2017 Patriots Defense

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Here are the plots updated through week 11, with Patriots 2001-16 in red, 2017 in green.

Per-drive points allowed vs yards allowed:


Pts allowed versus predicted points allowed considering yards allowed and average drive starting L.O.S.:


The residuals from the above plot:


Including turnovers:


They are super bend-but-don't-break-ish this year. It isn't the turnovers - they are 17th in the league at turnovers forced per drive. It's a bit of the starting L.O.S. - they are 1st in the NFL this year and currently have the 12th best average defensive starting LOS in the data set (out of 512 team-seasons; past NE teams also hold 4th, 6th, 8th, 14th, 22nd, and 29th places). However, even after accounting for DLOS and TOs, they have still been über BbDB-ish in 2017 thus far.
 

pokey_reese

Member
SoSH Member
Jun 25, 2008
16,247
Boston, MA
Great plots, thanks for posting those. This year, it's definitely also being affected at least somewhat by the number of missed field goals by opponents, even if you throw out the one the Pats blocked, so that's part of the over-performance. Not sure if the rest has to do with red zone scheming, or if teams are more likely to forego FGs in attempting to get TDs against the Pats because of the offense, or what.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Great plots, thanks for posting those. This year, it's definitely also being affected at least somewhat by the number of missed field goals by opponents, even if you throw out the one the Pats blocked, so that's part of the over-performance. Not sure if the rest has to do with red zone scheming, or if teams are more likely to forego FGs in attempting to get TDs against the Pats because of the offense, or what.
Nice insight. If I find time I'll look the missed FG rate against NE, compared to the league average, and try to work that into the regressions.

There is obviously something that drives the skew of the NE data from the control distribution that I'm not finding, so - in general - if anyone has thoughts on what it could be (especially if it can be encapsulated in data that I can easily obtain) feel free to post or PM me your ideas!
 

( . ) ( . ) and (_!_)

T&A
SoSH Member
Feb 9, 2010
5,302
Providence, RI
Nice insight. If I find time I'll look the missed FG rate against NE, compared to the league average, and try to work that into the regressions.

There is obviously something that drives the skew of the NE data from the control distribution that I'm not finding, so - in general - if anyone has thoughts on what it could be (especially if it can be encapsulated in data that I can easily obtain) feel free to post or PM me your ideas!
I’m curious what these would look like without the KC, Carolina and Houston game. It’s clearly bad form to just throw out their worse games but what I remember most from those games are the 30-50 yard plays and scores given up that were primarily results from coverage mistakes made by secondary. The defense has made several changes since the early part of the season but no longer allowing a receiver to be open downfield with no defender within 30 yards of him may be the biggest change. If I remember correctly Carolina alone had 3-4 big plays on multiple drives and multiple scores resulting from unforced errors.

I’m clearly cherry picking here but I don’t know how else to control for the drop in what looked like communication/coverage issues that allowed guys to rack up yards and points completely uncovered.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Yeah, for an apples-to-apples I'd also have to throw out three cherry-picked games from 511 other team-seasons, which I'm not doing. Were I better equipped to write code to grab the raw game data from the inter web I could then code something to do that for me, but, sadly, I'm not. It certainly could be the case that 2017 NE has a few outlier games that drives this seasons skew, but of course that wouldn't explain the trend. Then again, if many of the NE seasons contain more outliers than typical teams do, that would provide some interesting insight. I'm speculating/rambling now, though...
 

cmurphycode

Member
SoSH Member
Feb 9, 2009
1,305
Nice charts!

1) what does point size mean?
2) can we see the residuals for the turnover adjusted chart?
3) not sure if this is possible with your data source, but an interesting graph might be a correlation between residuals and the score delta at that time.


If you want to analyze how far out of the ordinary 2017 is, then you'd have to, at minimum, break this up into 9-game runs and graph all of those points. In other words, it's still quite possible for 2017's data point to regress towards the season mean. That being said, based on the graphs you've shown, it DOES look like the patriots seasons, in general, are off of mean, so if we just focus on completed seasons, the data is still quite interesting.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Nice charts!

1) what does point size mean?
2) can we see the residuals for the turnover adjusted chart?
3) not sure if this is possible with your data source, but an interesting graph might be a correlation between residuals and the score delta at that time.


If you want to analyze how far out of the ordinary 2017 is, then you'd have to, at minimum, break this up into 9-game runs and graph all of those points. In other words, it's still quite possible for 2017's data point to regress towards the season mean. That being said, based on the graphs you've shown, it DOES look like the patriots seasons, in general, are off of mean, so if we just focus on completed seasons, the data is still quite interesting.
1) Scaled to the average starting defensive drive L.O.S. See here for more explanation.
2) Yes you can. I just made a similar plot yesterday for my own interest. I'll get on that. It is indeed an useful/interesting thing to look at.
3) Great thought; absolutely agree. Though, as you suggest, I'm not sure I have the data right now.
 
Last edited:

BaseballJones

ivanvamp
SoSH Member
Oct 1, 2015
24,375
Update on the defense...

First 4 games (3 home, 1 road):
- 32.0 pts/g
- 456.8 yds/g
- 5 turnovers (1.3/g)

Last 7 games (3 home, 4 road):
- 13.1 pts/g
- 344.7 yds/g
- 11 turnovers (1.6/g)

The last 7 games, they've not allowed a single opponent to score more than 17 points. The past 5 games, they've not allowed a single opponent to gain more than 350 yards. Plus, 7 of Miami's points were scored directly by the defense, meaning the Pats' D has given up really 12.1 points a game over the past 7.

For what it's worth, if these were their season averages...

- 13.1 points a game would rank them #1 in the NFL in scoring defense
- 344.7 yards a game would rank them #19 in the NFL in yardage defense

As they sit right now, even with the awful first four games factored in, they rank #11 in scoring defense, and #30 in yardage defense.

And kind of quietly, the offense has gotten out of its funk. Scoring 41, 33, and 35 points over the last three, for an average of 36.3 points a game.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
They have also quietly started becoming more respectable in DVOA and drive stats. I will wait until DVOA gets updated tomorrow evening, then post some detail on there.

But a couple for now related to the pass D:
24th in completion % allowed (64.0%)
26th in YPA (7.6)
T-11 in INT (10)
T-19 in sack (24)
20th in passer rating against (92.3)

We said all along that if the D could be middle of the pack, they'd have a great chance. The D is getting closer in some advanced metrics.
 

RedOctober3829

Member
SoSH Member
Jul 19, 2005
55,298
deep inside Guido territory
As long as Van Noy and Flowers don't miss any time, this defense is going to continue to get better. They especially can't lose Van Noy. He is so important to this unit post-Hightower injury. I don't think there is anyone behind him who could do what he's been doing.
 

wilked

Member
SoSH Member
Jul 17, 2005
4,044
Stat question - how would you get Points Allowed for partial season (Say Weeks 5-11)? Is that through Pro-Football-Reference? I played around in there and couldn't figure it out
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
Stat question - how would you get Points Allowed for partial season (Say Weeks 5-11)? Is that through Pro-Football-Reference? I played around in there and couldn't figure it out
PFR -> Team Game Finder, search for Cumulative Season Games.

https://www.pro-football-reference.com/play-index/tgl_finder.cgi?request=1&match=single&year_min=2017&year_max=2017&game_type=R&game_num_min=0&game_num_max=99&week_num_min=5&week_num_max=11&temperature_gtlt=lt&c5val=1.0&order_by=points_opp

Pats are #2 behind Jax in that span, though TB and MIA have played one extra game.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
Speaking of, per Football Outsiders the Pats D would have ranked ~10th in the league based on their performance during the six game winning streak. Their week 12 update should be posted tonight.
 

Al Zarilla

Member
SoSH Member
Dec 8, 2005
58,867
San Andreas Fault
Stat question - how would you get Points Allowed for partial season (Say Weeks 5-11)? Is that through Pro-Football-Reference? I played around in there and couldn't figure it out
The other thing is yards allowed split, say, first half vs. second half. I feel a disproportionate amount of yards given up by the Pats are when they’ve got a good lead and go into bend don’t break mode. In the Denver game, the Pats gave up 339 yards, but I think considerably fewer than half of those were in the first half. Same with Oakland and their 344 total offense yards. Miami? 221 yards given up, no parsing necessary.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
The other thing is yards allowed split, say, first half vs. second half. I feel a disproportionate amount of yards given up by the Pats are when they’ve got a good lead and go into bend don’t break mode. In the Denver game, the Pats gave up 339 yards, but I think considerably fewer than half of those were in the first half. Same with Oakland and their 344 total offense yards. Miami? 221 yards given up, no parsing necessary.
In fact, the Patriots gave up more yards in the first half against Denver - 195 on four first-half drives (not including one kneeldown), 136 in four second-half drives (not including one final play). Oakland that was the case - 147 in five first-half drives, 176 in four second-half drives (again not including a final garbage-time play).

As you state, they just straight-up played great against Miami. It wasn't a BBDB performance that hinged on a couple timely plays or the O putting them in positions to succeed or a misleading performance where they faced few drives; they just pretty much shut the Dolphins down. It was probably the cleanest defensive performance all season.
 

Al Zarilla

Member
SoSH Member
Dec 8, 2005
58,867
San Andreas Fault
In fact, the Patriots gave up more yards in the first half against Denver - 195 on four first-half drives (not including one kneeldown), 136 in four second-half drives (not including one final play). Oakland that was the case - 147 in five first-half drives, 176 in four second-half drives (again not including a final garbage-time play).

As you state, they just straight-up played great against Miami. It wasn't a BBDB performance that hinged on a couple timely plays or the O putting them in positions to succeed or a misleading performance where they faced few drives; they just pretty much shut the Dolphins down. It was probably the cleanest defensive performance all season.
Thanks, SN. Where do you get the first half/second half yards given up (or gained) breakouts?
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
For what it's worth (spoiler: not much) I started looking at the play-by-play logs for the 2017 NEPs after the Oakland game and jotted down the yds/pts-allowed, DLOS, and lead/deficit at start of the drive. I only entered the first four games and the Oakland game, but there was a weak correlation with yds/DLOS-adjusted BbDB-ness and lead. Again, the correlation was weak, noisy, and not even very statistically significant, but that could change with more data. I gave up mostly because for any context I'd have to do the same for the rest of the league and I'm not going to put in the effort to catalogue all that data. Maybe I'll email F.O. and ask if they/Jim Armstrong would be willing to share their data.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
DVOA has been updated. Defense is up to #29 at 13.9%. SF is #27 at 12.6%, then Atlanta #26 at 9.9%. Pass defense is up to 25th at 22.3%. Plenty of time for the D to get to middle of the pack.

One thing I do NOT understand at all is how the Saints passed the Pats in offense this week. The Pats offense was much better than the Saints, albeit against a worse D.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
9th in PPG allowed
18th in opponent completion %
20th in opponent YPA
T-9 in INT
17th in opponent passer rating

Solidly middle of the pack. I even expect some DVOA improvement this week, ha.

They are going to finish top 6 or better in points allowed.
 

BaseballJones

ivanvamp
SoSH Member
Oct 1, 2015
24,375
Updating the numbers...

First 4 games (3 home, 1 road):
- 32.0 pts/g
- 456.8 yds/g
- 24.3 first downs/g
- 5 turnovers (1.3/g)

Last 8 games (3 home, 5 road):
- 11.9 pts/g
- 335.0 yds/g
- 19.4 first downs/g
- 12 turnovers (1.5/g)

Or to break it down by quarters of the season...

First 4 games (3 home, 1 road):
- 32.0 pts/g
- 456.8 yds/g
- 24.3 first downs/g
- 5 turnovers (1.3/g)

Second 4 games (2 home, 2 road):
- 12.8 pts/g
- 377.3 yds/g
- 21.5 first downs/g
- 4 turnovers (1.0/g)

Third 4 games (1 home, 3 road):
- 11.0 pts/g
- 293.0 yds/g
- 17.3 first downs/g
- 8 turnovers (2.0/g)
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA

Aaron Schatz: A couple weeks ago I wrote that the #Patriots had still not had an above-average defensive game according to DVOA. They now have 4. They've had their 2 best defensive games in Weeks 12-13, plus opponent adjustments changed slightly for NYJ and OAK.
 

eustis22

New Member
Nov 14, 2016
998
So, Eric Lee...a baller or no? Does he take Roberts spot when Van Noy is back to full health?
 

SMU_Sox

queer eye for the next pats guy
SoSH Member
Jul 20, 2009
8,878
Dallas
A problem with DVOA this year more than others is that this year there have been more injuries to key players during the year that have drastically impacted opponent adjustments. Schatz talked about it on the overall ratings last week. It’s an imperfect stat. My gut tells me when they are all healthy together (of guys not on IR) they are somewhere around an average unit or at least will be by the end of the season.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
I didn't see any of the game. Were the Pats playing to defend the pass and sort of daring the Bills to try and beat them with the run? I know they've done this in the past, one game against the Peyton Manning Broncos in Foxboro in particular.
Eh. Bills ran all over them first drive. They allowed several long wildcat runs. Taylor had a long scramble too. The total rushing yardage is a bit skewed since the Pats did seem content to let them run it at will once it got to 23-3 but not matter how you slice it they got gashed up pretty good.
 

Toe Nash

Member
SoSH Member
Jul 28, 2005
5,591
02130
It's almost like run defense barely matters in today's NFL especially if you have a team with a strong offense who can take the lead and make opponents play from behind.

It may have been different if the Bills scored on their first drive, but come on. The defense allowed 115 yards passing and after the first drive were pretty stifling. It was a good performance.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
It's almost like run defense barely matters in today's NFL especially if you have a team with a strong offense who can take the lead and make opponents play from behind.

It may have been different if the Bills scored on their first drive, but come on. The defense allowed 115 yards passing and after the first drive were pretty stifling. It was a good performance.
I didn't argue either of these points. I stand by the run defense being borderline awful on Sunday.
 

Bowhemian

Member
SoSH Member
Nov 10, 2015
5,698
Bow, NH
I didn't argue either of these points. I stand by the run defense being borderline awful on Sunday.
Agree with that. It was bad, and the wrapping/tackling was horrendous. Numerous times they made contact with McCoy in the backfield and failed to stop him, instead allowing plentiful gains. Ugly.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
Also, I know Buffalo didn't score on the opening drive, but that game was exactly how you beat Brady from an offensive perspective. Control the clock and limit possessions. The Pats only had 4 possessions in the first half. Now Buffalo's defense didn't play close to good enough to win that game, but from an offensive perspective if you can run for close to 200 yards and control the clock you give yourself a chance.
 

Toe Nash

Member
SoSH Member
Jul 28, 2005
5,591
02130
Also, I know Buffalo didn't score on the opening drive, but that game was exactly how you beat Brady from an offensive perspective. Control the clock and limit possessions. The Pats only had 4 possessions in the first half. Now Buffalo's defense didn't play close to good enough to win that game, but from an offensive perspective if you can run for close to 200 yards and control the clock you give yourself a chance.
This has always been true. If you're the underdog you limit possessions and hope that in the small sample size there's enough variance that you end up with a chance at the end.

But which teams can pull that off? If the Pats run D was/is bad...who cares? As long as the offense /special teams doesn't give them a short field, they're still going to need 10+ plays to score even if they're pulling off big gains on the ground. And so you have a good shot at them getting a penalty or a loss and stopping the drive.

The problem at the beginning of the year was that they were allowing 300 yard passing games and 20+ yard plays with regularity. That's no longer a problem, so the defense looks really good. We can quibble about the run D but that was never a strength of the team nor does it need to be.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,054
Hingham, MA
Well Jacksonville comes to mind. If they get Fournette rolling and their D can play like it is capable of it could have the makings of a playoff upset.
 

DJnVa

Dorito Dawg
SoSH Member
Dec 16, 2010
53,838
So we always hear that BB takes away the other team's #1 threat. Could there be a reason he deemed that to the Bills passing game?
 

pokey_reese

Member
SoSH Member
Jun 25, 2008
16,247
Boston, MA
So we always hear that BB takes away the other team's #1 threat. Could there be a reason he deemed that to the Bills passing game?
It could be that at this point, he always just considers the passing game to be the other team's #1 threat. If teams usually get 250 yards in the air, and 100 yards on the ground, and 70% of plays of 20+ yards are pass plays, he might just be willing to live with giving up 5 ypc instead of 4 ypc, especially given that the personnel he has on defense is stronger in the secondary.
 

Van Everyman

Member
SoSH Member
Apr 30, 2009
26,993
Newton
I actually would agree Yes, Lee had some trouble setting the edge -- they haven't really found a replacement for Nink yet (and Hightower's absence isn't helping). And Roberts got criticized in the game thread for trying to bull rush Taylor a few times when he would've been better showing some patience underneath. Even still, I'm not losing a ton of sleep about that performance -- or even the first drive. McCoy did escape some backfield tackles but he's also one of the best RBs in the league. The guy is going to put up some numbers against most teams -- and if BB thinks his team can stop a team like Buffalo on third and short, I'm inclined to trust him.
 

nothumb

Member
SoSH Member
Jul 27, 2006
7,065
yammer's favorite poster
It could be that at this point, he always just considers the passing game to be the other team's #1 threat. If teams usually get 250 yards in the air, and 100 yards on the ground, and 70% of plays of 20+ yards are pass plays, he might just be willing to live with giving up 5 ypc instead of 4 ypc, especially given that the personnel he has on defense is stronger in the secondary.
It could be that he thinks specifically the intermediate passing game is what makes BUF tick; if so, DVOA would seem to agree. FO considers their passing attack just a bit below average, their rushing attack worse than that. Seemed like they were content to cover the inside routes with extra defenders and match up on the outside. Lots of LBs and DBs in intermediate zones in the middle of the field will leave some running lanes but take away the easy completions and reads that Taylor needs.
 

johnmd20

mad dog
Lifetime Member
SoSH Member
Dec 30, 2003
61,996
New York City
Well Jacksonville comes to mind. If they get Fournette rolling and their D can play like it is capable of it could have the makings of a playoff upset.
Indeed. Jacksonville's D can win a game on the road in the playoffs. They are aggressive, fast, opportunistic, and relentless. Bortles is not a good QB, but the defense could overcome that with a couple of lucky plays/breaks. If the Pats do meet them in the playoffs, it won't be easy. But I would love to see that matchup.