Blind Power Rankings: Week 7, Plus College Football Idiocy

There didn't seem to be too many upsets last week (three according to the betting lines—but I don't know if anyone really thinks of Pittsburgh beating New York's less-storied team counts), but there were a lot of changes in the per-play numbers of a lot of teams, which precipitated more movement than a predictable week would normally involve.

St. Louis, Dallas and San Diego moved up significantly. San Diego and Dallas seem to be doing through offensive efficiency, while St. Louis simply stopped being bad on both sides of the ball and moved closer to average after an impressive performance against a quickly sinking Houston.

Teams have had some recent changes that aren't reflected in their rankings and therefore the subsequent strength-of-schedules, and Houston might be the poster boy for that, after moving from Matt Schaub to T.J. Yates without any real improvement.

Indianapolis' acquisition of Trent Richardson has seemingly increased it's run success rate, so Indianapolis' offensive efficiency may not fully be captured yet.

Cincinnati's rise is less due to a change in performance than a change in opponents, as most of their rise has to due with an increase in strength-of-schedule than anything else.

The opposite is true of the New York's disfavored (but somehow better) team, whose performance fell dramatically due to a decrease in efficiency and a worse opponent.

Everybody who has played Seattle or Kansas has received basically a one-point increase in their strength-of-schedule, which is somewhat significant. It helped Tennessee keep their eighth spot in efficiency despite a dismal performance and Oakland moved up two spots despite giving up ten sacks.

If you want a refresher on how the rankings were calculated or what they include, take a look here.

Again, I prefer an efficiency metric to rank teams as it generally does a good job of measuring team quality instead of past performance, and tends to reduce the influence of random events. But all the headers are sortable, so you can choose whichever method suits you.

Team	Efficiency	Points	Wins	Average	Average Rank
Kansas City Chiefs	2	2	2	2.00	1
Seattle Seahawks	1	3	4	2.67	2
Denver Broncos	7	1	1	3.00	3
New Orleans Saints	4	5	3	4.00	4
Indianapolis Colts	5	4	7	5.33	5
San Francisco 49ers	10	6	6	7.33	6
New England Patriots	14	10	5	9.67	7
Detroit Lions	6	13	11	10.00	8
Green Bay Packers	17	8	10	11.67	9
Tennessee Titans	8	14	14	12.00	10
Dallas Cowboys	11	7	18	12.00	10
Carolina Panthers	3	9	26	12.67	12
Cincinnati Bengals	16	16	9	13.67	13
Chicago Bears	15	17	13	15.00	14
Miami Dolphins	26	11	8	15.00	14
Arizona Cardinals	12	18	17	15.67	16
Cleveland Browns	18	19	15	17.33	17
Baltimore Ravens	29	12	12	17.67	18
Philadelphia Eagles	13	20	21	18.00	19
Buffalo Bills	9	23	24	18.67	20
San Diego Chargers	28	15	16	19.67	21
Oakland Raiders	21	22	22	21.67	22
St. Louis Rams	22	24	19	21.67	22
Atlanta Falcons	24	21	25	23.33	24
Washington Redskins	19	25	27	23.67	25
New York Jets	25	28	23	25.33	26
Houston Texans	27	30	20	25.67	27
Tampa Bay Buccaneers	20	26	32	26.00	28
Minnesota Vikings	23	27	30	26.67	29
Jacksonville Jaguars	30	32	28	30.00	30
Pittsburgh Steelers	32	29	29	30.00	30
New York Giants	31	31	31	31.00	32

There's a decent way to tell if a team is underperforming relative to their talent or overperforming, and that is simply by subtracting their efficiency rank from their "wins" rank.

In this case, you can tell that the Panthers had been underperforming until very, very recently. They have the largest difference between efficiency and wins. On the other side, the Miami Dolphins have been overperforming, with an eighth overall ranking in wins but a 26th overall ranking in efficiency.

Other big underperformers include the Buffalo Bills, the Tampa Bay Buccaneers and the Dallas Cowboys.

Overperformers include the Baltimore Ravens, San Diego Chargers, the New England Patriots, the Cincinnati Bengals and the Houston Texans.

There are two prevailing reasons a team will over- or underperform. The first is the most obvious: luck. A team might have an unusually high fumble recovery rate (it has always regressed to 50%), have the ball bounce the right way or have an unusual number of favorable calls. Other things that aren't accounted for include special teams performance, like good returns and a rangy kicker—while these aren't "luck," they are functionally "luck" for the purposes of this discussion.

The second reason is coaching. Is anyone really surprised that the Ravens and the Patriots are overperforming their talent? This is likely a more sustainable way to consistently have efficiency lag behind wins. On the other hand, the Panthers have consistently had disparate efficiency and wins scores and could very well shoulder the blame on Ron Rivera. He's well known for being too conservative in late-game decisions and notoriously poor with clock management.

I do not need to go into much discussion regarding the Buccaneers.

If you prefer recency weighting, perhaps to capture emerging trends in the data and emphasize relative health, here you go:

Team	Efficiency	Points	Wins	Average	Average Rank
Kansas City Chiefs	3	2	2	2.33	1
Seattle Seahawks	1	3	4	2.67	2
Denver Broncos	11	1	1	4.33	3
San Francisco 49ers	7	5	3	5.00	4
New Orleans Saints	4	6	6	5.33	5
Indianapolis Colts	6	4	9	6.33	6
Detroit Lions	5	11	10	8.67	7
Green Bay Packers	17	7	5	9.67	8
Dallas Cowboys	9	8	17	11.33	9
New England Patriots	15	12	8	11.67	10
Carolina Panthers	2	9	24	11.67	10
Cincinnati Bengals	14	15	7	12.00	12
Arizona Cardinals	10	17	15	14.00	13
Chicago Bears	13	18	13	14.67	14
Baltimore Ravens	28	10	12	16.67	15
Philadelphia Eagles	12	20	18	16.67	15
Tennessee Titans	16	16	19	17.00	17
Cleveland Indians	18	19	14	17.00	17
Miami Dolphins	27	14	11	17.33	19
Buffalo Bills	8	24	23	18.33	20
San Diego Chargers	26	13	16	18.33	20
St. Louis Rams	20	23	20	21.00	22
Oakland Raiders	21	22	21	21.33	23
Washington Redskins	19	25	26	23.33	24
Atlanta Falcons	24	21	27	24.00	25
New York Jets	25	29	25	26.33	26
Minnesota Vikings	23	27	29	26.33	26
Houston Texans	29	30	22	27.00	28
Tampa Bay Buccaneers	22	28	32	27.33	29
Pittsburgh Steelers	32	26	28	28.67	30
New York Giants	30	31	31	30.67	31
Jacksonville Jaguars	31	32	30	31.00	32

Evidently, laying only 16 points on the Jaguars was not good for Denver in the model. The table above seems to imply, shockingly, that Panthers' recent victory did them no good in the overall ranking, but that's not quite true. It has a similar average ranking as in the first table, but it simply is outweighed by the recent boosts of the teams above them that were washed out when the model eliminated recency bias.

Record against the spread:

Efficiency: 5-11
Points: 7-7
Efficiency (What Have You Done For Me Lately): 7-7
Points (WHYD4ML): 7-7

That is disappointing, I know.

Efficiency games against the spread:

Seattle over ARIZONA (+7)
ATLANTA (-8) over Tampa Bay
DETROIT (-3) over Cincinnati
Buffalo over MIAMI (-8)
New England over NEW YORK JETS (+5)
Dallas over PHILADELPHIA (-2.5)
Chicago over WASHINGTON (+0)
CAROLINA (-6) over St. Louis
JACKSONVILLE (+7.5) over San Diego
TENNESSEE (+4.5) over San Francisco
Cleveland over GREEN BAY (-10)
KANSAS CITY over Houston (-7)
Baltimore over PITTSBURGH (-2)
INDIANAPOLIS (+6.5) over Denver
Minnesota over NEW YORK GIANTS (-3.5)

Efficiency games against the spread (WHYD4ML):

Seattle over ARIZONA (+7)
ATLANTA (-8) over Tampa Bay
Cincinnati over DETROIT (-3)
Buffalo over MIAMI (-8)
New England over NEW YORK JETS (+5)
Dallas over PHILADELPHIA (-2.5)
Chicago over WASHINGTON (+0)
CAROLINA (-6) over St. Louis
JACKSONVILLE (+7.5) over San Diego
PUSH: San Francisco at TENNESSEE (+4.5)
Cleveland over GREEN BAY (-10)
KANSAS CITY over Houston (-7)
Baltimore over PITTSBURGH (-2)
INDIANAPOLIS (+6.5) over Denver
Minnesota over NEW YORK GIANTS (-3.5)

Points against the spread:

Seattle over ARIZONA (+7)
ATLANTA (-8) over Tampa Bay
DETROIT (-3) over Cincinnati
MIAMI (-8) over Buffalo
New England over NEW YORK JETS (+5)
Dallas over PHILADELPHIA (-2.5)
Chicago over WASHINGTON (+0)
CAROLINA (-6) over St. Louis
San Diego over JACKSONVILLE (+7.5)
TENNESSEE (+4.5) over San Francisco
Cleveland over GREEN BAY (-10)
KANSAS CITY over Houston (-7)
Baltimore over PITTSBURGH (-2)
INDIANAPOLIS (+6.5) over Denver
Minnesota over NEW YORK GIANTS (-3.5)

Points against the spread (WHYD4ML):

Seattle over ARIZONA (+7)
Tampa Bay over ATLANTA (-8)
DETROIT (-3) over Cincinnati
MIAMI (-8) over BUFFALO
New England over NEW YORK JETS (+5)
Dallas over PHILADELPHIA (-2.5)
Chicago over WASHINGTON (+0)
CAROLINA (-6) over St. Louis
San Diego over JACKSONVILLE (+7.5)
San Francisco over TENNESSEE (+4.5)
GREEN BAY (-10) over Cleveland
KANSAS CITY over Houston (-7)
Baltimore over PITTSBURGH (-2)
INDIANAPOLIS (+6.5) over Denver
Minnesota over NEW YORK GIANTS (-3.5)

I wanted to see if a more extreme situation created a different kind of ranking, so I applied the same points system to college football, and included every game between FBS schools, any games between an FBS school and an FCS school and any FCS games that included two teams that have played an FBS school. It produced the following unusual ranking that should give us some clues as to why the professional model could need tweaking.

Rank	Team
1	Florida State
2	Oregon
3	Baylor
4	Clemson
5	Alabama
6	Washington
7	Louisiana State
8	Georgia
9	Louisville
10	Missouri
11	UCLA
12	Stanford
13	Arizona State
14	Wisconsin
15	South Carolina
16	Ohio State
17	Georgia Tech
18	Florida
19	Utah
20	Auburn
21	Miami (FL)
22	Eastern Illinois
23	Texas A&M
24	BYU
25	Boise State
26	Coastal Carolina
27	Nebraska
28	Virginia Tech
29	Oklahoma State
30	Marshall
31	Mississippi
32	South Carolina State
33	Arizona
34	Central Florida
35	Indiana
36	Utah State
37	Oregon State
38	Kansas State
39	Texas Tech
40	Illinois
41	Southern California
42	Mississippi State
43	North Dakota State
44	Penn State
45	Tennessee
46	Northern Iowa
47	Oklahoma
48	Syracuse
49	Pittsburgh
50	Michigan

The extremely high rankings of Eastern Illinois, Coastal Carolina, Marshall and South Carolina State concern me, but none more so than Georgia Tech. At least EIU has Jimmy Garappolo.

Ohio State's low ranking is odd, too, and it kind of sucks not seeing Northern Illinois in there. MIchigan is clearly too low and Michigan State didn't even make the top 50. I suppose playing in the B1G isn't that good for this system (so far). Fresno State is missing as well, and Oklahoma took a huge hit.

I do enjoy that North Dakota State is not the top FCS school, but only half of their games were included.

To me, this means that the system works best when there are relatively congruent data sets (In the NFL, I add 14-16 games every week usually. In college, the number can vary wildly from 140 to 180) and when teams are all supposed to play enough games to properly determine a strength of schedule. Not a lot of teams played Grambling State, but those that did had a good time.

Another thing to think about is that pace is so much more important than college, so not accounting for their effects leads to significant differences. Football Outsiders' Fremeau Efficiency Index makes a lot more sense, and that's what they do.

So, for next year, it looks like I will ditch "Point Differential" for "Point Differential Per Drive" (excluding kneeldowns) and incorporate "Drive Success Rate" into the efficiency scores, while decreasing "run success rate" and increasing "yards per carry," which is functionally not making a difference at all.