Writers of Pro Football Prospectus 2008

04 May 2012

Seahawks Defense: Beyond Prototypical

On the heels of the controversial Bruce Irvin draft pick (as discussed earlier this week by Mike Tanier) comes this piece by Clare Farnsworth on the Seahawks' official site breaking down Pete Carroll's unorthodox defensive schemes. Carroll's own description: "Our defense is a 4-3 scheme with 3-4 personnel." Farnsworth also talks about the common threads between Carroll's defenses in college and the pros: A yin-yang set of defensive ends (one big run stuffer, one small pass rusher); cornerbacks known more for size than speed; and at least one big, physical safety who plays almost like an extra linebacker. Another tidbit: Farnsworth predicts that sixth-round draftee Winston Guy will see the field often in triple-safety formations.

Posted by: Vincent Verhei on 04 May 2012

29 comments, Last at 01 Apr 2013, 5:21pm by jam


by Ferguson1015 :: Fri, 05/04/2012 - 4:05pm

"Our defense is a 4-3 scheme with 3-4 personnel."

Sounds like the Raiders

by sundown (not verified) :: Fri, 05/04/2012 - 4:26pm

Interesting stuff. I like seeing teams not afraid to do things a little differently. And Seattle has had pretty good success with this scheme.

by Lance :: Fri, 05/04/2012 - 4:30pm

While I can't comment on what SEA is doing, I wonder if we'll start to see more and more unorthodox things as teams try to adjust to the new pass-happy NFL. My feeling is that it might be wise to start doing something like a 4-2-5 where the 5th DB is a freakishly athletic S who has near LB bulk, but DB speed so as to keep up with pass routes but still make solid hits.

by tuluse :: Fri, 05/04/2012 - 4:56pm

"My feeling is that it might be wise to start doing something like a 4-2-5 where the 5th DB is a freakishly athletic S who has near LB bulk"

So all it takes is to have a Troy Polamalu, and also have two starting caliber safeties in addition.

I think the Bears have been trying to do this for years (a safety drafted in each of the past 8 drafts), but again it's hard to find these players.

by Steve in WI :: Fri, 05/04/2012 - 5:30pm

In fairness to the theory, it's even harder to find these players when you have Jerry Angelo running your draft.

by tuluse :: Fri, 05/04/2012 - 5:47pm

Eh, I don't think Angelo was particularly bad at drafting (not good either mind you), unless you want to blame him for drafting injury prone players.

Also, I think the Bears are going to be quite good next year mainly due to contributions from last years draft. I think Conte and Carimi are going to be quite good, and I'm hoping Paea will come through.

by AnonymousD (not verified) :: Sat, 05/05/2012 - 9:31am

Concerned about the age of Peppers/Briggs/Urlacher?

by tuluse :: Sat, 05/05/2012 - 2:06pm

There were all good last year, and I'm not expecting them to fall off a cliff. I think a gradual decline over a few years is more likely.

by JasonK :: Sat, 05/05/2012 - 11:50am

The Giants have been using Deon Grant and Antrel Rolle in that kind of role over the past couple of seasons-- the "big nickel" has probably been their most common defensive personnel package. It covers receiving TEs pretty effectively, but it probably hurts run D a bit, and offenses that manage to get a slot WR in against this personnel tend to have a lot of success.

My guess at Perry Fewell's thinking behind this is that the modern NFL TE is more a receiver than he is a blocker, so in most cases it's better to counter that kind of player with a DB than with an LB. (Or it might just be that the Giants' linebackers are kinda terrible.)

by arias :: Mon, 09/03/2012 - 3:53am

Well they've got Earl Thomas (2nd team all-pro) in the Polamalu role and Kam Chancellor as pro bowl SS so I think the missing link of Jeron Johnson as the other starting caliber strong safety has been found.

by Shattenjager :: Fri, 05/04/2012 - 5:33pm

I've said before that it seems that the rover back should return, replacing one of the traditional linebackers.

by commissionerleaf :: Fri, 05/04/2012 - 5:46pm

Actually, I would argue that the addition of a third defensive back is weak tea compared to what defenses should be doing. The success of passing offenses even against nickel defense is an indication that too little attention is being paid. The whole concept of having three - or even two - players called "linebackers" who are neither pass-rushing nor pass-coverage specialists is, to put it bluntly, vestigial.

In 2011, teams passed for about twice as many yards as they ran for, on substantially more attempts (about 500 per team rather than 400). 6.3 net yards per pass (including sacks), 4.3 per run. There were 800+ turnovers, of which only about 500 were interceptions.

Game theory right now favors the pass, because teams put far too much effort into defending the run.

by tuluse :: Fri, 05/04/2012 - 5:48pm

I'm not convinced, I've seen teams sell out against the pass, and when they do they get ran all over.

by RickD :: Sat, 05/05/2012 - 12:58pm

That is the problem, isn't it? If the running game is not contained below a certain level, the pass defense doesn't matter at all. And the level it needs to be contained at is (roughly speaking) less than 4 yards/carry. A team that can rush for 4.5 yards/carry or more can march up the field consistently.

The reason that yards/pass attempt need not be held to the same constraint is the high variance of the passing play. As pointed out below, the defense is trying to get off the field, not trying to minimize expected yards/play.

Let's say I have a running game that gets 3.5 yards/carry, with no variance.

And let's say I have a passing game that has the following outcomes
1/3 of the time I get no yardage
1/3 of the time I get 4 yards
1/3 of the time I get 8 yards

I will average 4 yards per attempt, but there is still a positive probability that I will not gain a first down. The rushing game will gain the first down every time.

Variance matters.

by erniecohen :: Sun, 05/06/2012 - 11:15am

Variance does matter. However, the variance in the running game is much greater than most people realize. FO had an article about this a few years back (though it might have focussed on short yardage situations).

4.5 YPC is nowhere near what you need to march down the field rushing. This year, 8 teams averaged 4.5+ YPC (Carolina was 5.4), and none of them dominated with rushing. Jim Brown averaged 5.2 YPC for his career, but his teams didn't march down the field rushing like college teams in the 70's. To really dominate with rushing, you need to average more like 6-7 YPC.

by tuluse :: Sun, 05/06/2012 - 1:46pm

Against a standard defense it's variable, not necessarily against a defense selling out against the pass.

by Aaron Brooks Go... :: Mon, 05/07/2012 - 12:07am

The Brown Browns basically did march the ball downfield by rushing. They ran more than they passed, they ran for more yardage and 1st downs than they passed for, had fewer TOs rushing than passing, and outside of his first year, had comparable yards/attempt rushing versus passing.

The Browns ran over teams that sold out against the rush.

by LionInAZ :: Mon, 05/07/2012 - 12:15am

I'd love to hear tuluse explain to me how the Packers defense managed to get through the season. From what I could tell, they pretty much sold out against the pass, running 2-man DL setups more often than not, yet teams rarely ran over them -- and they had the worst pass defense (in terms of yardage) in the NFL.

by tuluse :: Mon, 05/07/2012 - 2:24am

They were 23rd by DVOA.

As for how they got through the season? Not very well, and played with the best offense in the NFL.

Combine the Bears offense and Packers defense, and you would have seen a lot more running and such a team would be lucky to win a game.

Also, 2 d-lineman doesn't by itself mean you're selling out against the pass. The Packers ran a lot of run blitzes from the secondary, and Charles Woodson is basically a linebacker at this point.

by Jimmy :: Tue, 05/08/2012 - 11:45am

What it tells us is that you are only going to play two linemen then make sure they are both really, really fat.

by Jeff M. (not verified) :: Sat, 05/05/2012 - 1:45am

The per-play averages don't actually give us enough information to determine optimal strategy. Offenses aren't simply trying to maximize an expected sum of yards over a sequence of plays; they're trying to maximize the probability of getting >10 yards in the next three plays (more complicated than this in actuality of course, but think "successful plays").

There absolutely could be a mean-variance tradeoff (I'm almost certain passing plays would be more variable, but again this is my point that we would need the relevant information) where three run plays have a lower expected total yardage but still have the same or greater probability of picking up at least a first down.

by Pat (filler) (not verified) :: Sat, 05/05/2012 - 10:20pm

It's worse than that. For strategy, you can't really pull out the plays and examine them independently. We already know there's intentional play-calling sequencing in the NFL (many teams script the first plays). It's entirely likely (and I'd say entirely probable) that the success of a play depends, quite a bit, on what the team has done prior to that in the game.

If you stop thinking of football as a sequence of independent zero-sum games, and start thinking of it as a linked series of them, it's entirely possible that you can get bizarre situations like teams defending against the run quite a bit, because the marginal benefit to the passing game of allowing success in the running game is enough to equal the costs.

Any time I see someone try to do a game-theory analysis on running/passing in football, I just roll my eyes. You can't do it. It's waaaay too simplistic.

by erniecohen :: Sun, 05/06/2012 - 10:54am

There is nothing wrong with doing game-theory analysis of running/passing in football. What you call a "linked series" is captured by the concept of *state*, which identifies what you need to know about the past to predict (statistically) the future to sufficient degree of fidelity. The way you figure out what has to be in the state is that you try to add corrections for additional stuff in the past and see if it changes things in a statistically significant way. (Of course we have the problem that there is a rather limited amount of real game data to work with, and you can't do lots of experiments on demand.)

For example, most studies ignore the score differential (which is certainly a relevant part of the state), but its one of those things that you can add to the state, do a regression to find out how to weight it in the run/pass decision, and see whether you get a better formula. Similarly, you might consider something like the yards per rushing/passing attempt as a state parameter and put that into you model, and see if it really makes a difference.

by Alex (not verified) :: Mon, 05/07/2012 - 6:28pm

I think the Eagles tried following your logic last year...

by Theo :: Sat, 05/05/2012 - 3:46pm

Google "4-3 over defense pete carroll" and there are a few nice diagrams of a defense he ran at USC.

by North Coast (not verified) :: Sat, 05/05/2012 - 4:06pm

Great comment by Rick D.

by erniecohen :: Sun, 05/06/2012 - 11:05am

It's a nice article, but I think it gives a false impression that Seattle is unique in this regard. Every good coach designs schemes that are designed to take best advantage of the particular skills of the players they have. Tomlin might say "next man up" and "the standard is the standard", but PIT will call very different plays if Taylor or Polamalu is out.

by Jimmy :: Tue, 05/08/2012 - 11:44am

True. The Seattle scheme is not very far removed from the scheme Clancy Penderghast used when he was the DC of Arizona. CP wanted to line his players up where they could play up to their potential instead of play down to their limitations. A team was in the Superbowl recently using similar ideas (from the same bloody division) yet it is written about as though it were quantum physics.

by jam (not verified) :: Mon, 04/01/2013 - 5:21pm

The success of passing offenses even against nickel defense is an indication that too little attention is being paid. The whole concept of having three - or even two - players called "linebackers" who are neither pass-rushing nor pass-coverage specialists is, to put it bluntly, vestigial. psychic phone reading