponents are doing?  If you play a team lining up 4 wides, you are gonna put more DBs out there, right?   And how much of it can you infer from the game situation, e.g., if you have a large lead?

GB had a lot of one score games and used Dime half the time, so not sure how much....

2 The overall trend is…

The overall trend is certainly governed by offensive trends, for sure -- as three-wide sets have become more and more common, nickel and dime packages have increased accordingly.  There's a 0.97 correlation between 11 personnel frequency and nickel personnel frequency.

But on the individual team level?  Not so much.  For example, if we look at defenses versus 11 personnel in the first half of games with the margin at two scores or less, teams used nickel anywhere from 98.7% of the time (Minnesota) to 26.9% of the time (Green Bay), and teams are fairly evenly spaced between those points.

3 22%/7%/0% in 2019 I thought…

22%/7%/0% in 2019

I thought only Detroit fielded zero defense on 71% of all snaps, but it looks like Buffalo joined them.

4 The Chargers went from…

The Chargers went from playing dime+ 64% of the time in 2018 to just 19% last season, due in part to Derwin James' injury and also to actually having healthy linebackers for once.

It wasn't just Derwin. Adrian Phillips (dime LB/S) also got IR'ed and last season's other dime LB/S was starting at FS this year (Reyshawn Jenkins) because last year's Free Safety Addae went to the Texans.

So, in a year, they lost 3 key contributors to that huge Dime % - 1 in FA and 2 to IR.

They even tried Teamer, an UDFA, as a SS/dime LB but he wasn't good. I think the plan was always to play a lot of dime. Instead, they got stuck playing Thomas Davis, a guy slower than a grandma in flip flops, for most of the year. Predictably, it didn't went so well. 

5 Teams do actually have a…

Teams do actually have a better DVOA against 11 personnel in base than in nickel or dime, but that's almost entirely due to the offense's run/pass ratios; defenses use base personnel when they think the opposing offense is going to run. So while nickel and dime+ defenses ended up facing pass attempts on 72% of their snaps against 11 personnel, base defenses faced passes only 59% of the time -- and those passes were, on average, a full yard shorter than the ones faced by nickel packages. That right there explains most of base's relative success against 11 personnel; they faced easier situations.

This analysis suggests there is a down-and-distance bias to both the offensive and defensive tendency here (and also some potential inefficiency in offensive playcalling). Is that the case?

8 A little bit.  Base…

A little bit.  Base personnel was about 6% of all snaps against 11 personnel with three or fewer yards to go, and about 4% of snaps beyond that; it goes down fairly uniformly for each yard further back you go from the first down marker, with the exception of 10 yards to, 1st-and-10 being  it's own special case.

Teams also mostly used base against 11 on first and second downs; it made up about 6% of all snaps against 11 personnel on downs one and two, and just about 2% on downs three and four.

9 & score?

I think it's likely that game-situation plays a big role as well. All the top teams for use of dime+ defense are good team - something that makes pretty straight forward sense. Not sure how much difference it makes to efficacy of the defense in question, but probably some.

11 Oh, certainly.  We can look…

In reply to by sbond101

Oh, certainly.  We can look at New England, for instance.  Overall, they used Dime+ 41%.  In one-score games, however, that drops to 36%, with Nickel rising from 42% to 46%; base stays pretty much steady, going from 15% to 16%.  As they take a lead, they know their opponents are passing more, so they can shift more and more into dime formations. 

You see the same sort of trend with most of the teams -- no one swings by massive amounts or anything, but yeah; when you're sitting on leads, it can be safer to put more defensive backs onto the field.

14 Normalization

Is it possible to normalize the data for game-situation? Not sure how far it would cut down the sample sizes but I would be very curious how the numbers would shake out when the game was playing "straight" rather than in a pass-bias situation or a run-bias situation - as I mentioned all the top dime-defense teams are all teams that frequently played ahead.

10 Why is no one calling the 4…

Why is no one calling the 4-DB set a "Penny" defense? Makes more sense than "Base" for a groupling whose percentage is so low.

15 What would you call a 4-4…

What would you call a 4-4 then?

i mean, we still call QBs a quarterback even though taking snaps under center is a dying phenomenon, and fullbacks haven’t lined up full back since the 60s. Tight ends are really slot ends or blocking backs half the time.

13 Are You Sure?

“Base defense continues to vanish against 11 personnel, the most effective offensive personnel group.”

One of your other articles has a tweet that points out the reason 11 has better results is due to sheer volume, and that—compared to other passing formations—it is the least efficient.

16 Ah, there's a bit of a…

In reply to by Raiderfan

Ah, there's a bit of a misunderstanding, I think.  11 is by far the most efficient and effective formation.  What it is not is the most efficient passing formation, though I believe it to be the most effective passing formation.

A significant part of the reason why 11 personnel has a higher DVOA than 12 or 21 is because it is the most frequent passing formation.  It is true that 12 and 21 both have better passing DVOA than 11, but that's more due to a) the element of surprise and b) the increased amount of play action in heavier sets.  The best play in football, of course, is the play your opponents don't think you're going to do.  Heck, running out of 11 personnel has a better DVOA than running out of 12 personnel, despite the apparent help of the extra beef blocking up front. 

Football, in it's basic form, is about doing what the other team doesn't expect you to do.  Flea flickers had a DVOA of 155.4%, so you could say that it is the most efficient play in football.  I strongly suspect, however, that if there were more than 22 of them last season, that efficiency would drop significantly.

I would suspect that if teams passed out of 12 and 21 formations at the same frequency they do out of 11 personnel sets, they would be worse, and significantly so -- for most teams, a third receiver is going to allow them to do more in the passing game than a second tight end or running back.

17 Flea flickers had a DVOA of…

Flea flickers had a DVOA of 155.4%, so you could say that it is the most efficient play in football.  I strongly suspect, however, that if there were more than 22 of them last season, that efficiency would drop significantly.

Do you guys calculate a sample-weighted error rate on your DVOA results?

18 Dunbar will play this season

Florida's courts are closed down due to Coronavirus and this has created a huge backlog.  It will probably be at least two years before Dunbar goes to court.