Writers of Pro Football Prospectus 2008

Most Recent FO Features

MelvinRas14.jpg

» Four Downs: AFC South

The Jaguars and Titans both won playoff games last year. Are the days of this division being the NFL's laughingstock over? Maybe not -- all four clubs failed to address critical weaknesses in the draft.

30 Jan 2018

QR Bonus: Keys To Beating Philadelphia

by Vincent Verhei

Last week, we looked at the worst games New England played this year, searching for weaknesses that might cost the Patriots another Lombardi Trophy. Today, we're going to do the same for New England's Super Bowl opponents, the Philadelphia Eagles.

By DVOA, Philadelphia's worst games this season were:

  • Week 2: Philadelphia Eagles 20 at Kansas City Chiefs 27. What a strange year it was for Kansas City, beating the eventual AFC and NFC champions in Weeks 1 and 2 and then going just 8-7 (including the playoffs) the rest of the way. This game was not as close as the final score would indicate; Kansas City scored two touchdowns in a little more than four minutes of game time to take a 27-13 lead in the fourth quarter before Philadelphia scored a final touchdown with just eight seconds left on the clock. (They actually recovered an onside kick, but Carson Wentz's pass on the last play of the game was incomplete.) The Chiefs got an efficient day out of Alex Smith (21-of-28, 251 yards, one touchdown, no interceptions) and big numbers from Kareem Hunt (13 carries for 81 yards and two scores; three catches for 28 yards) and Travis Kelce (eight catches for 103 yards and a touchdown). Wentz led the Eagles with 55 rushing yards. Philadelphia's top running back that day, Darren Sproles, hasn't played since tearing his ACL and breaking his arm against the Giants in Week 3.
  •      

  • Week 14: Philadelphia Eagles 43 at Los Angeles Rams 35. It was thought to be the emptiest playoff-clinching victory of all time, as Wentz's torn ACL seemed certain to end the Eagles shot at a Super Bowl win. That theory proved to be overly pessimistic, but DVOA was still surprisingly unimpressed by this road win over a very good Rams team. The Eagles trailed 35-31 early in the fourth quarter, but added a pair of Jake Elliott field goals and then a fumble-return touchdown by Brandon Graham on the game's final play to ice the win. There were three fumbles in the game (one by Philadelphia, two by Los Angeles), and the Eagles recovered all of them. The Rams' Todd Gurley made a big impact on a relatively small amount of touches: 13 carries for 96 yards and two touchdowns, plus three catches for 39 more yards.
  • Week 15: Philadelphia Eagles 34 at New York Giants 29. A lot happened in this game. The Giants scored touchdowns on each of their first three drives to jump out to a 20-7 lead. Philly rallied to go up 21-20, and then Aldrick Rosas put New York back in front 23-21 -- and that was just the first half. The game was close throughout the second half, realistically ending when Eli Manning's pass to Evan Engram on fourth-and-goal from the 11 fell incomplete with 48 seconds to go. Manning finished with 57 attempts, 37 completions, 434 yards, and three touchdowns, each of which was his highest total of 2017. His leading target, Sterling Shepard, finished with 11 catches for 139 yards and a score.
  • Week 17: Philadelphia Eagles 0 at Dallas Cowboys 6. With home-field advantage in the NFC already sewn up, the Eagles had nothing to play for here, and it showed. Nick Foles threw only 11 passes before yielding to Nate Sudfeld. Thirteen different Eagles had at least one target; none gained more than 25 yards receiving. The defense had a much better day, forcing punts on seven straight drives at one point, but then surrendered a 99-yard touchdown drive and was fortunate when Dan Bailey missed a 23-yard field goal in the fourth quarter. Ezekiel Elliott was the offensive star for Dallas, rushing 27 times for 103 yards and catching three passes for 38.

Some readers might be wondering if it's fair to include that Week 17 game. The Eagles' next worst game was a 19-10 home win over the Raiders on Christmas night in Week 16. Philadelphia trailed 10-7 in the second half before rallying with a pair of Jake Elliott field goals and a fumble return touchdown on the last play of the game. (Yes, that actually happened to Philadelphia twice in three games.) Including that game is problematic, though, because it's not similar to the Eagles' other bad games. In the Kansas City, Los Angeles, New York, and Dallas games, Philadelphia's pass defense DVOA was never better than 5.7%; against Oakland, it was -107.9%, easily the best day they had all year. If we're looking for ways to beat the Eagles, the only lesson from this game would be "hope their quarterback has an even bigger meltdown than your quarterback has," which doesn't seem to be helpful advice for Tom Brady and the New England Patriots. So we will stick with the Dallas game, but we will also make sure to highlight that Nick Foles had a very, very bad game against the Raiders, going 19-of-38 for 163 yards with one touchdown, one interception, two sacks, and a fumble.

That said, he wasn't much better in our four-game set, going 34-of-59 (57.6 percent) for 318 yards (5.4 yards per pass) with four touchdowns, one interception, and two sacks. That's a small sample size -- two games, essentially -- but it does show that the Eagles' worst games usually included a poor performance from their quarterback. (This is true for most teams, of course.) Defenses generally gave Philadelphia problems by taking away the wide receivers. Alshon Jeffery, Nelson Agholor, and Torrey Smith each had 20-plus targets in our four-game sample, but none had a positive DVOA. Jeffery and Smith were the deep threats; the average target to each came more than 14 yards downfield, but neither had a catch rate of even 50 percent. Agholor's usage was much different. He caught 73 percent of his targets, but those targets only traveled 6.2 yards past the line of scrimmage on average. The only other Eagles player with 20-plus targets in these games was tight end Zach Ertz, and he produced with a 19.0% DVOA, but that didn't do the Eagles much good. New England's top three cornerbacks -- Malcolm Butler, Stephon Gilmore, and Jonathan Jones -- each tended to cover receivers on deeper routes this year. However, former Eagles corner Eric Rowe's average target came just 9.2 yards downfield, so Agholor might be his personal responsibility. (Rowe missed eight games in the regular season but has essentially replaced the injured Jones in the playoffs, with five tackles against Tennessee and four more against Jacksonville.) If New England can take away the deep routes of Jeffery and Smith and make their tackles on Agholor, they should be able to stifle the Philadelphia air attack.

     

They should also be able to stifle the Philadelphia ground attack. In their four worst games, the Eagles ran 94 times for 424 yards, an average of 106.0 yards per game and 4.5 yards per carry. Those numbers look awfully good, but they are boosted by the performance of Wentz (seven carries for 71 yards) and Sudfeld (22 yards on his one carry against Dallas). Take out the first- and third-string quarterbacks, and you're left with 86 carries for 331 yards -- 82.8 yards per game, 3.8 yards per run. Jay Ajayi, Corey Clement, and LeGarrette Blount -- the top trio in Philadelphia these days -- each had 11 carries or more, but each finished with negative DYAR and a success rate of 45 percent or worse. One would think the Eagles would have an edge against New England's No. 30 DVOA run defense, but Philadelphia's running game has hardly been dominant this season.

The ground game was especially useless in the red zone, averaging 2.4 yards on 15 carries with no touchdowns and a DVOA of -68.1%. The red zone passing attack was much more successful, averaging 4.7 yards and scoring ten times in 33 plays, with a DVOA of 118.0%. Obviously, a lot of that is Wentz. But even if we only look at Weeks 15-20, including the playoffs, the Eagles have 49.1% passing DVOA and -51.8% rushing DVOA in the red zone. When the Eagles reach the red zone in the Super Bowl, they should count on Foles to win or lose the game for them, not Ajayi or Blount.

While the Philadelphia offense had plenty of issues, things were even worse on the other side of the ball. The Eagles average defensive DVOA in these four games was 11.1%, and the average run defense DVOA was 3.3%. Both figures would have been worst in the league this year. The average pass defense DVOA was 17.5%, which would have been in the bottom ten. Alex Smith, Jared Goff, Eli Manning, and Dak Prescott combined to go 91-of-141 against the Eagles (64.5 percent) for 1,063 yards (7.5 per pass) with seven touchdowns, one interception, and eight sacks. Seven touchdown passes isn't a ton in four games, but it's more than enough considering the Eagles also gave up five rushing touchdowns. Those scores on the ground came among 91 carries for 425 yards -- 106.3 yards per game, 4.7 yards per carry.

There was no real secret about the players who had success running against Philadelphia -- each team's top back had a big day. By sheer volume, Elliott had the best performance, but at 3.8 yards per carry and a 37 percent success rate, it was hardly the most efficient. Gurley (7.4-yard average, 77 percent success rate), Hunt (6.2, 46 percent), and New York's Wayne Gallman (4.9, 63 percent) were all better by DYAR. There was no significant difference in runs by direction either. So it's not clear precisely why the Eagles struggled to stop the run in these contests, but it is clear that they were vulnerable to such plays. That might lure the normally pass-heavy Patriots into more of a ground-based attack.

There might be more to be learned in looking at the receivers who had big days against Philadelphia. We mentioned Shepard's 139 yards already; he was followed by L.A.'s Cooper Kupp (118 yards), Kelce (103), and Engram (87). That's two tight ends in the top four, which is an extremely bad omen for a team that is preparing to play against Rob Gronkowski. As far as DYAR goes, the top five wide receivers include Kupp and Shepard, but also third receivers like New York's Tavarres King (two catches for 70 yards and two touchdowns) and Dallas' Brice Butler (two catches for 50 yards and a touchdown). There might be room here for someone on the bottom of New England's wide receiver totem pole (Phillip Dorsett? Kenny Britt?) to come through with a big play. More likely, we're talking here about Danny "Playoffs" Amendola.

Finally there's the kicking game, which also hurt the Eagles at times this year. Jake Elliott missed a 30-yard field goal against Kansas City, and the Philadelphia coverage teams surrendered a 40-yard kickoff return to Tyreek Hill. The Eagles forced five punts against the Rams, but Kenjon Barner managed only 11 total return yards on those five punts.

We mentioned last week that Philly's strengths didn't necessarily match up with New England's weaknesses, but the Patriots are a more versatile team. They have gone run-heavy at times this season, with 190-plus rushing yards against Buffalo (twice) and Miami. They will use all their weapons; witness Dorsett's three catches for 68 yards against New Orleans. And Bill Belichick has always specialized in taking away what an opponent does best. He'll likely be content to limit the Eagles' big-play passing attack and make the Eagles drive the field to get into scoring range, then clamp down on Foles in the red zone. It's all just further reasoning that the Eagles are again underdogs in this postseason -- but then, that's a role that Philadelphia has both embraced and overcome.

Posted by: Vincent Verhei on 30 Jan 2018

26 comments, Last at 21 Feb 2018, 8:15pm by Obat Pembesar Penis

Comments

1
by Pat :: Tue, 01/30/2018 - 1:57pm

There's also something else in common between 3 of those games: most of the opponents' scoring drives were pretty short.

Why is that? Because of the 50+ yard TDs the Eagles gave up all year (by yards: 77, 75, 67, 63, 57, 53) - 4 of those 6 were in those 3 games. The other two? Chargers and Raiders. The Cowboys game was just weird.

To me, it's pretty clear what that says: you really want to hit on a few home run shots, and there probably will be opportunities - which isn't surprising on an aggressive defense.

4
by Aaron Brooks Go... :: Tue, 01/30/2018 - 5:37pm

Shepard's 77 yarder came in the week 3 NYG game, not the week 15 game.

6
by Pat :: Tue, 01/30/2018 - 6:26pm

Ha, that's what I had first, and then corrected it because I saw NYG there. Oops.

In any case, it's pretty clear there's a benefit to being aggressive against Philly, which is partly why I tend to think this will be NE's first Super Bowl blowout. Philly saw last week what happens when you have success being aggressive against an aggressive defense, and Foles is fully capable of falling apart if they have to press.

2
by Anon Ymous :: Tue, 01/30/2018 - 3:22pm

The Patriots won't necessarily need a big game from Tom Brady to win the Super Bowl

Given Philly's outstanding run defense and NE's propensity to forget how to run the ball in the playoffs, I doubt NE wins this game without a terrific Brady performance.

11
by DavidL :: Wed, 01/31/2018 - 10:13am

Are we talking terrific in terms of DVOA or terrific compared to other Tom Brady playoff games?

12
by Anon Ymous :: Wed, 01/31/2018 - 10:14am

I was just talking raw numbers. 300+ yards, 65+% comp, 3 TDs, no more than 1 pick.... that kind of thing.

3
by nat :: Tue, 01/30/2018 - 4:42pm

Personally, I'd rather look at the Eagles' four worst games by VOA instead of DVOA. As a Patriots fan, I'd be perfectly happy to see my team learn from how a good team beats the Eagles or comes close. DVOA prefers weaker teams that overachieve.

I assume that Eagles fans feel the same way about the keys to beating the Patriots. The two teams are too close in weighted DVOA to make preferentially copying weaker teams a good plan.

It might not change the lists of games much. Then again, it might.

Okay. I'll make that an actual question for Vince:

If you went by VOA, which games would drop from the list? What games would be added?

7
by Eddo :: Tue, 01/30/2018 - 6:26pm

Agreed. To beat a team, in general, you want them to get the least absolute value out of their plays, regardless of how good you are. The Giants could play a close game vs. the Eagles with a high-variance game plan, but that wouldn't be a good strategy for the Patriots to use (NOTE: not saying this is what the Giants did, just giving a hypothetical).

And if a team's worst VOA games come against other good teams, so that their DVOA is still respectable, well, that tells you something: you'd better be a good team to beat them!

8
by Roadspike :: Tue, 01/30/2018 - 10:51pm

I would guess that the Seahawks game would come in if you went by VOA, but I'm not sure.

14
by nat :: Wed, 01/31/2018 - 10:52am

I just remembered that Vince and I had a discussion about this topic last year.

https://www.footballoutsiders.com/quick-reads/2017/qr-bonus-keys-beating...

Vince's main point was that using VOA would lead to "be a good team or play like one" being the highlighted "strategy", which is not very enlightening.

My point was that "study the successful schemes and match ups that successful teams used" was the correct start for a strategy, even if some of those teams were good (by DVOA) and thus were expected to have good schemes and match ups in most games.

There's validity to both positions. In the end, I'm less interested in the bad teams that surprised us but still weren't great than I am in the good teams that were more successful but not a surprise. For this purpose, anyway.

21
by Vincent Verhei :: Wed, 01/31/2018 - 6:16pm

Ignoring special teams, because calculating that is a pain in the ass, the Eagles worst four games by VOA were:

* Week 17 vs. DAL
* Week 14 vs. LARM
* Week 2 vs. KC
* Week 13 vs. SEA

If you wanted to remove the DAL game, you would add Week 4 vs. LACH.

22
by nat :: Wed, 01/31/2018 - 6:17pm

Awesome. Thanks.

5
by Aaron Brooks Go... :: Tue, 01/30/2018 - 5:48pm

That week 17 Dallas game was meaningless, and the Eagles played like it was meaningless. Their sole goal for the game was to keep Elliott below 1000 yards on the season. I don't think it tells you anything about beating the Eagles.

It would be more interesting to use week 3 against the Giants, week 4 against the Chargers, or week 13 against the Seahawks.

They ran the ball all over the Giants and Chargers, and the Seahawks game was mostly decided by turnovers, so I see how that undermines some of the analysis above and/or has a trivial answer, but week 17 offers nothing of value to the analysis.

Philly's rush offense is weird. They ran it all over the poor rush-D of the Chargers and the Giants, but struggled with the poor rush D of Dallas, Washington, and KC. But they obliterated the great rush D of Denver and ran well against Minnesota and Arizona.

Much like how NE is as simple as their offense being better than their defense is bad, Philly is predicted by turnovers. If they lose the turnover battle, they are 2-3. If they win or tie, they are 13-0.

9
by Hextall_27 :: Tue, 01/30/2018 - 11:54pm

The Patriots have the best offense in football, but they have to face a very good defense.

The Eagles have a very good offense, but have to face a much improved...

Has anyone compared the splits around the bye on the Patriots season? They played all but 1 of the strong offensive teams before the bye. The schedule is not some big secret. Just take a look.

Could the vast improvement be 20% scheme/fit/coaching and 80% they just played some really poor teams after the bye? Maybe its 30/70 or 40/60? Does anyone really think that the Jets, Bills x2, Dolphins x2, Raiders, and Broncos were offensive juggernauts that were shut down by a mighty defense? The Steelers had 413 yards and missed out on 31 points with a goal line dribble and an epic meltdown.

Those offenses were ranked 32, 29, 29, 25, 25, 22, 19, and 2 in weighted DVOA. (as of week 19) The offenses they faced in the playoffs were ranked 18 and 20.

Is it at least possible that this Patriots D is not very good when they have to play a good offense?

10
by Anon Ymous :: Wed, 01/31/2018 - 9:59am

Could the vast improvement be 20% scheme/fit/coaching and 80% they just played some really poor teams after the bye?

No. A good argument could be made that points per game is inflated, but not DVOA. NE's defensive improvement is legitimate and has nothing to do with the schedule. It's because they finally stopped blowing assignments and leaving receivers so uncovered that they could call for fair catches on their receptions.

EDIT: Go to the "Keys to Beating NE" post and scroll down to the thread starting at comment #14 for more on this topic.

15
by Pat :: Wed, 01/31/2018 - 12:08pm

"No. A good argument could be made that points per game is inflated, but not DVOA. "

That's the argument he's making, though. Rather than the bye, the obvious breakpoint looks to be about week 5. If you just look at weeks 1-4 versus the rest of the year, they went from 32 ppg to 14.4 ppg. By "DVOA expected by points" they went from +45% to -34%: that would be going from a godawful defense to a great defense.

Except by DVOA, they *didn't* do that. They went from (roughly) 30% defensive DVOA to 4% defensive DVOA. So going from a godawful defense to a "not good" defense.

In fact, I'll revise his numbers: their apparent (allowed points per game) improvement was ~80% (about 17.5 ppg). Of that 17.5 ppg, about 32% was true improvement, and about 68% was the difference in quality of offenses between weeks 1-4 and weeks 5-17.

17
by Anon Ymous :: Wed, 01/31/2018 - 12:52pm

Where in his post does he specify a focus on points allowed?

FWIW, DVOA is notoriously unimpressed by NE's defense even when there aren't such obvious opponent factors, which muddies simplistic analyses like the breakdown at the end of your post. Even if I accept it, though, there seems to be something awry with your numbers. Both NYJ and Det were exactly at 4% defensive DVOA, and both were in the 23.5 ppg range. So, it would be more accurate to say that roughly half of NE's drop, or ~9 ppg, was legitimate and the remainder was artificial.

18
by Pat :: Wed, 01/31/2018 - 2:17pm

Going from 30% DVOA to 4% DVOA isn't really a vast improvement. The only vast improvement you see in NE's defense is in points allowed per game.

But let me clarify a bit: it's an exaggeration to say that 70% of the improvement was due to poor offenses - there are other effects going on. But it's not an exaggeration to say that 70% of the improvement was artificial.

"Both NYJ and Det were exactly at 4% defensive DVOA, and both were in the 23.5 ppg range. So, it would be more accurate to say that roughly half of NE's drop, or ~9 ppg, was legitimate and the remainder was artificial."

I think you're misunderstanding how DVOA works: defenses don't allow points, *teams* allow points. DVOA separates out the defense/offense/special teams contributions. A DVOA of 0% means your defense gives up as many points as league average, if you played league average teams and everything else equalled out. This tends to mean that 0% DVOA = ~22 ppg (league average was 21.7 ppg this year, so a little lower).

However, your *actual* points per game depends on the quality of the offenses you faced, along with other factors - defenses aren't solely responsible for preventing scoring, for instance. Good offenses => less opponent scoring because opponents start in worse field position. Opponents score less field goals than you'd expect. Turnovers. Etc.

A 4% defensive DVOA with Detroit and the Jets would give up way more points than a 4% defensive DVOA with the Patriots, because the Jets and Detroit have far worse offenses, and so opponents start in better field position. (Plus possibly special teams a bit, but dominantly offense).

Patriots opponent starting field position: 24.6
Jets opponent starting field position: 28.0
Detroit opponent starting field position: 27.6

Assuming ~12 drives/game, that's around 3-3.5 yards/drive, good for around 2.9-3.4 ppg. So a 4% defensive DVOA with the Patriots would be expected to allow 20.1-20.6 ppg, not 23.5 (due to the better offense). This is super-fudgy because I'm not splitting up starting field position by week, however, the correction only gets *stronger* because in the first 4 games, the opponents had better starting field position than the average.

In addition, as was noted, the Patriots faced a weak slate of opposing offenses from weeks 5+: on average, -3.13%, which now puts the Patriots at an expected 17.9-18.4 ppg, not 23.5 ppg (due to the weak opposing offenses).

Finally, the Patriots also benefited from poor opposing field goal kicking. In weeks 1-4, opponents went 8/9. In weeks 5+, they went 14/22. (On the whole season, they had the second-worst opponent field goal kicking percentage in the league). By league average, opponents should've scored 1.1 ppg more. So now we're down to 16.8-17.3 ppg. Obviously there's some fudge here because the starting field position means longer field goals, etc. etc.

But you can see how a 4% defensive DVOA can still end up with a low points per game: because it doesn't all come from defense.

---

But, back to your question.

How did I figure out the 70%/30%? Easy. The Patriots allowed 32 ppg from weeks 1-4. They allowed 14.4 ppg from weeks 5+. That's a change from +47% to -34%, relative to league average. Or, said another way, a percentage change of 81%. DVOA, however, has them going from +30% to +4%, or a percentage change of 26%. 26/81 = 32%, so 32% of the change is 'real', and 68% of it is artificial.

OK, so suppose you don't like DVOA. You see a similar thing in expected points. From weeks 1-4, Patriots defense was -16.65 expected points/game. More negative = bad. This means defensive plays *reduced* the scoring margin of Patriots games by 16.65 points. In other words: first 4 games, Patriots scored 32.25 points, and allowed 32 points, a margin of +0.25. Absent the defense, that should've been +16.9.

From weeks 5+, Patriots defense was -0.885 expected points/game: in other words, the defense was still *hurting* the scoring margin in general.

By points per game, they went from godawful (32 ppg) to one of the better defenses in the league (14.4 ppg). But looking at the actual results of the play of the defense, either by DVOA or by expected points, they went from godawful to still-not-good. That's the entire point. The majority of the improvement is artificial.

19
by Anon Ymous :: Wed, 01/31/2018 - 2:38pm

I think you're misunderstanding how DVOA works

Not at all. Frankly, I'm confused how you could even make that inference. Amusingly it's understanding the distinction between points and DVOA or relative "goodness" of a unit that prompted my response in the first place!

Going from 30% DVOA to 4% DVOA isn't really a vast improvement.

It isn't? The difference between the best and worst defenses this year is 27.8%.

If we are going to focus on points, then allow me to supplement this with some data. Starting at the bye week and excluding the Jet game since they didn't have an NFL quarterback in that game, the following is NE's points allowed relative to points scored for those opponents: -2, -10, 0, -16, 10, 6, -3, -7, -6, -7

A few notes:

* I agree the better demarcation line is week 5, but I split it at the bye to be conservative (the figures look better for NE the couple weeks prior).

* Five of the first six games listed above were road games, with Miami and Pitt being the last of that leg.

* I gave Pitt the James TD in the figures above.

* Only the first two games above could be considered "over at half-time". The rest were competitive into the 3rd quarter, at least.

Draw whatever conclusion about causation you want, but NE consistently held teams to below their season average. The only real exception was in Miami, where the offense was pitiful and NE always seems to struggle. Whether you credit that to the offense or special teams, something about the team dynamic allows NE to limit points as if they were above average while appearing to be below average.

20
by Pat :: Wed, 01/31/2018 - 3:43pm

"Not at all. Frankly, I'm confused how you could even make that inference. "

Because you thought 4% DVOA with a 27% offense implies 23.5 ppg.

"Going from 30% DVOA to 4% DVOA isn't really a vast improvement.

It isn't? The difference between the best and worst defenses this year is 27.8%."

Going from 30% to 4% isn't nearly as significant as going from allowing 47% more points to allowing 34% *fewer* points. We just have a different definition of vast: if you look at points per game, they went from one of the worst ever to one of the best ever, and that's *not* what really happened.

"Draw whatever conclusion about causation you want, but NE consistently held teams to below their season average. "

Again: good offense + special teams allow fewer points than average. Field position matters, game situation matters. Teams have longer to go, they score less points. They get fewer turnovers, they score less points. They're down more points, they take bigger risks.

23
by Anon Ymous :: Wed, 01/31/2018 - 6:22pm

Because you thought 4% DVOA with a 27% offense implies 23.5 ppg.

No, I gave two examples of teams with that DVOA in the same range - one of which was a team with a better-than-average offense and an excellent special teams - whose points allowed indicate that your figures are off. You can offer supplemental data or alternate hypotheses, but what you said before was, at best, incomplete.*

Going from 30% to 4% isn't nearly as significant as going from allowing 47% more points to allowing 34% *fewer* points.

When did I said it did? Did you miss the part where I said a 50/50 split was more apt?

We just have a different definition of vast

Are you suggesting that a change that accommodates the entire span of NFL defenses this year isn't vast?

Again: good offense + special teams allow fewer points than average.

Again: nothing I've said indicates that this is lost on me. You can consider that point understood and stop repeating it now.

It bears noting that NE's offense wasn't particularly awe inspiring during the stretch of games mentioned in my prior comment, so the effect there would be minimal. The question also remains why the same offense and the same special teams didn't do much to slow down opposing offenses in September. Sure, the offenses were better, but NE didn't just allow more points, they allowed more points than the opposing offense's season averages. Aside from NO, who hadn't hit their stride yet, the relative totals were +16, +10 and +6 (compared to Houston's Watson games).

* ETA: KC and Dallas might be better comps for NE, each with solid O/ST and poor Ds, and each at ~21ppg allowed. This, of course, is concordant with other areas helping the D minimize points, and the two point difference superficially supports your 65/35 suggestion. The fact remains, though, that the O and ST for NE were there all along, so the defense didn't get more assistance from those units in December than it was getting in September. Since we're comparing NE to NE, that factor is already normalized in the samples. (And, if anything, NE was actually better on offense when the D was struggling, but I'm willing to overlook that for the sake of this discussion).

24
by Pat :: Thu, 02/01/2018 - 12:20pm

"No, I gave two examples of teams with that DVOA in the same range"

Yes, and then you used their points per game to suggest what range New England's should've been. That's why I said you're misunderstanding something. DVOA linearly correlates to points. 0% DVOA = league average points allowed/scored. You don't use individual teams to compare DVOA to points, because they've got too much noise. You use league average.

Even in that case, though, you then made a false comparison, where you compared the observed week 1-4 to the expected week 5+, and that's apples to oranges. In fact, DVOA didn't think that they were *that* bad during Weeks 1-4, which is where the difference comes in. 30% DVOA is ~28.2 ppg, not 32 ppg.

If you call the first time period A, and the second B, you did (OA-EB)/(OA-OB), whereas you need to do (EA-EB)/(OA-OB).

"When did I said it did? Did you miss the part where I said a 50/50 split was more apt?"

Going from 30% to 4% is a 26% change. Going from +47% to -34% is an 81% change. DVOA is directly related to points, so these are on the same scale. Therefore the DVOA change is 32% of the point per game change. That's all I'm saying. There's nothing funky or weird about that math. It's really simple.

The fact that we're literally arguing about the difference between 32/68 and 50/50 is honestly a special kind of silly. 32/68 is what DVOA says it is. Just math. If you want to adjust it in your head to 50/50 based on what you've seen, go ahead.

"Are you suggesting that a change that accommodates the entire span of NFL defenses this year isn't vast?"

Yes, because that ~25-30% span of defense values for this year is a total outlier. There aren't any super-bad or super-good defenses this year. Normally there are. Last year the full span was ~37%. The year before it was 51.9%. The year before that, it was 32.5%. Before that, 43.4%. In fact, as Aaron noted: "More than anywhere else, the parity of the 2017 season is seen in the defenses, with the smallest-ever gap between the best defense (Jacksonville at -16.1%) and the worst defense (Tampa Bay at -11.7%)."

25% changes from early on->late happen, just about every other year or so. Two years ago Philly's defense went from "pretty good" (-14%) early on to "bad" (+11%) by the end of the year. However, 80% swings in points per game *don't* usually happen. That same Philly team went from 21.5 ppg (-6%) to 32.5 ppg (+43%). That's ~50%, which is big, and still much bigger than the DVOA swing - but *much* smaller than the 80% swing that New England had.

Like I said, it's probably just a different definition of 'vast'. To me, if you look at just points allowed, it looks like New England just had one of the largest turnarounds of all time. But they didn't - it's a mirage, borne of poor opposing offenses, a great offense, and great special teams.

"Sure, the offenses were better, but NE didn't just allow more points, they allowed more points than the opposing offense's season averages"

Because the defense was worse! A +30% offense isn't going to compensate for a +30% defense. I'm not saying they didn't get better. I'm not even saying they didn't get *a lot* better. They just didn't get all the way to "good."

"(And, if anything, NE was actually better on offense when the D was struggling, but I'm willing to overlook that for the sake of this discussion)."

Not really. They scored more points, yes, but by DVOA they're actually *better* on offense than they were then. Scoring 27 against Kansas City looks cool, it's 27% more than their ppg on average... except Kansas City gives up 1.92 points per *drive* and that game was 14 drives long. So it's... basically exactly their points per drive on average. (And their offensive DVOA for that game was about 0% as well).

Ditto with the Texans game. 36 points is 32% above their season points per game allowed. But 7 of those points were basically scored by the defense (yay defense), and the remaining 29 points took 12 drives to score: which is only about 10% above the Texan's 2.18 points per drive. (And again, their offensive DVOA that game was about ~5%, so pretty close).

Weeks 2 and 4 were awesome (week 2 was their best offensive performance of the year) but not so awesome that the first 4 weeks collectively weren't pretty much dead on in line with the rest of the year.

Please keep in mind what I'm saying there in case you think I have some sort of strange anti-Patriots bias: their defense hasn't gotten as *good* as it appears over the year, but their *offense* hasn't actually gotten any worse and is just as scary-good as it has been all year.

25
by Anon Ymous :: Thu, 02/01/2018 - 3:34pm

Ditto with the Texans game. 36 points is 32% above their season points per game allowed. But 7 of those points were basically scored by the defense (yay defense), and the remaining 29 points took 12 drives to score: which is only about 10% above the Texan's 2.18 points per drive. (And again, their offensive DVOA that game was about ~5%, so pretty close).

Granted, though it needs to be noted that Houston accrued most of those poor defensive statistics after losing Watt and Mercilus, who both played against NE. This, of course, also serves to artificially suppress NE's offensive DVOA as well.

I'm not even saying they didn't get *a lot* better. They just didn't get all the way to "good."

Then we largely agree. I think they've become a good defense with a disproportionate ppg allowed, but the gap between our positions is minimal.

FWIW, as a refresher, I initially responded to this comment:

Could the vast improvement be 20% scheme/fit/coaching and 80% they just played some really poor teams after the bye? Maybe its 30/70 or 40/60? Does anyone really think that the Jets, Bills x2, Dolphins x2, Raiders, and Broncos were offensive juggernauts that were shut down by a mighty defense? The Steelers had 413 yards and missed out on 31 points with a goal line dribble and an epic meltdown.

Seeing as there was nothing about points in this comment, I interpreted it as asking if the DVOA numbers needed parsing, not ppg. Had I known that was what the OP was referring to I wouldn't have pushed back.

13
by Badfinger :: Wed, 01/31/2018 - 10:50am

The Eagles DO have an edge against the Patriots Run Defense!

http://www.sharpfootballanalysis.com/blog/2018/to-secret-to-running-on-t...

16
by Pat :: Wed, 01/31/2018 - 12:14pm

That's really interesting - it's also worth noting that the Titans and Jaguars barely ever/never ran to the right side, which is also where the Patriots are noticeably worse compared to the left side. That behavior for the Titans and Jaguars is in line with their behavior as a team, rather than something game-specific, whereas the Eagles are tied for the most runs around right end in the NFL.

For the Titans, at least, it really may have been the case that the Patriots run defense just matches up really, really well against what they like to do.

26
by Obat Pembesar Penis :: Wed, 02/21/2018 - 8:15pm

https://www.kampungobat.com Thank you for the information good luck always..