Blind Power Rankings Week 11: Cohesion, Projecting Wins and Passer Identity

This week, a surprising win from Jacksonville was obviated by wins from Tampa Bay and Minnesota, allowing both the Buccaneers and Jaguars to continue competing for the top spot in the draft (or more accurately, compete out of that top position of the draft, a supposition that we here at the Daily Norseman agree on).

The biggest difference between performance and result may have had to have been the Cincinnati Bengals-Baltimore Ravens game, where the Bengals were a far more efficient team (a near identical turnover rate, a running game that was twice as successful and a passing game that was more than twice as efficient) but couldn't hold on in high-leverage situations.

Interestingly, the Colts-Rams game also was a big mismatch, but not because the Colts were more efficient—they were evidently twice as efficient as the score indicated, where the efficiency result of the teams would have predicted a nearly 60 point loss! Of course, the efficiency/point conversion rate falls apart at the extremes, so this is actually unsurprising.

Of note, Tampa Bay's win coincided with more efficient play than their opponents, while Jacksonville and Minnesota's did not (a surprising result, given how well Minnesota passed the ball in that game).

This isn't particularly odd. While efficiency scores have done a good job predicting future success, they do not do a particularly good job describing a current game score (just above 50%)—this similar to the difference between passer rating (highly explanatory and with a high correlation to the game that they are measured in, but a terrible statistic for predicting future performance) and yards per attempt (the opposite).

That's OK—if I wanted a statistic that counted current wins instead of future wins, I'd probably just use "wins," which is below. Otherwise I'd use quarterback kneels.

Power Rankings

Ranks below, first organized without recency bias. Again, if you want to get a primer on what these shortcut words mean, take a look here.

Like always, I strongly prefer the efficiency ranks over anything else, even the average I have presented below.

Remember, all headers are sortable:

Power Rankings - No Bias
Team	Efficiency	Points	Wins	Average	Average Rank
Seattle Seahawks	2	4	2	2.67	1
Denver Broncos	5	1	3	3.00	2
New Orleans Saints	3	3	4	3.33	3
Kansas City Chiefs	4	6	1	3.67	4
Carolina Panthers	1	2	9	4.00	5
San Francisco 49ers	8	5	6	6.33	6
New England Patriots	11	7	5	7.67	7
Indianapolis Colts	9	10	7	8.67	8
Detroit Lions	10	12	8	10.00	9
Arizona Cardinals	7	13	11	10.33	10
Dallas Cowboys	6	11	15	10.67	11
Cincinnati Bengals	15	8	10	11.00	12
Green Bay Packers	12	9	14	11.67	13
Chicago Bears	14	14	13	13.67	14
Philadelphia Eagles	13	15	16	14.67	15
Tennessee Titans	16	18	17	17.00	16
St. Louis Rams	18	16	20	18.00	17
Cleveland Browns	22	21	18	20.33	18
Buffalo Bills	17	22	24	21.00	19
New York Jets	23	28	12	21.00	19
Miami Dolphins	25	20	19	21.33	21
Baltimore Ravens	29	17	21	22.33	22
San Diego Chargers	27	19	21	22.67	23
Washington Redskins	20	26	25	23.67	24
Houston Texans	21	25	27	24.33	25
Tampa Bay Buccaneers	19	23	32	24.67	26
Atlanta Falcons	24	24	29	25.67	27
New York Giants	26	30	23	26.33	28
Pittsburgh Steelers	31	27	28	28.67	29
Oakland Raiders	30	31	26	29.00	30
Minnesota Vikings	28	29	31	30.00	31
Jacksonville Jaguars	32	32	30	31.33	32

Because of the relative wins of the other teams (or by playing a strong team, in the Falcons' case), Minnesota stays at 31 in the "wins ranking" despite having more wins than the team above it, as is often the case in SOS-adjusted ranks. Jacksonville's losses "meant less" and its win "meant more".

The rankings are converging, which is nice to see. In Week 8, the average difference between "efficiency" and "wins" in the rankings was about seven ranks. This week, the difference is about five ranks.

One thing these rankings don't signify is magnitude. I used a tier system recently to convey an interesting quirk of the rankings, but here we can use it to group teams in order of efficiency to mark how significant the differences between teams really are.

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

It really is that stratified at the top. The efficiency numbers indicate that Carolina walks into every game with a nearly 20-point advantage over an average opponent (although this would be mitigated by the fact that Carolina has the fewest drive per game of any team in the NFL, making their differences seem smaller by about 4.5 points).

I'm not entirely sure what to make of the fact that the Denver Broncos are in the "fifth tier" of efficiency, as that feels wrong. The efficiency scores at Football Outsiders and Advanced NFL Stats consistently rank the Denver Broncos in the top two, which feels more correct.

I suspect the incorporation of drive stats (an inverse "first down counter" that counts punts and field goals) will resolve a lot of this. I also think the Bengals' statistics would have them ranked higher, especially after their 49-9 win over the Jets (who beat the Saints, most notably) but you get what you get.

One unusual fact of the rankings is that the Packers do not drop all that much. Take a look at the recency-adjusted rankings (headers sortable, etc):

Power Rankings - Recency Bias
Team	Efficiency	Points	Wins	Average	Average Rank
Carolina Panthers	1	1	5	2.33	1
Seattle Seahawks	2	5	2	3.00	2
New Orleans Saints	3	4	4	3.67	3
Kansas City Chiefs	4	6	1	3.67	3
San Francisco 49ers	5	2	6	4.33	5
Denver Broncos	8	3	3	4.67	6
New England Patriots	10	7	7	8.00	7
Arizona Cardinals	7	10	10	9.00	8
Green Bay Packers	9	9	12	10.00	9
Dallas Cowboys	6	12	13	10.33	10
Cincinnati Bengals	14	8	11	11.00	11
Indianapolis Colts	12	13	9	11.33	12
Detroit Lions	15	11	8	11.33	12
Philadelphia Eagles	11	15	14	13.33	14
Chicago Bears	13	16	15	14.67	15
St. Louis Rams	16	14	20	16.67	16
Tennessee Titans	17	18	19	18.00	17
New York Giants	20	23	17	20.00	18
Cleveland Browns	23	21	18	20.67	18
Miami Dolphins	24	20	23	22.33	20
San Diego Chargers	27	19	22	22.67	21
Baltimore Ravens	29	17	22	22.67	21
Washington Redskins	18	28	24	23.33	23
Buffalo Bills	21	24	26	23.67	24
New York Jets	25	30	16	23.67	24
Tampa Bay Buccaneers	19	22	32	24.33	26
Houston Texans	22	25	29	25.33	27
Atlanta Falcons	26	26	28	26.67	28
Pittsburgh Steelers	31	27	27	28.33	29
Oakland Raiders	30	31	25	28.67	30
Minnesota Vikings	28	29	31	29.33	31
Jacksonville Jaguars	32	32	30	31.33	32

I looked at that case a bit more closely, and it is interesting for two reasons: the first is that the Packers' Week 6 performance (against Baltimore, not a well-regarded opponent in these rankings) was pretty poor and already weighed them down. The second interesting thing is that the Packers defense had been pretty good in Weeks 7, 8 and 9 (Cleveland Browns, Minnesota Vikings and Chicago Bears—all against QBs with fewer starts than there have been weeks in their schedules), as their offense started diving, hiding their issues.

I doubt Matt Flynn will revive their passing fortunes, Lions game aside. So we may see a more precipitous drop in the future.

Of note, the Panthers' recent wins have allowed their win totals to converge quickly with their efficiency in recent weeks and could signal the Panthers' ability to take the top spot in efficiency by season's end.

Naturally, there are no promises, as the Jaguars amply demonstrated this last week.

The models produce a few teams to watch for and a few teams to be wary of. Right now, the "most undervalued" teams are the Tampa Bay Buccaneers (I've mentioned in the past that the disconnect between coaching and talent can speak to the difference between efficiency and wins), the Dallas Cowboys, the Buffalo Bills and the Carolina Panthers (still). The "overvalued teams" are the New York Jets (who do in fact have a number of good wins), the Baltimore Ravens, the New England Patriots and the Miami Dolphins (but you already knew that).

In general, I may have to increase the overall value of passing and decrease the importance of running in these models to consistently create predictive value, but the models are not doing too badly so far.

Against the Spread

Week 10 ATS

Efficiency: 7-7
Efficiency ("What Have You Done For Me Lately"): 9-5
Points: 7-7
Points (WHY4DML): 5-9

Overall ATS

Efficiency: 34-35
Efficiency (WHYD4ML): 39-30
Points: 34-35
Points (WHY4DML): 33-36

The more I think about it, the more OK I am with the mediocre showing against the spread, given the haphazard nature of the construction of the efficiency scores and the unusual amount of injuries in the league this year, especially at QB. Next year should be better for a number of reasons.

Also, the models are doing a pretty decent job of picking the winner

Straight Up

Betting Market: 39-30
Efficiency: 43-26
Efficiency (WHYD4ML): 41-28
Points: 37-32
Points (WHY4DML): 37-32

As of right now, the efficiency model would do well on moneyline bets, and would come out ahead if you only bet the Vegas underdogs the model thinks would win outright (hitting them ten out of twenty-one opportunities, more than enough to make money on only underdogs, even with the vig).

Picks ATS

Efficiency

Indianapolis over TENNESSEE (+3)
BUFFALO (-1) over New York Jets
CHICAGO (-2) over Baltimore
CINCINNATI (-6) over Cleveland
PHILADELPHIA (-3.5) over Washington
Detroit over PITTSBURGH (+1.5)
TAMPA BAY over Atlanta (+0)
Arizona over JACKSONVILLE (+7)
Oakland over HOUSTON (-7)
San Diego over MIAMI (-1)
San Francisco over NEW ORLEANS (-3.5)
Green Bay over NEW YORK GIANTS (-7)
SEATTLE (-14) over Minnesota
Kansas City over DENVER (-8)
CAROLINA (-1) over New England

Efficiency (WHYD4ML)

Indianapolis over TENNESSEE (+3)
BUFFALO (-1) over New York Jets
CHICAGO (-2) over Baltimore
CINCINNATI (-6) over Cleveland
Washington over PHILADELPHIA (-3.5)
Detroit over PITTSBURGH (+1.5)
Atlanta (+0) over TAMPA BAY
Arizona over JACKSONVILLE (+7)
Oakland over HOUSTON (-7)
San Diego over MIAMI (-1)
San Francisco over NEW ORLEANS (-3.5)
Green Bay over NEW YORK GIANTS (-7)
SEATTLE (-14) over Minnesota
Kansas City over DENVER (-8)
CAROLINA (-1) over New England

Points

TENNESSEE (+3) over Indianapolis
BUFFALO (-1) over New York Jets
CHICAGO (-2) over Baltimore
CINCINNATI (-6) over Cleveland
PHILADELPHIA (-3.5) over Washington
Detroit over PITTSBURGH (+1.5)
TAMPA BAY over Atlanta (+0)
Arizona over JACKSONVILLE (+7)
Oakland over HOUSTON (-7)
MIAMI (-1) over San Diego
NEW ORLEANS (-3.5) over San Francisco
Green Bay over NEW YORK GIANTS (-7)
SEATTLE (-14) over Minnesota
Kansas City over DENVER (-8)
CAROLINA (-1) over New England

If they try hard enough, the Jaguars could be the worst team in NFL history. It certainly seems like they're still on pace for the worst average point differential since the merger, despite a win where they were averaging a 9.1 point differential!

We also could have included an AAFC team—the 1948 Chicago Rockets, who were abysmal—but it was a stretch to include pre-1960 teams in the first place (no AFL team was bad enough to get on the list either way) given our conception of modern football and its place in history.

The 2011 Buccaneers may be a surprise until we remember that the scores above are not adjusted for strength-of-schedule, and the Bucs had to go up against Drew Brees' record-setting season twice, along with an efficient year from Matt Ryan.

If you'll remember, that 2011 Buccaneers team DID beat the 2011 Minnesota Vikings.

Notable absences on that list are the 2012 Kansas City Chiefs, who were known for not having a second of gametime with a lead until late in the season, despite a win (it occurred in overtime against the New Orleans Saints, so no seconds were played with a lead as the game ended as soon as points were scored), the 2011 Indianapolis Colts (who were awful) and the 2008 Detroit Lions.

The most useful thing about these "game scripts" may be their ability to determine run-pass balance. It's a fairly simple method that doesn't seem to work quite as well at the extreme margins of point differential, but otherwise is really good at parsing out the fact that teams run when they're ahead and pass when they're behind.

Taking that into account, we can look at a team's run-pass balance from the perspective of their response to the clock. The scores are modified to treat "100" as average and every change in movement by 15 to be one standard deviation:

Rank	Team	Passer ID Score
1	Denver Broncos	134.7
2	Dallas Cowboys	131.8
3	New Orleans Saints	131.7
4	Atlanta Falcons	124.3
5	Kansas City Chiefs	115.6
6	Cleveland Browns	115.4
7	New England Patriots	111.6
8	Detroit Lions	111.0
9	Green Bay Packers	109.7
10	Miami Dolphins	109.6
11	Cincinnati Bengals	109.2
12	Carolina Panthers	106.0
13	Chicago Bears	105.5
14	San Diego Chargers	101.4
15	Pittsburgh Steelers	101.2
16	Baltimore Ravens	101.0
17	Arizona Cardinals	100.2
18	Indianapolis Colts	96.4
19	St. Louis Rams	94.9
20	New York Giants	94.2
21	Minnesota Vikings	92.5
22	San Francisco 49ers	92.2
23	Tampa Bay Buccaneers	91.3
24	Philadelphia Eagles	89.3
25	Houston Texans	88.6
26	Seattle Seahawks	87.5
27	Oakland Raiders	86.3
28	Tennessee Titans	86.0
29	Buffalo Bills	78.9
30	Washington Redskins	75.2
31	New York Jets	63.6
32	Jacksonville Jaguars	62.9

What this means is that the Denver Bronco and the Dallas Cowboys are particularly pass-heavy, while the Jets and Jaguars are very run-heavy, once you take into account time left and the score. For the most part this is very unsurprising. Run-heavy teams like the Eagles, 49ers, Texans and Seahawks are near the bottom of passer ID score, although as a Vikings fan you'd expect them to be even lower given that Adrian Peterson is the best running back in the league and their QB of the Week isn't really the focal point of their offense.

You can also use this to see what offenses are choosing to do against certain defenses.

Rank	Team	Passer ID Against
1	New York Jets	136.7
2	Arizona Cardinals	125.6
3	Minnesota Vikings	119.5
4	Washington Redskins	119.3
5	Detroit Lions	119.3
6	New York Giants	115.8
7	Philadelphia Eagles	113.0
8	Jacksonville Jaguars	112.2
9	Cleveland Browns	112.1
10	Tampa Bay Buccaneers	107.3
11	Dallas Cowboys	106.2
12	Baltimore Ravens	104.7
13	San Diego Chargers	102.4
14	Oakland Raiders	100.7
15	Miami Dolphins	100.3
16	Buffalo Bills	99.9
17	Kansas City Chiefs	97.8
18	Cincinnati Bengals	97.0
19	Atlanta Falcons	95.0
20	St. Louis Rams	92.4
21	Carolina Panthers	92.2
22	Tennessee Titans	90.9
23	Denver Broncos	89.3
24	Pittsburgh Steelers	88.7
25	Indianapolis Colts	87.9
26	New Orleans Saints	85.8
27	Houston Texans	85.0
28	Green Bay Packers	84.5
29	New England Patriots	83.0
30	Chicago Bears	82.0
31	Seattle Seahawks	80.9
32	San Francisco 49ers	72.6

Roughly this reads like this: people would much prefer to pass on the New York Jets and run on the San Francisco 49ers.

For the most part this makes sense: teams want to keep "quarterbacks off the field' (I have my own issues with that, but whatever), so teams with the top quarterbacks (although not necessarily the most efficient passers this year) rank "low" in passer identity against: the Broncos (23rd), Saints (26th), Packers (28th) and Patriots (29th).

It seems odd that the 49ers have such a heavy identity of teams running against them, but they have had a schedule against the Colts (18th), Rams (19th), Seahawks (26th), Titans (28th) and Jaguars (32nd). On average, teams run 26.9 running plays a game and 35.7 passing plays a game. The 49ers have had to deal with nearly exactly that despite ranking first overall in unadjusted game script—where you would expect significantly more passes.

Projecting Wins

Last week, we touched on "remaining strength of schedule" to look at what was upcoming for teams. With the efficiency scores, we can project future wins as well using that model:

Projected Wins
Team	Current Wins	Future SOS Rk	Projected Wins	Total	Rank
Kansas City Chiefs	9	19	5.6	14.6	1
Seattle Seahawks	9	12	5.3	14.3	2
Carolina Panthers	6	8	6.4	12.4	3
Denver Broncos	8	16	3.9	11.9	4
Detroit Lions	6	30	5.7	11.7	5
New England Patriots	7	17	4.3	11.3	6
New Orleans Saints	7	1	3.5	10.5	7
Indianapolis Colts	6	18	4.3	10.3	8
Green Bay Packers	5	23	4.4	9.9	9
San Francisco 49ers	6	5	3.6	9.6	10
Chicago Bears	5	23	4.8	9.9	11
Dallas Cowboys	5	24	4.4	9.4	12
Arizona Cardinals	5	9	4.1	9.1	13
Cincinnati Bengals	6	32	2.9	8.9	14
Philadelphia Eagles	5	11	3.5	8.5	15
New York Jets	5	26	3.3	8.3	16
Tennessee Titans	4	20	4.0	8.0	17
Cleveland Browns	4	31	3.1	7.1	18
Buffalo Bills	3	29	4.0	7.0	19
Miami Dolphins	4	21	2.8	6.8	20
Houston Texans	2	28	4.0	6.0	21
San Diego Chargers	4	15	2.0	6.0	22
Washington Redskins	3	13	2.9	5.9	23
Baltimore Ravens	4	25	5.6	5.6	24
St. Louis Rams	4	2	1.4	5.4	25
New York Giants	3	7	1.6	4.6	26
Oakland Raiders	3	10	0.4	3.4	27
Pittsburgh Steelers	3	22	0.4	3.4	28
Tampa Bay Buccaneers	1	4	2.4	3.4	29
Atlanta Falcons	2	3	1.0	3.0	30
Minnesota Vikings	2	6	0.7	2.7	31
Jacksonville Jaguars	1	14	0.0	1.0	32

You won't expect teams with significant differences in their makeup from the first week and the most recent week to continue at this pace (Green Bay, Chicago, Minnesota, Cleveland, etc.) but this should be a decent approximation of future win totals when blind to context.

I was surprised to see the New Orleans Saints as low as they were given their newfound defensive prowess, but a quick look at their schedule reveals why: a game against the 49ers, Seahawks and two games against the Panthers. In addition, the model overvalues Tampa Bay. Tough row to hoe.

On the other hand, it should be disappointing for the Bengals if they get the record that they are projected to get given the strength of their schedule, even if 9-7 is supposed to be respectable (and good enough for a playoff berth in that division). Ditto Browns.

This projects the strongest division to be the NFC West (huge surprise) and the weakest division to be the AFC North, who have dropped as quickly as the NFC West has risen, it seems. The AFC South seems sorry as well, and the poor Texans only would have one year in the limelight as the Colts return to prominence.

Without taking into account the actual tiebreakers for the playoffs, here are the projected playoff teams:

AFC

(1) Kansas City Chiefs
(2) New England Patriots
(3) Indianapolis Colts
(4) Cincinnati Bengals
(5) Denver Broncos
(6) New York Jets

NFC

(1) Seattle Seahawks
(2) Carolina Panthers
(3) Detroit Lions
(4) Dallas Cowboys
(5) New Orleans Saints
(6) Green Bay Packers

Like I've said, I doubt Green Bay makes it with Matt Flynn/Scott Tolzien and fully expect San Francisco to take that spot. The AFC's top four teams are very strong, but two of them are in the same division, so two "weak" AFC teams enter as a result (Bengals and Jets). The NFC would only have one "weak" team in the Cowboys.

But they don't play the games on my computer, so it will be interesting to see how the games actually develop.