Defense helps offense helps defense...

Let me begin by saying that this not specifically an article about Russell Wilson.  It's a broad statistical analysis that--

Oh, who am I kidding? Of course it's about Wilson. Russell Wilson is the future of the Seahawks and the future of the NFL. For the next decade, any NFL-related article which does not reference Wilson is either irrelevant to the game of football or a tangential fluff piece, like, say a story about seat warmers at Lambeau.

In particular, this is about whether or not (and in what manner) Wilson benefits from playing on the same team as the league's best defense.

Of course a good defense helps a quarterback win more games. That's a given. Less certain is the idea that a strong defense somehow protects the offense from being pressured into scoring more,  thus allowing the offense to avoid mistakes, run the football if they choose, and ultimately be more efficient.

Part I: Correlation Is a Fickle Mistress

We'll look specifically at the Seahawks further down. But to answer the general question, we'll start with a broad comparison that also includes all of those sad, boring games without Russell Wilson. Pro Football Reference dot com's Game Play Finder handily lets us search composite statistics in situations where, for example, a team has at least a 7-point or lead or at least a 7-point deficit. So let's cut straight to the chase and see what that looks like:

All Games, All Teams, 2012-2014

Situation	TD%	yds/play	TO%
2nd & 3rd Quarter Leading 7+	3.9	5.61	2.6
2nd & 3rd Quarter Trailing 7+	3.5	5.55	3.6

The table can be read this way:

Teams with a good lead are more efficient on offense, gaining more yards, scoring more touchdowns, and committing fewer turnovers than teams with a bad deficit.

That statement is correct, and supports the idea that better score differential leads to better offense. But simply flip it around, and you get:

Teams that are more efficient on offense are more likely to have a lead.

Well, duh. This is correlation, not causation. Having a 7-point lead doesn't necessarily make you play offense better, and it makes much more sense that playing good offense causes you to have a 7-point lead.

We can test that by looking at a similar situational breakdown, but limiting our data to average teams. For this, I used Football Outsiders' efficiency stats and looked only at teams within +/- 10% for total VOA and within +/- 6% for offensive VOA.

Selected Teams 2014 (Vikings, Chargers, 49ers, Texans, Bengals, Colts, Lions)

Situation	Yards/ play	Turnover%	TD%
Trailing 7+	5.36	3.7%	3.4%
Leading 7+	5.20	2.9%	3.3%

This table proves that the first table proved nothing. All the numbers changed, with yards & touchdowns actually flipping to favor the trailing team.

But what about that turnover gap? Surely, that must result from offenses playing conservatively or recklessly based on the score, right?

Maybe not. There is still correlation. We have to keep in mind that each team's efficiency and score are affected by their weekly matchup, which may involve a weaker/stronger opponent, personnel mismatches, and even injuries.

So maybe the problem here is we're looking at score differential, which itself is partly determined by the quality of offense. Perhaps we can get around that by comparing offensive performance to total points allowed:

All Games, All Teams, 2012-2014

Points Allowed	Points Scored	Yards/ Play	Turnovers
24 or more	22.4	5.47	1.96
all games	22.98	5.41	1.54
19 or fewer	23.7	5.35	1.06

That seems clear. When you give up more points, your offense has to take more risks, and you commit more turnovers. That settles it, right?

Guys, check your notes. Don't make me point at Pete again.

[This is where you all mumble "correlation"]

Every time you turn the ball over, you increase your opponent's scoring on the following drive by 1.134 points. What's more, you shorten two drives (yours and theirs) and cut out a special teams play for a combined 105 seconds of game clock saved, which translates to an additional .662 points for each team. Committing just one turnover, then, increases the opposition score by 1.8. A three-turnover game adds 5.4 points to your opponent's total.

We can easily verify that by examining all games from 2012 to 2014 where a team committed at least three turnovers. In that sample, teams surrendered an average of 28.6 points-- significantly more than the league-wide average of 23.

All that data and nothing conclusive. So, finally, let me offer what I think is the best way to see the effect of game state on offensive efficiency: A quarter-by-quarter comparison on each of two fixed score differentials:

Here, at last, I will concede some causality. There is definitely conservative play late in the fourth quarter when teams have a lead (lower yards/play), and an increased risk tolerance for teams that are trailing (more turnovers).

But that's just the fourth quarter. And there are two more important caveats:

First, we have gone back to examining score differential, which means the offense is at least half to blame for the current game state. If you commit a turnover while trailing 13-6 late in the fourth, the defense isn't to blame. If you avoid turnovers while leading 34-24 late in the fourth, it's not because of a strong defense.

Second, even assuming that the defense is responsible for the score differential, there is no evidence that your overall offensive play is made better or worse. That is to say, if you looked at a good composite measurement that includes yards gained, points scored, and turnovers committed, I do not think you'd find any difference. Risk tolerance means committing more mistakes in exchange for more gains (and vice versa for risk avoidance).

Interlude: The Myth of Statistically Dominant Garbage Time

Thanks to Fantasy Football, many casual fans have become attentive to the phenomenon of skewed statistical performances in late-game situations. But increased awareness often leads to flawed analyses.

So what is "garbage time", or at least, "altered strategy time"? Here are the number of plays run under select conditions (2012-2014):

Trailing 15+ in the 3rd quarter: 3211 plays (3.2%)
Trailing 8+ in the 4th quarter: 9520 plays (9.6%)
Leading 14+ in the 3rd quarter: 2824 plays (2.9%)
Leading 7+ in the 4th quarter: 7220 plays (7.3%)

Those cut-off points are arbitrary, but justifiable. If you need only one touchdown to catch up in the 4th quarter, or only two touchdowns to catch up in the 3rd quarter, you don't have to alter your offense to get there. Numbers on 4th-down suggest that teams are still risk-averse in most of those situations:

When trailing by exactly 14 in the 3rd quarter, teams went for it on 4th down just 14.4% of the time (23/160); when trailing by exactly 7 in the 4th quarter with at least 7:00 remaining, teams went for it on 4th down just 15.1% of the time (23/152)

All-in-all, that's only 23.0% of plays with questionable levels of desperation/conservatism.

And if you're curious, here are the Seahawks' average point differentials at the start of each quarter from 2012-2014:

Start of 1st: +0
Start of 2nd: +2.17
Start of 3rd: +4.25
Start of 4th: +6.96 (Final: +9.90)

Part II: Seahawks Before & After

Russell Wilson. Because I haven't typed his name in a while.

And because there will be one more short diversion before I focus on the Seahawks, as we ask the question of whether or not a good offense benefits the defense.

To be environmentally responsible, I'm going to recycle some research I did back in 2013 when previewing the Monday showdown against the Saints, and when looking into Seattle's seemingly disappointing defensive performance on 3rd down.

The 2013 Saints had a pretty good defense:

Plays/drive 5.4 (10th)
Yards/drive 27.0 (11th)
Points/drive 1.6 (9th)

But in my humble opinion, that Saints defensive talent was a long ways from being in the top 10. Rather, their success was part of a comprehensive plan. The offense, despite having the ability to run hurry-up, was among the league's most deliberate:

2013 Saints Offense:
Plays/drive 6.1 (2nd)
Yards/drive 35.3 (3rd)
Time/drive 2:53 (3rd)

And while the offense did their part to keep the defense rested, the defense wasn't afraid to take risks to get themselves off the field:

2013 Sacks:
league average = 6.7%
Saints = 8.8% (4th)

2013 16+ yd passing plays allowed:
league average = 14.2% (80.6 / 567)
Saints = 19.9% (101 / 507)

The combined result of these strategies is that the Saints defense was on the field for just 943 scrimmage plays, fewest in the league and more than a full game's worth (65) below the average of 1041.

While fascinating in itself, this also serves as a stark contrast to the Seahawks' defensive philosophy and execution. Take a look at this mind-blowing year-to-year consistency:

Seahawks Defense

	2012	2013	2014
Plays/drive	5.9	5.5	5.5
rank	27	12	12
Yards/drive	29	23	25
rank	12	1	2

More plays translates to more yards, so being 10 spots better in yards/drive (versus plays/drive) is remarkable enough. But to be consistently 10-15 spots better three years running is astounding. Most Seahawk fans are already familiar with the strategy behind this anomaly: Pete Carroll's #1 priority on defense is to avoid giving up the big play. Seattle doesn't take risks to get off the field, but is willing to let the opposition plod along for a few first downs until they ultimately make a mistake.

And clearly, it's working. But if you're going to be patient on defense, you need help on the other side of the ball.

The 2011 Seattle Defense did not lead the league in any significant categories. But the pieces were there.

Although Bobby Wagner had not yet arrived, David Hawthorne was solid. A case could be made that Hawthorne in his last Seattle year was every bit as good as a rookie Wagner. Hawthorne led the team in solo tackles and assists, while pitching in 3 interceptions, 6 passes defensed, and 2 sacks. He was joined by Leroy Hill, who was still playing well (4.0 sacks from the linebacker position).

Chris Clemons had 11 sacks, Seattle allowed the 4th-fewest yards per rush (3.8), and Richard Sherman was the only starter in the secondary who did not go to the Pro Bowl (Brandon Browner had his best season, joining Thomas and Chancellor as selections).

The 2011 Seahawks allowed the same number of yards/play (5.1) as their 2012 defense (ranking 7th and 8th, respectively).

Now look at what was happening on the other side of the ball:

2011 Seahawks Offense, per drive:
5.2 plays (28th)
2 minutes 18 s (29th)
24.4 yards (29th)

2011 Seahawks Defense:
Average starting field position: 26.8 (10th)
Total scrimmage plays for the season: 1086 (9th most, 24th fewest)

No rest for the weary. Simply put, Seattle's potentially-elite defense got worn down within each game, and worn out through the course of the season.

But that changed dramatically when Rusell Wilson took over as quarterback:

2012 Seahawks Offense, per drive:
5.9 plays (8th)
3 minutes 0 s (1st)
32.9 yards (8th)

2012 Seahawks Defense:
Average starting field position: 24.4 (1st)
Total scrimmage plays for the season: 993 (30th most, 3rd fewest)

Bear in mind that this is before the arrival of Michael Bennett and Cliff Avril.

From 2011 to 2012, the Seahawks defense went from 9th (5315) in total yards to 4th (4899).

Defensive DVOA went from 10th (-7.1%) to 2nd (-14.5%).

Points allowed went from 7th (315) to 1st (245).

Tentative research suggests the defense was pretty good from 2013-2014, as well. Wilson isn't the sole cause, of course. But unlike the inexplicable notion of point differential changing the absolute quality of offense, there is a sound causal explanation whereby a defense is affected by fatigue.