the watchtower: lions at vikings

Before we delve into a very, very special Watchtower—the first to feature an opponent that has already been Watchtower’d—let’s take a look at last week’s installment:

• Regardless of talent, Gunther Cunningham defenses disproportionately depress the scoring of Greg Knapp offenses, despite allowing typical rushing and passing per-play effectiveness.
• Given a definite talent advantage and a mild systemic points-denial disadvantage, I expect the Seahawks to mildly overperform their season averages: 20-23 points, with 6.00-6.25 YpA, and 3.75-4.00 YpC. I have medium-high confidence in this prediction.
• It seems as though Jim L. Mora’s conservative 4-3 disproportionately depresses the per-play effectiveness of Linehan’s balanced offense—unless scoring can come from big plays that get behind the defense.
• If "Dr. Jekyll" shows up--the Lions offense that features Matt Stafford, Calvin Johnson, and Kevin Smith all healthy and effective--the Lions should be able to push it deep and meet or slightly outperform their expectations: 17-20 points, 6.25-6.5 YpA, and 3.75-4.YpC. I have medium-low confidence in this prediction.

. . . and the results, in gorgeous table format:

Off.	PgG	YpA	YpC	Drnk	PpG	DYpA	DYpC	PTS	YpA	INT	YpC	Sack
DETp	19.3	5.84	3.51	31st	29.3	7.53	4.82	17-20	6.25-.5	-	3.75-4	-
DETa	19.3	5.84	3.51	31st	29.3	7.53	4.82	20	4.83	5	5.43	2-17
SEAp	16.1	5.42	3.76	14th	21.0	6.49	4.25	20-23	6-6.25	-	3.75-4	-
SEAa	16.1	5.42	3.76	14th	21.0	6.49	4.25	25	6.62	1	3.06	1-2

First, of course, the points: I projected 17-20 for the Lions; they scored 20. I projected 20-23 for the Seahawks; their offense scored 25.  I got into a protracted argument with Seahawks fans about whether or not it is ethically legitimate to subtract a defensive TD from “actual points scored” while evaluating a projection of offensive points; I won.

Oddly, as close as the projection was on points scored, it was wildly off-base on run/pass effectiveness--in both directions, for both teams.  I projected 6.25-to-6.50 yards per pass attempt for the Lions; they produced a miserable-even-for-them 4.83.  Of course, that’s partially explicable by the five attempts that went for zero yards . . . on the ground, I projected a mildly better-than-average day, 3.75-to-4.00 YpC.  Instead, the Lions’ ground game was breathtakingly effective, gashing Seattle for 5.43 YpC.

For Seattle’s part, I projected 6.00-6.25 YpA, and instead they averaged almost a half-yard more than that: 6.62 YpA.  On the ground, I projected they’d just about match their average, 3.75-to-4.00 YpC—and instead, with run-stuffer Grady Jackson a healthy scratch, the Lions held the ‘Hawks to just 3.06 yards per carry.

Interestingly, this run/pass effectiveness reversal seems to happen a lot.  Might it be because defenses load up to stop where teams are strong, and then the offense outperforms expectations the other way?  Figuring out if this is a statistically quantifiable phenomenon might make for an interesting offseason project.  Either way, this run/pass Whack-a-Mole effect produced the projected final results: a regrettably foreseeable, no-less-heartbreaking loss.

Brad Childress vs. Gunther Cunningham

Chilly	Gun	Ornk	PgG	YpA	YpC	Drnk	PpG	DYpA	DYpC	PTS	YpA	INT	YpC	Sack
PHI	TEN	4	25.9	6.18	4.54	11th	20.2	6.30	3.83	24	5.89	2	3.64	1-1
PHI	KCC	18	19.4	5.93	3.92	16th	20.3	6.58	4.10	37	7.68	1	1.64	3-1
MIN	DET	2nd	30.5	7.18	4.15	31st	29.6	7.36	4.65	27	6.04	0	4.48	2-1

In the Watchtower for the first Vikings game, I identified a systemic advantage for the Lions' blitzing 4-3 defense against Brad Childress' conservative flavor of the Bill Walsh offense.  Sure enough, the Lions stymied the Vikings in the first half, shutting them out completely.

I expected the Vikings to overcome this disadvantage by airing it out over the Lions' suspect secondary--but instead, they nibbled underneath. Brett Favre was sparklingly efficient while dinking and dunking, completing 23 of 27 pass attempts.  Adrian Peterson, while very effective on a per-carry basis (6.13 YpC), was effectively contained, gaining only 92 yards on 14 carries; his sole TD accounted for 27 of those.

Since that game, the Vikings have proven that they’re a force to be reckoned with: they’re the second-best scoring offense in football, averaging 30.5 points per game.  They’re excelling in both dimensions of offense, netting 7.18 YpA and 4.15 YpC.  Meanwhile, the Lions’ defense has been anything but forceful: they’re the second-worst scoring defense in football, allowing 29.6 points per game.  They’re failing in both dimensions of defense, allowing 7.36 YpA and 4.65 YpC. 

Of course, it's tempting to turn around and predict a 40-point explosion for the Vikings--but, that didn’t happen the first time.  In fact, the Lions held the Vikings to 3.5 points beneath their season average, partially thanks to that aggressive passing D.  The Vikings fell more than a yard short of their season per-attempt average, and only outgained their season per-carry average by a third of a yard.

So, what to do? The strongest possible data point of all, a prior in-season meeting between two teams, seems to trump everything--and yet, when two teams meet twice in a season, it never results in the same game being played twice.  Part of this is adjustment; both teams learned a lot about each other from the first meeting.  That might lead to some of the Whack-a-Mole effect I discussed above: if the Lions come out defending the short pass with press man coverage, and linebackers in zones behind them, the Vikings might immediately attack the sidelines deep, and rush AD often, without fearing the run blitz.

However, we could spend all day playing the “I know that YOU know that I know that YOU know . . .” game: 

Rather than attempt to decide which dimension of the Vikings' offense holds the iocaine, let's go right back to the data.  Despite the second-best offense in the NFL meeting the second-worst defense, that offense underperformed its season averages.  I originally concluded that the Gunther Cunningham 4-3 disproportionately disrupts Childress’ conservative flavor, and that conclusion was indisputably correct.

Now, I didn’t make an actual prediction for the game beyond “a medium-to-low scoring slugfest”; I was just starting this feature, and hadn’t refined it to the level I have now.  Therefore, I’ll simply refine my original prediction: Given a huge talent and execution advantage, but a definite systemic disadvantage, I expect the Vikings will meet or slightly underperform their season averages: scoring 27-30 points, passing for 6.75-7.0 yards per attempt, and rushing for 3.75-4.0 yards per carry.  I have very high confidence in this prediction.

Mitigating/Augmenting Influences:

I’ve noticed that when there are two in-season games between two teams, two primary factors cause the results to vary between games: weather tipping the run-pass balance towards “run”, and variances in quarterback play.  Obviously, these are two dome teams, so weather isn’t a factor.  Brett Favre has been playing generally better as the season wears on, and has developed a much stronger rapport with Vikings wideouts like Sidney Rice. 

However, he was essentially flawless in the first meeting between two teams; how much better could he play?  The Lions’ defense, meanwhile, has been playing generally better as the season wears on as well (their current averages are still skewed by the Saints and Bears blowouts).  Their pass rush, especially, has been markedly better.

The final result of this should be that Favre has a lower completion percentage, gets sacked more often, and possibly commits a turnover or two—but he should also complete more deep passes, possibly for devastating quick-strike TDs.  I’m going to cite the Whack-a-Mole principle and call this all a wash; in the end, the Vikings should meet the objective statistical predictions.

Scott Linehan vs. Leslie Frazier

Lin	Fraz	Ornk	PgG	YpA	YpC	Drnk	PpG	DYpA	DYpC	PTS	YpA	INT	YpC	Sack
MIN	TBB	8th	24.4	6.60	5.3	1st	12.2	4.88	3.79	24	7.78	2	7.52	3-11
MIN	IND	6th	25.3	7.16	4.71	19th	21.9	7.15	4.43	28	8.89	0	5.75	2-1
MIA	TBB	16th	19.9	5.94	3.69	8th	17.1	6.15	3.46	13	6.21	0	3.56	3-2
DET	MIN	26th	16.6	5.27	3.92	17th	21.8	6.89	4.14	13	5.07	2	3.79	2-1

In the previous installment, I also identified a mild systemic advantage for Scott Linehan offenses against Dungy-style Tampa 2 defenses.  When Linehan had a talent advantage over the Dungy D, the running game was even more effective.  When Linehan had lesser talent, this systemic advantage seemed very mild or nonexistent.

In Week 2, this analysis was confirmed: the Lions scored 13 points, just a little beneath their season average.  The Lions managed 5.07 YpA against the Vikings; just a little beneath their season average.  The Lions ran for 3.79 YpC; just a little beneath their season average.  Considering that the Lions’ offense is the 26th-best in the NFL, and the Vikings are the 17th, underperforming averages across the board is exactly what you’d expect.  It could be argued that the Lions underperformed their averages less than expected--indicating that a systemic advantage affected the outcome--but that would be assigning big significance to some very, very small variances.

I originally concluded that given lesser talent, Scott Linehan’s balanced offense meets or slightly exceeds expectations against a Dungy-style defense, even while allowing more sacks and/or turnovers.  This conclusion was confirmed by the results of Week 2.

Again, I’ll refine my original prediction: with lesser talent, and therefore a small-to-zero systemic advantage, the Lions will slightly underperform, or approach, their season averages: 14-17 points scored, 5.0-5.25 YpA, and 3.75-4.0 YpC.  I have very high confidence in this prediction.

Mitigating/Augmenting Influences:

Again, what will be the biggest factor in differentiating Week 10’s results from Week 2’s?  Effective quarterback play.  Now, Matt Stafford’s Week 2 game against the Vikings was no great shakes: 18-of-30 for 152 yards, 1 TD, and 2 INTs.  However, that was a notable improvement from Week 1.  Tellingly, it was not yet as good as his best game as a pro: his 21-of-36, 241-yard, 1 TD/0 INT Week 3 win over the Redskins.

Once he’d established this very, very steep upwards trend line, he got hurt, and the offense completely fell apart.  It was clear against St. Louis that Stafford was rusty; his balls were inaccurate, and very frequently dropped.  It was similarly clear that Kevin Smith was not 100%; he looked hesitant and slow.  It was sparklingly, scintillatingly, absolutely crystal clear that without the threat of Megatron, defenses could simply smother the Lions with in-the-box defenders and press coverage.

However, against the Seahawks, it was clear that the offense that had started the season--the one that had averaged over 20 points in the first four games--had returned, leaving behind the miserable 10 PpG offense we'd seen against the Steelers, Packers, and Rams. With Stafford and Megatron both in the lineup, this offense would the 19th-best in the league, virtually tied with Seattle on a per-game basis. Without them both, it would be very nearly the worst; ahead of only the Rams, Browns, and Raiders.  This is the “Dr. Jekyll/Mr. Hyde” phenomenon I referred to last week.

Matthew Stafford's five interceptions cost the Lions the Seattle game; there's no doubt about that. Two of those picks came when Stafford failed to see a extra-deep linebacker playing centerfield--the cornerstone principle of the Tampa 2 defense.  The Seahawks just installed the Tampa 2 this season--and the Vikings are an established T2 team.  One would hope that this is a point of coaching emphasis this week!  If Stafford can cut the picks down from five to the two he threw in Week 2, and move the ball with the effectiveness he did against Seattle, the Lions’ O could drastically exceed expectations.

Would it surprise you to learn that the Seahawks are allowing only 20.88 points per game on defense; 13th-best in the league?  Would it further surprise you to learn that Minnesota’s is the 17th-best in the league, allowing 21.75?  I hope not; we’ve covered those figures very recently!  But even so, it’s shocking to see it in black and white: this season, Seattle is executing the Tampa 2 slightly better than Minnesota is.

So: if we continue down this primrose path, we can conclude that this matchup is "really" between Dr. Jekyll, the 19th-best scoring offense, and the 17-best scoring defense--and therefore, the systemic advantage that Linehan offenses possess against T2s should express itself, and the Lions should strongly outperform their "real" season averages, especially on the ground. The averages for the "Dr. Jekyll" Lions offense (Weeks 1, 2, 3, 4, and 9) are 20.6 PpG, 5.55 YpA, and 3.80 YpC.  Therefore, in Fantasyland, the Lions will score 27-30 points, pass for 6.00-6.25 YpA, and rush for 4.5-4.75 YpC.

Conclusion:

Unfortunately, even if you buy tickets to the "Fantasyland" I've constructed, the Lions and Vikings are both projected to score 27-30 points.  Further, with the way the Vikings are cutting through the league right now, and the way the Lions have been playing, I simply cannot see the Lions playing 60 minutes of “Dr. Jekyll” football in the Metrodome.  Moreover, one of the “Dr. Jekyll” games was against Minnesota, and as competitive as that game was (I was there), 27-13 is not the same as winning.  I’m going to stick with my original projections—the data is rock-solid—and say that the Lions will lose this week, scoring 14-17 points against Minnesota’s 25-30.