Blind Power Rankings: Week 9 Power Rankings, Drive Stats, Average Margin of Victory

Naturally, the Vikings have been a depressing team to watch, but they can at the very least take heart in the fact that they aren't the worst team in the league. Just close to it.

Once again, if you want an explanation for how the rankings are calculated, take a look here.

One thing I've noticed is that the rankings put a similar amount of weight on the relative capabilities of the offense and defense, so a team that is dominant in one respect won't do well without the other half of their unit performing. In this sense a few models may overrate teams like Carolina (who should be in the top five, but perhaps shouldn't be ranked first overall, like one model presumes) and underrate teams like Denver (eighth seems low).

As I remind everyone every week, I prefer raw efficiency over any other metric, including the average rank these are sorted by. But I include everything, so you can choose which you prefer.

First, we'll look at the rankings without a recency bias. Remember, the headers are all sortable:

Team	Efficiency	Points	Wins	Average	Average Rank
Seattle Seahawks	1	2	2	1.67	1
Indianapolis Colts	5	3	3	3.67	2
Denver Broncos	6	1	5	4.00	3
Kansas City Chiefs	4	8	1	4.33	4
New Orleans Saints	3	5	6	4.67	5
San Francisco 49ers	8	4	4	5.33	6
Green Bay Packers	10	6	8	8.00	7
Carolina Panthers	2	7	19	9.33	8
Cincinnati Bengals	12	10	7	9.67	9
Dallas Cowboys	7	9	14	10.00	10
Detroit Lions	11	11	10	10.67	11
Arizona Cardinals	9	14	13	12.00	12
New England Patriots	17	13	9	13.00	13
Tennessee Titans	13	15	12	13.33	14
San Diego Chargers	22	12	11	15.00	15
Chicago Bears	14	17	15	15.33	16
St. Louis Rams	15	21	24	20.00	17
Oakland Raiders	24	20	16	20.00	17
Buffalo Bills	19	22	20	20.33	19
Cleveland Browns	20	19	22	20.33	19
Miami Dolphins	28	18	17	21.00	21
Philadelphia Eagles	16	24	26	21.33	22
Baltimore Ravens	30	16	21	22.33	23
Washington Redskins	18	25	25	22.67	24
Houston Texans	21	27	23	23.67	24
Atlanta Falcons	26	23	27	25.33	26
New York Jets	27	31	18	25.33	26
New York Giants	23	28	29	26.67	28
Pittsburgh Steelers	31	26	28	28.33	29
Tampa Bay Buccaneers	25	30	32	29.00	30
Minnesota Vikings	29	29	31	29.67	31
Jacksonville Jaguars	32	32	30	31.33	32

The Seahawks, despite taking a massive hit in their efficiency against a mediocre St. Louis team (one that will hurt them more later after St. Louis continues to suffer at the hands of Kellen Clemens), are hitting on all cylinders. It's not correct to say they are "massively ahead" of the others because they are highly ranked in all three metrics, but that their performance is matching their results.

Others who are performing to their metrics: nearly everyone at the bottom of the rankings. The most interesting differences are at the top, with Carolina, San Francisco, Kansas City and Denver.

There seems to be clear tiers forming in the power rankings. The tiers are not particularly useful, of course, because we are averaging three completely different (but not unrelated) ranks. Still, it's a fun concept.

Tier One: Seattle Seahawks
Tier Two: Indianapolis Colts, Denver Broncos, Kansas City Chiefs, New Orleans Saints
Tier Three: San Francisco 49ers
Tier Four: Green Bay Packers, Carolina Panthers, Cincinnati Bengals, Dallas Cowboys, Detroit Lions
Tier Five: Arizona Cardinals, New England Patriots, Tennessee Titans, San Diego Chargers, Chicago Bears
Tier Six: St. Louis Rams, Oakland Raiders, Buffalo Bills, Cleveland Browns
Tier Seven: Miami Dolphins, Philadelphia Eagles, Baltimore Ravens, Washington Redskins, Houston Texans
Tier Eight: Atlanta Falcons, New York Jets
Tier Nine: New York Giants
Tier Ten: Pittsburgh Steelers, Tampa Bay Buccaneers, Minnesota Vikings
Tier Eleven: Jacksonville Jaguars

There's nothing to read into the tiers. The Seahawks are a high-efficiency team whose results are matching their relative performance. Tier Two consists of slight differences in performance/results while Tier Three has an overperformer. After that, it's open season on what they mean. Carolina and Green Bay, for example, are sort of opposites in their character here.

The Eagles will not turn it around without a quarterback, and the efficiency model overstates their ability (as it does in a small way with Carolina and understates Denver) because of how it de-emphasizes turnovers, which are difficult to project due to low repeatability.

But I think it is fair to say that the reason turnovers are rarely repeatable is because players at the NFL level are so elite that turnovers are low-probability events. They perform at one tail end of the bell curve, and most of our debates are separating the best 0.00001% from the 0.00002%. With the Eagles, I am not so sure their quarterbacks fit into that tail end, and therefore their turnover rate might be more repeatable.

This is why I think Chip Kelly should not be held accountable for the difference in efficiency and wins, unlike Ron Rivera (who is turning it around in his new Riverboat Rivera persona).

This model evidently hates Baltimore. I was extremely pessimistic on them before the season, but not this pessimistic. This is very interesting, to say the least.

I would argue that San Diego, New England and Baltimore benefit from extremely intelligent coaching, but that is just speculation and I could simply be inventing a narrative that fits what we already know (Bill Belichick and John Harbaugh are good coaches, and Mike McCoy might be) instead of more accurately attributing these differences to luck.

There's another reason those teams may be performing better than their efficiency scores. I will detail it near the bottom at the "Drive Stats" portion.

Here are the recency-weighted metrics. They are still slow to incorporate data like QB injuries, but are better about it:

Team	Efficiency	Points	Wins	Average	Average Rank
Seattle Seahawks	2	3	2	2.33	1
San Francisco 49ers	5	2	3	3.33	2
Indianapolis Colts	4	4	4	4.00	3
Kansas City Chiefs	7	8	1	5.33	4
Denver Broncos	8	1	7	5.33	4
New Orleans Saints	3	7	8	6.00	6
Green Bay Packers	9	5	6	6.67	7
Carolina Panthers	1	6	15	7.33	8
Cincinnati Bengals	11	10	5	8.67	9
Dallas Cowboys	6	9	13	9.33	10
Detroit Lions	13	11	10	11.33	11
Arizona Cardinals	10	13	12	11.67	12
New England Patriots	18	15	9	14.00	13
San Diego Chargers	20	12	11	14.33	14
Chicago Bears	12	17	17	15.33	15
Tennessee Titans	15	16	16	15.67	16
St. Louis Rams	14	18	24	18.67	17
Oakland Raiders	24	20	14	19.33	18
Washington Redskins	16	24	22	20.67	19
Baltimore Ravens	28	14	20	20.67	19
Cleveland Browns	23	21	19	21	21
Buffalo Bills	21	22	21	21.33	22
Philadelphia Eagles	17	23	26	22	23
New York Giants	19	27	27	24.33	24
Houston Texans	22	28	23	24.33	24
Miami Dolphins	29	19	25	24.33	24
New York Jets	30	31	18	26.33	27
Atlanta Falcons	26	25	29	26.67	28
Pittsburgh Steelers	27	26	28	27	29
Tampa Bay Buccaneers	25	30	32	29	30
Minnesota Vikings	31	29	31	30.33	31
Jacksonville Jaguars	32	32	30	31.33	32

Big differences here from the neutral rankings are San Francisco, who saw elevated QB play; San Diego, whose performance against the Raiders is dragging them down while not benefiting by playing Jacksonville; Philadelphia, who I've discussed; and Houston and Tampa Bay, whose new quarterbacks are being folded into the data better in the second set of rankings. Also significant is Miami, who simply lost their hot start.

Against the Spread

Last Week

Efficiency: 9-4
Efficiency (What Have You Done For Me Lately): 8-5
Points: 10-3
Points (WHYD4ML): 10-3

Woohoo!

Total

Efficiency: 18-24
Efficiency (What Have You Done For Me Lately): 21-21
Points: 21-21
Points (WHYD4ML): 24-18

There we go. These things come in bunches.

Next Week

Efficiency

Cincinnati over MIAMI (+3)
Kansas over BUFFALO (+3.5)
DALLAS (-10.5) over Minnesota
Tennessee over ST. LOUIS (-3)
WASHINGTON (-0) over San Diego
CAROLINA (-7.5) over Atlanta
SEATTLE (-18.5) over Tampa Bay
NEW ENGLAND (-7) over Pittsburgh
CLEVELAND (+3) over Baltimore
INDIANAPOLIS (-3) over Houston
Philadelphia over OAKLAND (-2.5)
Chicago over GREEN BAY (-11)

Efficiency (WHYD4ML)

Cincinnati over MIAMI (+3)
Kansas over BUFFALO (+3.5)
DALLAS (-10.5) over Minnesota
Tennessee over ST. LOUIS (-3)
WASHINGTON (-0) over San Diego
CAROLINA (-7.5) over Atlanta
Tampa Bay over SEATTLE (-18.5)
Pittsburgh over NEW ENGLAND (-7)
CLEVELAND (+3) over Baltimore
INDIANAPOLIS (-3) over Houston
Philadelphia over OAKLAND (-2.5)
Chicago over GREEN BAY (-11)

I am somewhat skeptical of the recency-biased rankings' differences with the unbiased rankings (Tampa Bay/Pittsburgh over Seattle/New England) but those spreads are both very large so I understand.

Points

Cincinnati over MIAMI (+3)
Kansas over BUFFALO (+3.5)
DALLAS (-10.5) over Minnesota
Tennessee over ST. LOUIS (-3)
San Diego over WASHINGTON (-0)
CAROLINA (-7.5) over Atlanta
SEATTLE (-18.5) over Tampa Bay
NEW ENGLAND (-7) over Pittsburgh
CLEVELAND (+3) over Baltimore
INDIANAPOLIS (-3) over Houston
OAKLAND (-2.5) over Philadelphia
GREEN BAY (-11) over Chicago

Points (WHYD4ML)

Cincinnati over MIAMI (+3)
Kansas over BUFFALO (+3.5)
DALLAS (-10.5) over Minnesota
Tennessee over ST. LOUIS (-3)
San Diego over WASHINGTON (-0)
CAROLINA (-7.5) over Atlanta
SEATTLE (-18.5) over Tampa Bay
NEW ENGLAND (-7) over Pittsburgh
CLEVELAND (+3) over Baltimore
INDIANAPOLIS (-3) over Houston
OAKLAND (-2.5) over Philadelphia
GREEN BAY (-11) over Chicago

The picks for the point differential models are identical, but the confidence levels are not. The recency model really likes Cincinnati and Green Bay, but is hesitant about Baltimore. The unbiased model is very hesitant on Green Bay and not too confident in Cincinnati, but is reasonably confident about Baltimore.

The biggest confidence levels were for Dallas and Indianapolis. It seems that Vegas predicts the betting markets won't play action on Dallas at a more realistic level or don't understand how good Dallas is.

Drive Stats

As I've alluded to previously, drive stats are probably a better way to evaluate team quality than pure point totals. Some teams grind out the clock and decrease the number of drives in a game but get points on the board, like Carolina. Some teams relish having as many drives as possible, like Buffalo, so their offensive point total does not represent them very well.

While this doesn't affect point differential in significant ways (although it may overstate magnitude for drive-heavy teams like Miami and understate magnitude for drive-averse teams like Atlanta), it does affect the individual offensive and defensive points and therefore our ability to adequately assess offensive and defensive strength.

The table below is a point-specific set of drive data, with a few caveats. The first is that all special teams touchdowns are eliminated, because they are generally unpredictable and nonrepeatable. But the affect that special teams have on field position are already "priced in". I have also eliminated defensive scores.

The second caveat is that I've eliminated any drives that are "garbage time," which I've defined as any drive that takes place after a team has either a 99% chance of winning or a 1% chance of winning, based on historical point differential/time remaining data found at Advanced NFL Stats.

After poring over the data over the last few years, no team has won after hitting that game state, but they have put up points. So I am comfortable with doing that.

I have also eliminated kneeldowns.

Finally, I created an "average game" in the adjusted point differential column that has 11.7 drives, but that's not entirely accurate because the 11.7 figure include garbage time and kneeldowns while the data do not. There is an additional misleading point that I will discuss about this data in a second. For now, look at it:

Overall				Offense			Defense
Tm	Rk	PD/Dr	Adj PD/Gm	Rank	Pts	Pts/Dr	Rk	Pts	PD/Dr
Denver Broncos	1	1.31	15.32	1	313	3.10	21	179	1.79
Carolina Panthers	2	1.17	13.66	5	171	2.19	2	84	1.02
Green Bay Packers	3	1.11	13.01	2	192	2.49	7	105	1.38
New Orleans Saints	4	0.91	10.62	4	186	2.33	8	112	1.42
San Francisco 49ers	5	0.87	10.16	9	185	2.01	4	112	1.14
Seattle Seahawks	6	0.84	9.79	13	181	1.93	3	98	1.09
Indianapolis Colts	7	0.56	6.51	6	168	2.18	13	130	1.63
San Diego Chargers	8	0.55	6.44	3	168	2.33	20	123	1.78
Kansas City Chiefs	9	0.49	5.78	22	127	1.43	1	97	0.93
Cincinnati Bengals	10	0.43	5.00	12	177	1.97	11	137	1.54
New England Patriots	11	0.37	4.35	16	165	1.60	5	123	1.23
Dallas Cowboys	12	0.26	3.02	7	203	2.14	25	186	1.88
Detroit Lions	13	0.25	2.96	8	194	2.02	18	175	1.77
St. Louis Rams	14	0.09	1.04	17	131	1.58	9	140	1.49
Tennessee Titans	15	0.03	0.37	26	103	1.32	6	107	1.29
Arizona Cardinals	16	-0.03	-0.36	18	147	1.50	10	150	1.53
Philadelphia Eagles	17	-0.07	-0.78	15	168	1.66	16	173	1.73
Baltimore Ravens	18	-0.09	-1.04	21	130	1.46	12	141	1.55
Chicago Bears	19	-0.28	-3.22	10	162	2.00	28	182	2.28
Buffalo Bills	20	-0.31	-3.67	19	160	1.50	23	199	1.81
Pittsburgh Steelers	21	-0.35	-4.07	23	115	1.40	17	140	1.75
Oakland Raiders	22	-0.36	-4.19	28	105	1.28	14	136	1.64
New York Giants	23	-0.39	-4.53	29	134	1.28	15	168	1.66
Atlanta Falcons	24	-0.42	-4.93	10	138	2.00	31	184	2.42
Miami Dolphins	25	-0.44	-5.14	24	145	1.37	22	150	1.81
New York Jets	26	-0.44	-5.16	25	134	1.33	18	175	1.77
Cleveland Browns	27	-0.52	-6.05	27	110	1.31	24	179	1.83
Washington Redskins	28	-0.54	-6.27	14	140	1.75	29	192	2.29
Houston Texans	29	-0.64	-7.51	30	106	1.26	26	139	1.90
Minnesota Vikings	30	-0.91	-10.65	20	121	1.48	30	198	2.39
Tampa Bay Buccaneers	31	-0.99	-11.57	31	92	1.05	27	179	2.03
Jacksonville Jaguars	32	-1.67	-19.55	32	79	0.84	32	226	2.51

If I've coded correctly, the subheaders should be sortable, but the large multiple column headers won't be affected. It will be fun to find out, I'm sure.

The reason this is misleading is that while it does measure offensive and defensive output, it doesn't entirely measure the quality of those units. A huge reason is because it eliminates the effect of field position that one unit will have on another, along with the effect that special teams has on field position. I wrote an article for the Bleacher Report today that explains the significance of that as it pertains to the Vikings:

On a surface level, measuring offensive and defensive scoring opportunities is a simple way to determine whether or not a team is taking advantage of their drives.

But even that doesn't speak to the larger offensive problems, because the Vikings offense cannot create many of its own points—it needs more help from field position than other teams and by a significant amount.

The Vikings are the fourth-worst in the league at converting drives into scores from inside their own 20 and have the vast majority of their points come from favorable position.

They have needed more help than any other team except the Jacksonville Jaguars and the Dallas Cowboys in creating scoring opportunities and have the third-most favorable field position on touchdowns.

Simply put, they cannot put the ball in the end zone unless they have significant help in the field-position battle-a function of special teams and defense.

The offense is much worse than people might think, based on points scored.

I discussed ways to isolate performance from field position. One is almost captured in the efficiency rankings above, with a heavy bias on yardage and success rate. The other is more direct and already done by Football Outsiders:

That's not to say that the offense is worse than the defense, simply that the two are closer in quality than many people would think after looking at the surface-level statistics.

No single metric captures this better than Football Outsider's "Drive Success Rate"statistic, a field-position independent statistic that measures how well a team can gain first downs and touchdowns as a percentage of its total first-down opportunities.

The table below is, to date, the Drive Success Rate of every team in the league. Like the table above, the one below is not adjusted for opponent. Next week, perhaps.

Team	NET DSR	NET Rank	OFF DSR	OFF Rank	DEF DSR	DEF Rank
Denver Broncos	0.116	1	0.797	1	0.680	17
Carolina Panthers	0.104	2	0.751	3	0.648	5
Kansas City Chiefs	0.072	3	0.650	22	0.578	1
New Orleans Saints	0.055	4	0.716	7	0.661	9
San Diego Chargers	0.054	5	0.767	2	0.713	25
Detroit Lions	0.053	6	0.726	5	0.673	13
Seattle Seahawks	0.042	7	0.702	11	0.660	8
Indianapolis Colts	0.038	8	0.716	7	0.678	16
Green Bay	0.037	9	0.718	6	0.681	18
Baltimore Ravens	0.030	10	0.631	28	0.601	2
Atlanta Falcons	0.022	11	0.744	4	0.722	29
Houston Texans	0.021	12	0.681	15	0.661	9
San Francisco 49ers	0.021	12	0.656	19	0.635	4
New York Jets	0.011	14	0.627	30	0.617	3
Cincinnati Bengals	0.011	14	0.686	14	0.675	14
Dallas Cowboys	0.008	16	0.715	9	0.707	23
New England Patriots	0.007	17	0.656	19	0.648	5
Washington Redskins	-0.008	18	0.708	10	0.716	27
Pittsburgh Steelers	-0.017	19	0.661	17	0.677	15
Tennessee Titans	-0.017	19	0.645	23	0.662	11
Chicago Bears	-0.019	21	0.701	12	0.721	28
Cleveland Browns	-0.021	22	0.662	16	0.682	20
Arizona Cardinals	-0.022	23	0.659	18	0.681	18
Buffalo Bills	-0.034	24	0.621	31	0.655	7
Philadelphia Eagles	-0.035	25	0.693	13	0.728	30
Tampa Bay Buccaneers	-0.038	26	0.635	25	0.672	12
St. Louis Rams	-0.057	27	0.655	21	0.712	24
Miami Dolphins	-0.059	28	0.636	24	0.694	21
Oakland Raiders	-0.077	29	0.628	29	0.705	22
New York Giants	-0.082	30	0.632	27	0.714	26
Minnesota Vikings	-0.104	31	0.634	26	0.737	32
Jacksonville Jaguars	-0.150	32	0.586	32	0.736	31

This is generally considered quite predictive. I assume it's a bit better after accounting for strength of schedule. This also reveals some things that disagree with my models a lot, which is why I'm so excited to fold the concept into my model next year. Baltimore ranks highly here and does not in my efficiency model. Other teams that do better in DSR than in my efficiency scores are San Diego, Atlanta and the New York Jets.

Notably, all those teams except Atlanta are "overperforming" their wins.

The teams who do much better in efficiency than they do in drive success rate include St. Louis, Arizona, Philadelphia and in a small way, Chicago.

Two of those teams are "underperforming" relative to their efficiency.

The only worry I have is that these drive statistics, by nature, fold in turnovers. They are extremely important to winning games, but do not do a good job replicating themselves so are not necessarily predictive. It's embedded into the type of data you need to gather for it, but when constructing the model there are ways around it by using "field goals and punts per drive" as a negative instead of "first downs per drive" as a positive.

You all have no idea how excited I am by this.

But next year.

Finally, one of the more interesting statistics I've come across: average margin of victory, which I explained last week. Again, it comes from Football Persepective's work on Game Scripts (which is the real term for it, but I'm using it for a bit of a different purpose.

It might turn out to be another statistic worth incorporating, although I'm not sure of it given its emphasis on "time" instead of drives. I could modify it to drives for next year however.

Woohoo.

Current strength-of-schedule adjusted average margin of victory:

Rank	Team	Average MOV
1	Green Bay Packers	6.6
2	Carolina Panthers	6.3
3	San Francisco 49ers	6.2
4	New Orleans Saints	4.5
5	Denver Broncos	4.4
6	Seattle Seahawks	4.0
7	Dallas Cowboys	3.9
8	Cincinnati Bengals	3.6
9	Indianapolis Colts	3.6
10	Kansas City Chiefs	2.8
11	New England Patriots	1.7
12	San Diego Chargers	1.2
13	Oakland Raiders	1.0
14	Detroit Lions	0.9
15	Atlanta Falcons	0.6
16	Arizona Cardinals	0.3
17	Buffalo Bills	-0.7
18	Miami Dolphins	-0.8
19	Tennessee Titans	-1.0
20	Cleveland Browns	-1.1
21	Chicago Bears	-1.5
22	Baltimore Ravens	-2.4
23	Philadelphia Eagles	-2.6
24	New York Giants	-2.6
25	Minnesota Vikings	-2.7
26	Tampa Bay Buccaneers	-3.0
27	St. Louis Cardinals	-3.1
28	Washington Redskins	-4.4
29	Houston Texans	-4.6
30	New York Jets	-5.2
31	Pittsburgh Steelers	-5.3
32	Jacksonville Jaguars	-10.8

Unsurprising that the Packers maintained their lead given their game against the Vikings. The Vikings, however, moved UP a spot because they did better, on average, than most teams did against Green Bay. More importantly, the other teams the Vikings have played against in the past (the Giants, Bears and Browns specifically) had positive average margins of victory this past week and that helped the Vikings' historical strength of schedule.

This is perhaps the most encouraging ranking. But the most encouraging really just means least discouraging, honestly.

It means that the Vikings don't lose games by a lot until late into the game.

There is also probably no one number that better underscores how bad the Jaguars really are.