Writers of Pro Football Prospectus 2008

31 Jul 2013

The Defensive Scheme Type Indicator

This is a really good idea. Ted Bartlett of Denver Broncos blog "It's All Over Fatman" has come up with an idea to describe defensive schemes using a series of letters, as if they were personalities in the Myers-Briggs Indicator. You end up with descriptions that are much more detailed than just "3-4" vs. "4-3," not to mention more applicable in a league where teams now have to play nickel or dime the majority of the time. The splits:

Stack (2-gap) vs. Penetrate (1-gap)
Downhill run D vs. Flow run D
Man vs. Zone
Coverage-oriented vs. Blitz-oriented

You may disagree with some of Bartlett's designations of which team is which. For example, I've always thought of the Patriots are more of a Flow run D, because they tend to teach read-and-react, rather than a Downhill run D. In addition, he's made some adjustments for how teams are changing in 2013 but missed a few. Jacksonville will now be playing the same defense as Seattle, for example, and even if they weren't, the Jaguars have always been a primarily-zone team. Nonetheless, it's a pretty good go for a guy who's mainly a Broncos fan. And the idea is really sound. We may have to steal this idea for next year's book.

Well, steal with attribution, anyway.

Posted by: Aaron Schatz on 31 Jul 2013

26 comments, Last at 04 Aug 2013, 3:11pm by Scott Crowder

Comments

1
by Independent George :: Wed, 07/31/2013 - 5:55pm

That is the single greatest blog name I've ever read.

13
by Danish Denver-Fan :: Thu, 08/01/2013 - 5:09am

Agreed. The content is great too. Broncos centric of course, but still useful for non-Broncos fans.

2
by Dr. Mooch :: Wed, 07/31/2013 - 6:33pm

I don't know how well it's going to hold up. I remember a long article detailing Green Bay's base Cover 7 defense. The basis there was a single high safety, two CBs in man coverage on the outside, and variable zones for everyone else. I'm not sure the man/zone distinction is so clear.

3
by Intropy :: Wed, 07/31/2013 - 7:00pm

I'm not so sure the blitz/coverage distinction is either. Consider a team like the Saints. They blitz by bringing a lot of guys. That's an aggressive way to blitz and it results in pressure quickly. Compare to the Steelers. They blitz by bringing four, but making it hard to tell which four. This is accomplished through means such as stunts or dropping a lineman to blitz a safety. This is a safer way to blitz, but the pressure is not usually going to materialize as quickly. For that reason I would tend to consider a zone blitz team, like the Steelers to be actually more on the coverage side of things despite blitzing a lot.

Maybe there needs to be a direct/indirect axis or blitz needs to be blitz percentage x blitz aggressiveness. Which is all to say that the idea is interesting and has enough merit to pick at rather than dismiss.

6
by deflated (not verified) :: Wed, 07/31/2013 - 7:59pm

If you read their definition of blitzing they keep it simple - more than 4 rushing its a blitz, less than 4 it isn't. Sounds good to me.

Just because the Steelers use a particular type of misdirection should not affect the blitz vs coverage split.

7
by Intropy :: Wed, 07/31/2013 - 8:10pm

I did read that section, which is why I thought zone blitz teams would lean coverage. Yet the breakdown said blitz not coverage.

10
by deflated (not verified) :: Wed, 07/31/2013 - 10:22pm

Then they brought 5 or more pass rushers more often than 23 other teams in the eyes of the author. Define a percentage of plays where a team brings 5+ to be a 'Blitz' team and we can talk.

11
by Shattenjager :: Thu, 08/01/2013 - 1:52am

The Steelers were second in the league in 5-man rushes last year and 13 in 6+. They sent 5+ a little over 40% of the time, which I think led the league. In this scheme, that pretty clearly defines them as a blitz team.

22
by kylesotx (not verified) :: Thu, 08/01/2013 - 2:49pm

Bringing four is not classified as a blitz. Even in the DSTI Ted writes that even if the D in questiin is a 3 man line a fourth rusher does not make it a blitz even if it is a safety.

4
by DisplacedPackerFan :: Wed, 07/31/2013 - 7:01pm

That is still the core concept for most of Green Bays coverage regardless of front. They will play that same concept in a 2-4-5, a 3-4-4, a 3-3-5, a 1-5-5, a 2-3-6, etc. Yes they use all those fronts too. But they do a lot of man outside, zone inside, but that outside man coverage might actually be from the slot corner shifting out while the outside corner is covering the flat zone or blitzing. I'd say that it's the style of D they play 60% of the time even if it's not obvious. That's part of the point though, to disguise things.

I'm OK calling them a man coverage team though, in part because they play pure zone infrequently and they will do straight man without any zone concepts more frequently. The outside man coverage is also really the key to much of the D as well. When Williams played most of the year hurt and and Shields had his down year in 11 they had lots of issues because the man play didn't work. In the NYG game last year when they started House and his man coverage failed they collapsed as well. They aren't relying on a player like Revis to "take away half the field" but they are relying on the two outside man guys to "take away half the field". So you have 2 players, with one more who can help, covering half the field and then 2 - 4, with that same one for help, for the other half.

If you can add a good slot guy like Woodson (and what Hayward is shaping up to be) to that you get that 09 and 10 level defense. A MLB with great coverage skills (something they have never had) or a great pass rusher with a decent compliment also works. But the idea is that you only need 3 players, with one of them needing to be a good man corner, and everything else covers for the weaknesses. If all three players are cover guys, then the pressure happens because of extended time to apply it. If two of the guys are pressure players that should make up for the weak coverage. If two coverage and one pressure, hopefully the pressure gets there or you have enough weak to OK zone guys to hold up, etc. If you can get 4 guys like they had with Tramon's career year, Woodson's last great year, Matthews being Matthews, and then Shields being above average, with Jenkins also given extra pass rush you get outstanding defense and win the Super Bowl. :)

19
by Jimmy :: Thu, 08/01/2013 - 11:14am

I thought the Pack played man free most of the time - ie CBs on outside shoulders leveraging receivers towards their help. Then LBers not blitzing or inn man coverage drop into hook zones to help take away anything underneath.

25
by Arkaein :: Fri, 08/02/2013 - 12:20pm

Don't forget about Nick Collins. His terrific range was really key to playing their outside man defense, since GB was able to leave him as the only deep safety backing up the CBs in man and the LBs who rarely dropped deep. Made it a lot easier for the average players to smother the middle of the field, or blitz without risking much on the back end.

If anything, I'd say Collins was more important than Woodson in 2010, since Woodson had a down year by his standards, at least in coverage.

5
by Karl Cuba :: Wed, 07/31/2013 - 7:22pm

There are always going to teams (and bits of teams) that are hard to put in either box but I think this is a very good, and welcome, attempt to provide a shorthand for classifying defenses. In the spirit of FO's anniversary, genuflections.

8
by deflated (not verified) :: Wed, 07/31/2013 - 8:12pm

Really like this - I'm sure there can be endless discussions on exactly where to draw the dividing lines but having a shorthand for defensive tendencies that goes beyond 4-3/3-4 is a great idea. There will always be some teams that are tough to classify but I think for any normal defense 3 of the 4 splits will be obvious.

9
by NHPats (not verified) :: Wed, 07/31/2013 - 9:45pm

Hmmm.....can the game-charting data be used to extract these values on a play-by=play basis that can be rolled up to see if there's an objective basis to describe teams in this framework?

Once you're got that grading done, you have a dataset that would let you look at DVOA by style for a team, and look at style versus game context...and...

Wow, lots of possibilities

14
by CBPodge :: Thu, 08/01/2013 - 6:41am

In theory you could do it for most of them now, I think. Blitz V Cover is tracked through ESPN. Man V Zone isn't in charting, but it could be done. It'd be tricky, and might make charting take significantly longer. You can't do it from the standard game angle, so you'd need to use all-22, which isn't (IMO) any use for standard charting (because its not good enough quality to easily ID players), so it might take a separate run through of the game film. It wouldn't take as long as normal charting (which takes roughly 2 hours per half) as you wouldn't need to rewind at all - you'd just mark whether it was clearly zone, clearly man, or neither on each play. A run through of the game on condensed, skipping run plays, would only take half an hour per game or so. I guess, with enough charters, you could assign extra charters 4 games a week to just mark man or zone coverage.

Say you throw out 50% of passing plays where you aren't certain of whether its man or zone, that would still leave an absolute minimum of about 200 plays per team (and likely much more). Even if you then factor out plays where the score isn't close to factor out prevent defence, it should still give a number that would be at least vaguely statistically viable.

That'd be doable, I think, but it'd rely on getting enough willing charters, which ain't easy.

Stack V Penetrate - I don't really see how charting could do this, without making it take hours. You'd have to look at each play from the endzone view, and decide whether a majority of linemen were lined up over an OL or between them. And then what would you do for times when half them are, and half them aren't? Or when they are, but then at the snap seem to look to penetrate. Or when the OL blocks left or right, changing where the gaps are? This one will always have to be broadly subjective. You can get it from film study, but you wouldn't (I think) ever be able to say that a team is 2-gap 93% of the time and 1-gap 7% of the time.

Downhill V Flow - I guess in theory you could do this, if you worked out a common way to mark it. I guess you could mark it as whether a particular defender (say the MLB) or combo of defenders (all LBs? MLB and SLB?) heads straight downhill at the snap on running plays. I think it'd be too subjective to be particularly accurate across charters though.

That being said, I think both of these could be estimated from the stats on stop/defeat rate and average yards making a tackle. If you threw out every tackle more than say 8 yards downfield (to avoid plays from a defender hustling back up the field to stop a longer run), you could potentially say that teams that penetrate will have DLs with higher stop rates and lower yards per tackle than teams that stack, and teams that play downhill will have LBs with higher stop rates and lower yards per tackle.

Maybe you could divine it a bit from location of tackles? If you find that a team has all its linebackers making plays in specific areas (say particular LBs making a high percent of run tackles on plays marked as "right", "left" or "middle"), it could indicate a more downhill scheme, and if there wasn't any sort of correlation between play direction and who makes the tackles it could indicate a more flow scheme. That seems a bit wooly though. I think it might be something worth looking at, but I'm not sure if you'd see anything on these two though. They seem more subjective.

16
by CBPodge :: Thu, 08/01/2013 - 7:16am

Bah. I ran the numbers for run plays for DL and LBs for the Ravens (Stack, Flow, Cover) and the Chargers (Penetrate, Downhill, Blitz) and tried to put them in with some thoughts, but its getting blocked by the spam filter. Will try again, with less stuff:

Ravens: 65% stop rate, 3.3 yards per tackle
Chargers: 68% stop rate, 2.7 yards per tackle

Backs up the guesses on those two Ds in the article, I guess.

12
by allybhoy :: Thu, 08/01/2013 - 5:09am

No Jets or Browns :(

15
by rfh1001 :: Thu, 08/01/2013 - 6:49am

Love this. On practical terms, 16 TLAs are going to confuse anyone who isn't really concentrating BUT I can imagine a world where the dichotomies are understood widely and defences can be more subtly and usefully described now in terms of their PRIMARY characteristics on this scale - thus X run primarily a Penetrative, Coverage style D.

But I love it.

17
by CBPodge :: Thu, 08/01/2013 - 7:22am

I agree with that. There's no point calling the Rams a PFZC, or the Steelers a SDZB, but if you called the Rams a penetrative, zone coverage defence, and the Steelers a downhill, zone blitzing D, that'd be pretty descriptive, and feels quite accurate to how I view those teams. The Rams D feels defined by a strong DL and the rest of the D trying to contain while the DL gets to work. The Steelers D feels defined by zone blitzing and players flying down to the line against the run.

18
by Jimmy :: Thu, 08/01/2013 - 11:10am

As far as I know 2gap is flow and 1gap is penetrate - or at least they are if you are doing it right.

Essentially a one gap flow team is another way of writing 'terrible run defense' similarly for two gap penetrate, you are just going to end up with huge running lanes.

I think there is some value in the process, just three options not four.

20
by EastCoastBronco (not verified) :: Thu, 08/01/2013 - 1:40pm

Jimmy,

You are using the wrong labels. Flow is describing what the LBers do (Downhill vs Flow), whereas penetrate is a DL label (Stack/2gap vs Penetrate/1gap).

21
by EastCoastBronco (not verified) :: Thu, 08/01/2013 - 1:49pm

I am assuming you were strictly talking about the DL. In that case obviously a 1 gapping, penetrating DL scheme should not "flow" to the ball, and visa verse. However there are plenty of teams whose DL 1gap while their LBs flow (read and react). Examples that Ted Bartlett (the Author) uses for this are the Bengals and the Lions. Both teams have good defenses.

Examples of teams that stack (2gap) their DL but have their LB attack downhill are Patriots and Redskins. Both of these teams have at least decent run defenses.

If you were strictly talking about the DL, then I think you just accidentally mixed up the labels. If you were talking about the general front 7, then you are wrong. Teams can penetrate and flow or stack and attack. The first label is the DL, the second is the LB.

23
by Karl Cuba :: Thu, 08/01/2013 - 8:04pm

I think you are wrong, there are two approaches to run defense: Force and Contain, and Fill and Spill. It's about gap responsibility and you don't mix the two. Here's a good article explaining it:

http://www.ninersnation.com/2012/8/24/3243940/breaking-down-the-49er-run...

When you start trying to pick apart whether or not a 'force' team's linebackers move more aggressively downhill or not is splitting hairs between coaching points. There is no defense that plays part of one or the other on a given play, they might switch their approach between plays but as Jimmy points out, playing different concepts on different levels is a recipe for giving up huge running plays on a very regular basis.

"Examples of teams that stack (2gap) their DL but have their LB attack downhill are Patriots and Redskins. Both of these teams have at least decent run defenses." - This is my point, how is this any different from the 49ers linebackers who will head down field in a hurry the second they read the play? Basically defenses either look to set the edges and contain the runner within that boundary or look to clog the middle and force the runner to head to the sidelines where the defensive pursuit will chase them down. The Bengals and Lions are Fill and Spill while the Pats are Force and Contain.

The article does a better job than me in explaining it.

24
by Jimmy :: Fri, 08/02/2013 - 9:55am

I haven't mixed up anything. Run defense schemes encompass all the front seven players (or however many remain in the box when offenses spread out) every player has to be playing the same scheme. You seem to be implying that in a one gap scheme a lineman will have a gap but a linebacker may be free to flow. That is nonsense. In a one gap scheme all seven players in the front seven will have responsibility for (wait for it......) one gap. In a two gap system all seven guys will have two gaps - which in reality means either side of one blocker, so they get a player not a gap.

And linebackers are supposed to attack downhill no matter what system they are in. The alternative is standing several yards behind the line waiting for a guard to come and de-cleat them (see the FO article about Buffalo's run defense last year). In a one gap they are supposed to be trying to get through their gap if at all possible but if not shed and pursue. In a two gap scheme they are supposed to go and stack the blocker as close to the line (and any other blockers) as possible to reduce the space available to the back. The only way to do this to guys 50lbs bigger than they are is to head downhill and get under the blocker's pads (Willis and Bowman are brilliant at this but if you watch Lance Briggs you will see the same technique).

26
by Scott Crowder (not verified) :: Sun, 08/04/2013 - 3:11pm

Something I noticed is that the top 5 defenses per FO are all Flowing Coverage teams (while Bartlett projects Seattle as a blitzing team, in 2012 they were primarily a coverage team, depending upon their 4 man rush.).