The Best Defense is a Good Offense - Pats 2016 Defense Discussion

BrazilianSoxFan

Member
SoSH Member
Apr 11, 2006
3,751
Brasil
I don't think against a neutral schedule they are a legit top 5 defense, they certainly beneftited from the offense scoring so much that it made other teams one-dimensional for long stretches, and their special teams didn't give up points.
The defense actually performed better when the games where still close, giving up only 6.8 first-half points per game, by far best in the league.

https://www.teamrankings.com/nfl/stat/opponent-1st-half-points-per-game

And didn't they go the whole season without giving up a first quarter TD?

Edit: they actually gave up two first quarter TDs, week 4 vs Buffalo and week 5 @ Cleveland.
 
Last edited:

eustis22

New Member
Nov 14, 2016
998
>but that's about their only weapon.

That kid Shepherd is GOOD. And Cruz still has something left in the tank.

Am I the only one who thinks the bend but don't break D is because BB is motherfucking sick and tired of motherfucking long miracle catches in the motherfucking Super Bowl and has resolved never to be beaten (or almost beaten) by one ever again?
 

johnmd20

mad dog
Lifetime Member
SoSH Member
Dec 30, 2003
62,085
New York City
>but that's about their only weapon.

That kid Shepherd is GOOD. And Cruz still has something left in the tank.

Am I the only one who thinks the bend but don't break D is because BB is motherfucking sick and tired of motherfucking long miracle catches in the motherfucking Super Bowl and has resolved never to be beaten (or almost beaten) by one ever again?
Yes, you are. You cannot in any way prepare for miracle catches. You can't plan for them and you can't plan against them. Because they are miracles.
 

Rico Guapo

New Member
Apr 24, 2009
2,174
New England's Rising Star
>but that's about their only weapon.

That kid Shepherd is GOOD. And Cruz still has something left in the tank.

Am I the only one who thinks the bend but don't break D is because BB is motherfucking sick and tired of motherfucking long miracle catches in the motherfucking Super Bowl and has resolved never to be beaten (or almost beaten) by one ever again?
Cruz is a shell of his former self, Shepherd is decent, but the Giants can't run the ball...it's just a bad offense.
 

johnmd20

mad dog
Lifetime Member
SoSH Member
Dec 30, 2003
62,085
New York City
Cruz is a shell of his former self, Shepherd is decent, but the Giants can't run the ball...it's just a bad offense.
Their running game is especially bad. Jennings runs like he's in quicksand. And he's their best runner. Dion Lewis makes Jennings look like he's in junior high.
 

bigsid05

Member
SoSH Member
Jul 15, 2005
2,327
Fine. I'll be the only one. But that's what BB thinks. He's thinking "Fuck Miracle Catches, I'm taking them all"



Watch Perkins.
Uh... bend but don't break D was what was beaten by those miracle catches. It's been the philosophy through the wins and losses.
 

ShaneTrot

Well-Known Member
Lifetime Member
SoSH Member
Nov 17, 2002
6,447
Overland Park, KS
>but that's about their only weapon.

That kid Shepherd is GOOD. And Cruz still has something left in the tank.

Am I the only one who thinks the bend but don't break D is because BB is motherfucking sick and tired of motherfucking long miracle catches in the motherfucking Super Bowl and has resolved never to be beaten (or almost beaten) by one ever again?
I think the Pats play a lot of bend but don't break defense because their secondary is very good, they tackle well and BB believes most QBs will screw up before they march their team down the field. I also think its a by product of having OK not great edge rushers.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Some of you might remember my having posted some plots of season-avg defensive allowed points per game versus allowed yards per game, for all teams seasons during the Pats/BB-era. The Patriots always allowed fewer PPG than would be expected given their allowed YPG, and I interpreted this as evidence for a successful "bend but don't break" defense. Others pointed out that having a good offense and special teams can skew that simple way to look at the data. Well, on a whim I today looked at the data again, but factoring in average defensive drive line-of-scrimmage starting position. It turns out that starting LOS explains most of the trend. The Patriots still do better at preventing points given their defensive starting LOS and yards allowed, over the span 2001-2016, but it's only a 1-sigma signal. In the 2016 regular season they prevented ~5pts/game more than would be expected, which is a ~3-sigma signal, but considering the fact that there are many fluky things that haven't been factored into the calculation of that sigma, it probably isn't very interesting.

N.B.: this isn't at all to say that the Patriots don't have a good defense this year. It's just to say that the discrepancy between their yards-allowed and their points-allowed rankings is largely if not completely due to the fact that their opponents have farther to go for those points compared to the average team.

If anyone has any interest in the methodology, I can post some plots and numbers later.
 

InstaFace

The Ultimate One
SoSH Member
Sep 27, 2016
22,261
Pittsburgh, PA
Very cool stuff, dbn. Worth its own thread with the detail you offer, perhaps some day this week when things are slow for you and everyone's getting antsy about Sunday.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,024
Mansfield MA
N.B.: this isn't at all to say that the Patriots don't have a good defense this year. It's just to say that the discrepancy between their yards-allowed and their points-allowed rankings is largely if not completely due to the fact that their opponents have farther to go for those points compared to the average team.
To add context, they were 8th in yards allowed this year, so still pretty good. That ranking jibes more with my subjective assessment of the team; closer to a fringe top-10 D than a real top-of-the-league elite unit. I'd take them over any of the Ds remaining in the playoffs though (especially over either NFC team).
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
To add context, they were 8th in yards allowed this year, so still pretty good. That ranking jibes more with my subjective assessment of the team; closer to a fringe top-10 D than a real top-of-the-league elite unit. I'd take them over any of the Ds remaining in the playoffs though (especially over either NFC team).
In fact, there is - unsurprisingly - an anti-correlation between starting defensive LOS and defensive yards allowed. The Patriots were #1 in the NFL for def starting LOS, so their #8 ranking in yards allowed more impressive in that light.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,024
Mansfield MA
In fact, there is - unsurprisingly - an anti-correlation between starting defensive LOS and defensive yards allowed. The Patriots were #1 in the NFL for def starting LOS, so their #8 ranking in yards allowed more impressive in that light.
There is a mild such correlation (~ -0.2 in 2016).
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
(un-edit: nope, it looks okay)

I'm still playing around with the data, but IMO it's getting rather interesting. The strong correlation between points-allowed/dive and yards-allowed/drive lets one use the latter to predict the former, but with significant scatter, of course. Factoring in defensive starting line-of-scrimmage tightens up the scatter, and makes the Patriots' putative bend-but-don't-break defense look - on average over 2001-2016 - more like an artifact of their good defensive starting field position.

However... the five of the ten largest differences in average predicted-minus-actual points allowed per drive over that period are Patriots teams. Thats 5 of the 16 Patriots seasons are in the top 10 of the 511 data points, or 31% of the Patriots' seasons are in the top 2% of all NFL seasons during that span. So, maybe there is something to this BBDB defense after all?

I'd be happy to write this all up in a ITP-style article with cool plots, equations, and discussion, if ITP is interested. Otherwise, I'll start a new thread and put it there.
 
Last edited:

DJnVa

Dorito Dawg
SoSH Member
Dec 16, 2010
54,114
Factoring in defensive starting line-of-scrimmage tightens up the scatter, and makes the Patriots' putative bend-but-don't-break defense look - on average over 2001-2016 - more like an artifact of their good defensive starting field position.
Perhaps I'm missing something obvious, but isn't that the point? By playing the field position game, and playing a bend but don't break philosophy, you make teams make more plays/drive, which gives them more opportunities to make mistakes.

That, coupled with an offense that scores a lot of points, forces offenses into more predictable patterns on offense, making it even more likely that the defense is well-positioned for what the offense does.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Maybe I'm missing something too; what do you mean by "playing the field position game"?

On defense, playing the field position game - i.e., being willing to give up field position when on D in hopes that you lower the expected points given up - seems like the opposite of a BBDB defense. That's moot, however, because that affects the field position of the Patriots' offense.

Do you mean the Patriots' offense plays the field position game? How so?

I think the reason they consistently are around the best in the league at defensive drive starting field position is because they simply have a good offense (and thus are rarely punting from deep on their own side of the field) and have good special teams kick coverage.

Maybe they just play defense (not BBDB), and the discrepancy between their points and yards allowed rankings are an artifact of their good O and ST. What I'm finding so far is that in most seasons their (points allowed - predicted points allowed from yards allowed and defensive starting LOS) is right around where you'd expect, but that five of 16 seasons lie at the envelope of the overall NFL distribution suggests that there is more to it - perhaps a BBDB philosophy, or perhaps something else.

edit: I'll try to look into whether those 5 outliers are also seasons with significant "average lead at start of defensive drives". Not sure what that would mean in terms of yards vs points, but could be interesting.
 

FL4WL3SS

Member
SoSH Member
Jul 31, 2006
14,926
Andy Brickley's potty mouth
Some of you might remember my having posted some plots of season-avg defensive allowed points per game versus allowed yards per game, for all teams seasons during the Pats/BB-era. The Patriots always allowed fewer PPG than would be expected given their allowed YPG, and I interpreted this as evidence for a successful "bend but don't break" defense. Others pointed out that having a good offense and special teams can skew that simple way to look at the data. Well, on a whim I today looked at the data again, but factoring in average defensive drive line-of-scrimmage starting position. It turns out that starting LOS explains most of the trend. The Patriots still do better at preventing points given their defensive starting LOS and yards allowed, over the span 2001-2016, but it's only a 1-sigma signal. In the 2016 regular season they prevented ~5pts/game more than would be expected, which is a ~3-sigma signal, but considering the fact that there are many fluky things that haven't been factored into the calculation of that sigma, it probably isn't very interesting.

N.B.: this isn't at all to say that the Patriots don't have a good defense this year. It's just to say that the discrepancy between their yards-allowed and their points-allowed rankings is largely if not completely due to the fact that their opponents have farther to go for those points compared to the average team.

If anyone has any interest in the methodology, I can post some plots and numbers later.
Wouldn't any successful analysis of "bend don't break" defenses have to include success in the red area and number of scoring plays outside of red zone (i.e. the breaking part of 'bend don't break').
 

FL4WL3SS

Member
SoSH Member
Jul 31, 2006
14,926
Andy Brickley's potty mouth
Perhaps I'm missing something obvious, but isn't that the point? By playing the field position game, and playing a bend but don't break philosophy, you make teams make more plays/drive, which gives them more opportunities to make mistakes.

That, coupled with an offense that scores a lot of points, forces offenses into more predictable patterns on offense, making it even more likely that the defense is well-positioned for what the offense does.
He's including back to 2001 and the Patriots did not have an elite offense until 2007. I don't think an offense that scores a lot of points is necessarily a requirement to the field position game. The Patriots have always been able to move the ball efficiently under Brady (he's always been at an elite level completion percentage), so they didn't necessarily score a lot of points early in his career, but they could move the ball.
 

Stitch01

Member
SoSH Member
Jul 15, 2005
18,155
Boston
Interesting stuff. Id have guessed that most of the "bend but dont break" benefit was a mirage that really came from having good offense/ST, so will be interesting to see if something stands out in the outlier years. Wonder how the turnover rate fits in, the '11/'12 Pats teams in particular stand out as having unsustainably high turnover rates.
 

Stitch01

Member
SoSH Member
Jul 15, 2005
18,155
Boston
He's including back to 2001 and the Patriots did not have an elite offense until 2007. I don't think an offense that scores a lot of points is necessarily a requirement to the field position game. The Patriots have always been able to move the ball efficiently under Brady (he's always been at an elite level completion percentage), so they didn't necessarily score a lot of points early in his career, but they could move the ball.
Plus they've had good to elite ST's for some absurd amount of years in a row, like back to the Parcells era.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Wouldn't any successful analysis of "bend don't break" defenses have to include success in the red area and number of scoring plays outside of red zone (i.e. the breaking part of 'bend don't break').
I thought about that but decided not pursue it, for a combination of reasons. Mostly because it complicates things a lot, and I like clean analysis.

edit: as "red zone" is an arbitrary cut-off, I think the thing to look at would be average length of TD-scoring play (i.e., not drive) against. If a team gives up few yards but many long TDs, they're breaking but not bending. I'm not sure where to find that data.
 

FL4WL3SS

Member
SoSH Member
Jul 31, 2006
14,926
Andy Brickley's potty mouth
I thought about that but decided not pursue it, for a combination of reasons. Mostly because it complicates things a lot, and I like clean analysis.
But that's the whole point right? If you give up big plays, then you're breaking more than bending. A bend but don't break defense would have a very high percentage of it's points against scored in the red zone**, but also have a very high ranking of red zone efficiency.

**Counter-intuitive, but if you're not giving up big plays, then the points that are actually scored against you would be from the red area.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
But that's the whole point right? If you give up big plays, then you're breaking more than bending. A bend but don't break defense would have a very high percentage of it's points against scored in the red zone**, but also have a very high ranking of red zone efficiency.

**Counter-intuitive, but if you're not giving up big plays, then the points that are actually scored against you would be from the red area.
Indeed, that would be an intriguing way to approach the question.

I approached it from the yds-vs-pts allowed direction, because it seems that discrepancy has been what has driven the BBDB narrative. If I can figure out a way to get big-play data, it will be fun to take that approach as well!
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
I should also mention that - since this is a thread about the 2016 NEP defense - that in 2016 the Patriots gave up about as many points as would be expected given the combination of the yards they gave up and their average defensive starting field position. (Their best season in the data set for beating their expected points given given allowed YDS and DSFP was 2001; their other noteworthy seasons were 2010, 2004, 2006, and 2011.)
 

DJnVa

Dorito Dawg
SoSH Member
Dec 16, 2010
54,114
Maybe I'm missing something too; what do you mean by "playing the field position game"?

Do you mean the Patriots' offense plays the field position game? How so?
I mean on offense, but I'm not sure all the ways they may do that. Early in the season it was Gostkowski's kickoffs. I'm sure there are other ways. You mentioned the Patriots good starting defensive field position. Are they generally near the top in this? Would we think that's coincidence?

He's including back to 2001 and the Patriots did not have an elite offense until 2007.
Gotcha. Maybe not "elite", but they've been top 10 in points scored every year since 2001 except 2003.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
I mean on offense, but I'm not sure all the ways they may do that. Early in the season it was Gostkowski's kickoffs. I'm sure there are other ways. You mentioned the Patriots good starting defensive field position. Are they generally near the top in this? Would we think that's coincidence?



Gotcha. Maybe not "elite", but they've been top 10 in points scored every year since 2001 except 2003.
Yes, they have been elite at this, year after year.

Their ranks from 2001-2016: 2, 15, 9, 17, 6, 5, 2, 6, 2, 6, 2, 6, 3, 2, 2, 1.
 

SMU_Sox

queer eye for the next pats guy
SoSH Member
Jul 20, 2009
8,945
Dallas
I want to know if there is a correlation between the average number of drives in a game and points-per-drive. Same analysis but looking at number of plays per game and then one looking at both plays per game and drives per game. I'd also want to see score differential at the time of the drive built in. I feel like there are a lot of contextual factors to investigate. Great initial work though.
 

BaseballJones

ivanvamp
SoSH Member
Oct 1, 2015
24,714
This is a big outcome of the philosophy BB has taken with roster spots. He has consistently dedicated 2-3 spots for core ST'ers and the fruit bears out in these numbers.
Yep. It also helps that the Patriots are typically among the league's elite at not turning the ball over. Turnovers, on average, have a significant impact on field position.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
I want to know if there is a correlation between the average number of drives in a game and points-per-drive. Same analysis but looking at number of plays per game and then one looking at both plays per game and drives per game. I'd also want to see score differential at the time of the drive built in. I feel like there are a lot of contextual factors to investigate. Great initial work though.
I want a pony.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,024
Mansfield MA
Yes, they have been elite at this, year after year.

Their ranks from 2001-2016: 2, 15, 9, 17, 6, 5, 2, 6, 2, 6, 2, 6, 3, 2, 2, 1.
They have also generally had some of the best starting field position on offense, too (they're third in 2016). I think all this stuff goes together. O produces and doesn't turn the ball over. They kickoff after a score or punt after a failed drive but generally have good coverage. The D has favorable field position to start, might give up a few yards, but then generally forces a punt (which is usually better than a kickoff for field position). The punt return team is good, so the O gets good field position to start ...
 

RetractableRoof

tolerates intolerance
Lifetime Member
SoSH Member
Dec 1, 2003
3,836
Quincy, MA
Maybe I'm missing something too; what do you mean by "playing the field position game"?

On defense, playing the field position game - i.e., being willing to give up field position when on D in hopes that you lower the expected points given up - seems like the opposite of a BBDB defense. That's moot, however, because that affects the field position of the Patriots' offense.

Do you mean the Patriots' offense plays the field position game? How so?

I think the reason they consistently are around the best in the league at defensive drive starting field position is because they simply have a good offense (and thus are rarely punting from deep on their own side of the field) and have good special teams kick coverage.

Maybe they just play defense (not BBDB), and the discrepancy between their points and yards allowed rankings are an artifact of their good O and ST. What I'm finding so far is that in most seasons their (points allowed - predicted points allowed from yards allowed and defensive starting LOS) is right around where you'd expect, but that five of 16 seasons lie at the envelope of the overall NFL distribution suggests that there is more to it - perhaps a BBDB philosophy, or perhaps something else.

edit: I'll try to look into whether those 5 outliers are also seasons with significant "average lead at start of defensive drives". Not sure what that would mean in terms of yards vs points, but could be interesting.
They have also generally had some of the best starting field position on offense, too (they're third in 2016). I think all this stuff goes together. O produces and doesn't turn the ball over. They kickoff after a score or punt after a failed drive but generally have good coverage. The D has favorable field position to start, might give up a few yards, but then generally forces a punt (which is usually better than a kickoff for field position). The punt return team is good, so the O gets good field position to start ...
Yep. It also helps that the Patriots are typically among the league's elite at not turning the ball over. Turnovers, on average, have a significant impact on field position.
This is where I am. Every coach talks the talk about turnovers. BB seems to walk the walk as well; he takes steps to make sure his team is better at it. His roster decisions factor in ball protection (or lack thereof). Lewis pulls off the 3 TD post season trifecta and his first couple of sentences to the media is pre-emptively talking about his failure to protect the ball. In the wake of deflategate, fumbles were looked at and stories came out about wet football fumble drills, and the extent the Patriots go to practice protecting the ball. Everything about Brady's approach seems geared to not turning the ball over (2 INTs the last game be damned). Safer, surgical approach to going down field at the expense of flashy numbers - unless like the Texans - a team is able to force him to have to try to beat them in a more aggressive manner. Also, as noted by others, his allocation of roster spots and use of starters on special teams to create competitive advantages and contribute to field position.

I don't think there is any doubt that the all three phases of the game are designed to take every opportunity to swing the odds and percentages in their favor and it seems job 1 on offense is to not turn the ball over - which even in a failed drive still often results in a positive for the defense - a long field to defend. I think it is vastly underrated part of the team approach, the offensive approach, special teams approach, etc. in support of the defense. I know other teams try to do it as well - but even as far as allocation of cap resources to make sure that the Patriots aren't star or top heavy via the cap and have enough "middle class" depth in case of injuries or to maintain flexibility in schemes. Rex Ryan for example builds teams that are star heavy, and if they are fortunate with injuries they can often be dangerous, but if they encounter normal attrition, they rarely seem to have the depth to over come it.

Supporting the percentage play or Bend But Don't Break seems the design of the team, not just the defense.

The remaining games should be real interesting. There are 4 teams that have a very high octane offense, led by a QBs who are all capable of pulling a rabbit out of their hat. That should equalize some of the field position advantages contributed by the offenses. Special teams advantages could remain a difference maker - go Lewis and Edelman. And the final piece will obviously be the defenses and how each team is able to slow the other team because I don't see any of these teams really stopped offensively - short of some matchup scheme/advantage I'm not knowledgeable enough to identify. The margins of error will be smaller due to the QB skills on all the teams... and should make for some real stomach wrenching games. And as disconcerting as it sounds to me... games with margins of error this small are susceptible to being decided by a single freak play, like a Kearse catch, or a helmet catch or what have you...
 

SMU_Sox

queer eye for the next pats guy
SoSH Member
Jul 20, 2009
8,945
Dallas
I want a pony.
Sorry, that came off the wrong way, I think. I meant in a hypothetical world where all you have to do is press an easy button and the answer is given to you. Not where you ask someone to do a shit-ton of work and don't volunteer to help at all. I didn't mean to degrade you or your effort at all.
 

BaseballJones

ivanvamp
SoSH Member
Oct 1, 2015
24,714
The Pats' playoff run for the defense:

vs. Houston:
- 16 points (Hou averaged 17.4)
- 302 yards (Hou averaged 314.7)
- 14 first downs (Hou averaged 19.0)

vs. Pittsburgh:
- 17 points (Pit averaged 24.9)
- 368 yards (Pit averaged 372.3)
- 22 first downs (Pit averaged 21.0)

vs. Atlanta
- 28 points* (Atl averaged 33.8)
- 344 yards (Atl averaged 415.8)
- 17 first downs (Atl averaged 23.7)
*The D only allowed 21 points; 6 came on an Atlanta defensive TD

So in every single case (except Pittsburgh's first downs) the Patriots held their opponent in the playoffs to less than their regular season averages.

In the second half of each of these games, the Pats allowed:

- 3 points to Houston
- 8 points to Pittsburgh
- 7 points to Atlanta

That's an average of 6.0 points per game allowed in the second half. Over that time, the Patriots outscored their opponents 61-18 (+43). Amazing.

Consider these first half/second half rushing numbers from the Super Bowl.

First Half: 9 carries, 86 yards (9.6 avg), 1 TD
Second Half: 9 carries, 18 yards (2.0 avg), 0 TD

They completely shut down the Falcons running game in the second half. Beautiful.