Writers of Pro Football Prospectus 2008

Most Recent FO Features

GurleyTod15.jpg

» OFI: Letdown Saturday

A week after big upsets of Stanford and Ohio State, the USC and Virginia Tech themselves fell to less-talented opponents. Georgia also fell to South Carolina after pummeling Clemson in the opener.

27 Sep 2005

Week 4 DVOA Ratings

by Aaron Schatz

Here are the first DVOA ratings of the year, and the commentary now resides on FOXSports.com. You'll find it here. We're still working on how much I'll do comments per team like the old FOX Power Ratings, and how much I'll do the longer essay that FO readers have come to know and love. For this week, I did a little of both.

If we end up with a little less hardcore DVOA analysis on Tuesdays, don't worry. I'm going to try to do a mailbag weekly this season, and that will let me toss some extra analysis at you on Fridays.

The last two years, I did not introduce opponent adjustments until Week 4. I decided that I would introduce them a week early this year so that I wouldn't have to explain DVOA to FOX readers without the D. The opponent adjustments are at 30% this week, and they'll go up 10% per week until Week 10.

The other change due to FOX: The ratings will now be titled based on the week to come, not the week past.

Oh, and we moved back to a dedicated server again, since we're getting hit by lots of new visitors. The current slowness should be gone by Wednesday.

* * * * *

Here are the Football Outsiders team efficiency ratings through Week 3 of 2005, measured by our proprietary Defense-adjusted Value Over Average (DVOA) system that breaks down every single play and compares a team's performance to the league averaged based on situation in order to determine value over average. (Explained further here.)

OFFENSE and DEFENSE DVOA are adjusted based on strength of opponent as well as to consider all fumbles, kept or lost, as equal value. Opponent adjustments are currently set at 30% and will increase each week until they are full strength after Week 10. SPECIAL DVOA is adjusted for type of stadium (warm, cold, dome, Denver/Mexico City) and week of season. NON-ADJ TOTAL VOA does not include these adjustments.
As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE. LAST WEEK represents rank after Week 2, even though there were no ratings posted to the website for Week 2.

All of the pages -- including individual player stats -- have now been updated for 2005 except offensive and defensive lines.


TEAM
TOTAL
DVOA
LAST
WEEK
NON-ADJ
TOTAL VOA
W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
SPECIAL
DVOA
S.T.
RANK
1 CIN 92.0% 3 98.9% 3-0 26.6% 5 -61.9% 1 3.4% 9
2 PIT 63.3% 1 70.9% 2-1 50.9% 1 -14.6% 8 -2.2% 21
3 TB 59.1% 6 70.0% 3-0 1.0% 15 -56.1% 2 2.0% 12
4 PHI 46.1% 4 35.2% 2-1 30.7% 4 -24.1% 5 -8.7% 27
5 NYG 35.8% 5 47.1% 2-1 21.7% 6 9.6% 22 23.7% 1
6 IND 31.7% 8 27.3% 3-0 14.7% 9 -19.3% 6 -2.3% 22
7 CHI 29.6% 2 20.6% 1-2 -26.4% 26 -50.0% 3 6.0% 6
8 MIA 28.9% 10 24.5% 2-1 -1.1% 16 -26.6% 4 3.4% 10
9 SEA 27.3% 14 21.5% 2-1 35.4% 2 5.4% 20 -2.7% 23
10 JAC 26.3% 12 38.1% 2-1 -3.1% 17 -18.1% 7 11.4% 3
11 SD 21.1% 16 14.0% 1-2 32.9% 3 15.6% 25 3.9% 7
12 DAL 11.8% 11 23.4% 2-1 13.7% 10 1.0% 17 -1.0% 17
13 KC 11.4% 7 5.0% 2-1 18.3% 8 7.0% 21 0.1% 15
14 NE 10.5% 18 -0.7% 2-1 6.7% 13 -4.2% 12 -0.5% 16
15 BUF 6.2% 9 9.5% 1-2 -14.1% 21 -9.4% 10 10.9% 4
16 STL 5.1% 15 11.7% 2-1 8.8% 12 0.0% 14 -3.7% 24
TEAM
TOTAL
DVOA
LAST
WEEK
NON-ADJ
TOTAL VOA
W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
SPECIAL
DVOA
S.T.
RANK
17 ATL 4.0% 22 13.3% 2-1 10.6% 11 4.5% 19 -2.1% 20
18 CAR 1.9% 13 8.4% 1-2 -6.6% 19 -5.5% 11 2.9% 11
19 DEN -7.7% 25 1.0% 2-1 -5.9% 18 0.3% 16 -1.6% 19
20 CLE -9.7% 17 -12.5% 1-2 4.6% 14 17.8% 27 3.5% 8
21 OAK -13.3% 21 -12.0% 0-3 20.5% 7 25.0% 30 -8.8% 28
22 WAS -18.0% 19 -18.2% 2-0 -18.0% 23 -11.6% 9 -11.6% 31
23 TEN -19.7% 23 -14.2% 1-2 -19.0% 25 9.6% 23 8.9% 5
24 NYJ -36.3% 24 -19.8% 1-2 -37.0% 28 -1.8% 13 -1.0% 18
25 GB -38.4% 26 -44.0% 0-3 -18.2% 24 14.7% 24 -5.5% 26
26 NO -38.8% 20 -48.0% 1-2 -9.8% 20 24.0% 28 -5.0% 25
27 MIN -40.3% 32 -46.0% 1-2 -41.8% 29 0.0% 15 1.5% 13
28 SF -46.9% 29 -49.3% 1-2 -27.3% 27 32.1% 31 12.6% 2
29 ARI -48.7% 27 -52.1% 0-3 -14.7% 22 24.9% 29 -9.1% 29
30 DET -66.7% 28 -68.7% 1-1 -51.2% 31 3.0% 18 -12.6% 32
31 BAL -70.9% 30 -69.1% 0-2 -44.4% 30 15.7% 26 -10.8% 30
32 HOU -95.3% 31 -106.8% 0-2 -53.2% 32 42.2% 32 0.1% 14

  • FOX RANK represents the FOXSports.com Power Ratings which are one-third 2005 DVOA, one-third 2005 pre-season projection, and one-third a special weighted DVOA for 2004 that includes the playoffs.
  • ESTIMATED WINS uses a statistic known as "Forest Index" that emphasizes consistency as well as DVOA in the most important specific situations: red zone defense, first quarter offense, and performance in the second half when the score is close.  It then projects a number of wins adjusted to a league-average schedule and a league-average rate of recovering fumbles.  It is based on the number of games a team has played so far, so teams which have not had their bye will appear higher.
  • PAST SCHEDULE lists average DVOA of opponents played this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • VARIANCE measures the statistical variance of the team's weekly DVOA performance. Teams are ranked from least consistent (#1, highest variance) to most consistent (#32, smallest variance).

TEAM TOTAL
DVOA
W-L FOX

RANK
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VARIANCE RANK
1 CIN 92.0% 3-0 3 3.0 1 -6.8% 22 -13.0% 29 13.8% 22
2 PIT 63.3% 2-1 1 2.6 3 -34.8% 31 -1.1% 18 61.3% 8
3 TB 59.1% 3-0 5 3.0 2 -24.2% 30 -12.7% 28 1.8% 31
4 PHI 46.1% 2-1 2 2.1 9 -18.7% 27 2.3% 12 90.7% 6
5 NYG 35.8% 2-1 15 2.5 4 -22.1% 29 1.2% 15 48.7% 9
6 IND 31.7% 3-0 4 2.4 5 -18.1% 26 -6.1% 23 7.4% 25
7 CHI 29.6% 1-2 21 1.7 14 2.4% 15 -20.2% 32 155.3% 1
8 MIA 28.9% 2-1 14 2.2 7 -14.1% 24 0.9% 17 9.5% 24
9 SEA 27.3% 2-1 9 2.3 6 -6.1% 21 -13.7% 30 35.3% 12
10 JAC 26.3% 2-1 13 2.1 10 7.6% 13 -17.1% 31 14.1% 21
11 SD 21.1% 1-2 8 2.1 8 13.3% 12 9.3% 3 16.5% 20
12 DAL 11.8% 2-1 16 1.7 13 -14.6% 25 4.2% 8 2.8% 30
13 KC 11.4% 2-1 11 1.8 12 -19.1% 28 10.7% 2 24.3% 15
14 NE 10.5% 2-1 6 2.0 11 17.3% 11 6.0% 4 19.7% 16
15 BUF 6.2% 1-2 7 1.0 21 -10.8% 23 5.6% 5 108.6% 4
16 STL 5.1% 2-1 20 1.5 18 -38.4% 32 -6.3% 24 5.4% 27
TEAM TOTAL
DVOA
W-L FOX

RANK
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VARIANCE RANK
17 ATL 4.0% 2-1 17 1.7 16 26.5% 4 -5.3% 21 28.6% 13
18 CAR 1.9% 1-2 10 1.7 15 0.2% 16 -7.3% 25 16.7% 19
19 DEN -7.7% 2-1 12 1.3 19 20.5% 10 1.3% 14 25.9% 14
20 CLE -9.7% 1-2 27 1.6 17 28.4% 3 -5.7% 22 19.6% 17
21 OAK -13.3% 0-3 22 1.1 20 22.7% 6 2.9% 9 3.6% 29
22 WAS -18.0% 2-0 19 0.6 26 20.7% 9 13.1% 1 4.0% 28
23 TEN -19.7% 1-2 25 0.7 25 -0.8% 17 -3.5% 20 110.7% 3
24 NYJ -36.3% 1-2 18 0.9 23 22.2% 7 1.3% 13 11.2% 23
25 GB -38.4% 0-3 28 0.5 27 -5.8% 20 2.8% 10 0.1% 32
26 NO -38.8% 1-2 24 1.0 22 -0.9% 18 5.1% 6 44.6% 10
27 MIN -40.3% 1-2 23 0.4 28 37.4% 1 -9.3% 26 84.8% 7
28 SF -46.9% 1-2 32 0.8 24 21.0% 8 0.9% 16 97.5% 5
29 ARI -48.7% 0-3 29 0.1 30 22.8% 5 -11.0% 27 6.1% 26
30 DET -66.7% 1-1 30 0.3 29 -4.4% 19 -1.8% 19 132.7% 2
31 BAL -70.9% 0-2 26 0.0 31 6.0% 14 4.5% 7 18.0% 18
32 HOU -95.3% 0-2 31 0.0 32 34.7% 2 2.6% 11 39.0% 11

Posted by: Aaron Schatz on 27 Sep 2005

62 comments, Last at 03 Oct 2005, 4:21pm by Freddie Falcon

Comments

1
by JasonK (not verified) :: Tue, 09/27/2005 - 5:58pm

Nice!

Am I correct in assuming that the FOX Rank calculations will shift to be more DVOA-dependant as the preseason projections and 2004 results become less and less relevant?

2
by pm (not verified) :: Tue, 09/27/2005 - 6:00pm

When does it appear on foxsports.com.

3
by Pat (not verified) :: Tue, 09/27/2005 - 6:05pm

I said it elsewhere, and I'll say it again: CIN and TB are the early-season flukes. They'll drift downwards really quickly, especially once their opposition continues to drift downwards (Chicago, Buffalo).

4
by Pat on the Back (not verified) :: Tue, 09/27/2005 - 6:06pm

Hazzah! Cincy at number 1. Who would have thunk it?

5
by Pat (not verified) :: Tue, 09/27/2005 - 6:08pm

Oh, and note that Green Bay variance: 0.1%! That's not a good sign. Green Bay's sucked on almost every play.

6
by Xian (not verified) :: Tue, 09/27/2005 - 6:09pm

I'm thinking TB is fluke-ish, but I'm totally buying Cincinnati. Which probably means that they'll tank, but I'm up for a feel-good story right now.

I'm assuming the GB Special Teams rating is due to:
a) Longwell's traditionally non-exceptional/slightly-worse-than-average kickoffs
b) two bad snaps/holds resulting in an aborted field goal and a missed XP.

Rather than, say, the punting? Though the punting could be mediocre too, but that's what I had specifically noticed about GB's special teams.

7
by Nick (not verified) :: Tue, 09/27/2005 - 6:27pm

How can the Bears have the #6 ranked special teams unit when Doug Brien is currently 1-4 on field goals? Wade has been good on punt returns, and Maynard is fine, but have they really been that good?

8
by Trogdor (not verified) :: Tue, 09/27/2005 - 6:31pm

Cincy is a good team, they were pretty good last year and seem to have improved on both offense and defense so far this year. However, I have to think they'll regress a little for one reason: turnovers. They aren't going to get 5 INTs a game, every game.

In the 1999 DVOA article, it was mentioned that perhaps the reward/penalty for turnovers is too high. This may also explain why Cincy is not just ranked 1st (which shouldn't be *that* surprising this year), but ranked so freaking far ahead of everyone else. My goodness, that is a huge lead. Even if they play really well the next few games, their DVOA will likely go down if they "only" force 2-3 turnovers. Or maybe I'm just on crack.

9
by B (not verified) :: Tue, 09/27/2005 - 6:39pm

Yea, TB should drop in the rankings, but it's not like the rest of the NFC (Philly excluded) is a bunch of world-beaters. Somebody has to play the Eagles in the championship game.

10
by Budman (not verified) :: Tue, 09/27/2005 - 6:39pm

Wow, it looks weird seeing NE so low. I have grown accustomed to seeing them at top last few years.

And I KNEW Cin would come out 1st after watching two of their games and seeing how well they have played both. But as others have pointed out expect a regression.

11
by Matthew Furtek (not verified) :: Tue, 09/27/2005 - 6:40pm

The 'Skins had the hardest schedule, won both games, yet are ranked 22nd in DVOA. Huh?

Or we could try this song... (staring at the NFL standings)... "One of these things is not like the others, one of these things just does not belong."

I'm just too giddy hoping they will beat Seattle at home on Sunday.

12
by Pat (not verified) :: Tue, 09/27/2005 - 6:42pm

Cincy and TB are almost the definition of an early season fluke that looks really good this year. Look at their past schedules: -10% on each! And it'd be a lot lower if Chicago wasn't fluky good in Week 2.

Cincy's played Cleveland (-9.7%), Minnesota (-40.3%), and Chicago (29.6%). Note the variance in the only good team they played (Chicago): 155%! That's an early season fluke. I'm sure they'll end up down at -10% to -20% on the year.

I'm not saying that Cincinnati's going to fall off the map, but they're going to drop to between, say, 8-15 by the end of the season.

But Dear God, could Cincinnati's schedule get any easier for the next weeks? Houston (at home!), Jacksonville, and Tennessee. That's 1 team out of their first 5 that I believe will have a positive DVOA by the end of the year.

Cincinnati will make the playoffs this year, definitely - especially since they've got 10 almost guaranteed wins on their schedule (the 3 already, 2X Baltimore, Houston, Tennessee, Cleveland, Green Bay, and Detroit). But I wouldn't count them much more than a wild card, and I doubt they'd get past the first round.

First four teams, and their past schedules:

CIN: -13%
PIT: -1.1%
TB: -12.7%
PHI: 2.3%

Two of these things are not like the others...

13
by Pat (not verified) :: Tue, 09/27/2005 - 6:46pm

The ‘Skins had the hardest schedule, won both games, yet are ranked 22nd in DVOA. Huh?

That's because they played really bad in both games. They shouldn't've won the second one. They did win, but that doesn't make you feel any better about the way they played.

The fact that the Skins had the hardest schedule doesn't really apply yet. We don't know if that's really true. They faced Dallas and Chicago: Dallas and Chicago *look like* strong teams, but Chicago also looks like a pile of steaming dung as well (155% variance!). Which are they? We'll find out in time. The adjustments for opponent are only at 30%, so that wouldn't affect their rankings.

14
by Sean (not verified) :: Tue, 09/27/2005 - 6:46pm

All Tampa is doing is finally meeting their projections. For the last two years, Tampa Bay has been one of the most underachieving teams in the league when you look at DVOA and their Pythagorean Win totals. The odds were strong that they were going to be notably improved this year, and the fact that they radically upgraded their personnel in the backfield doesn't hurt, either.

I won't say that the Bengals are the best team in the league, but I would say that it's possible. I think they would give any elite team a real handful, home or away.

15
by jim's apple pie (not verified) :: Tue, 09/27/2005 - 6:49pm

Here is the Dr. Z Memorial Ranking with Teams Always Above the Teams They've Beat:

1. Bengals
2. Colts
3. Bucs
4. Redskins
5. Jags
6. Seahawks
7. Falcons
8. Eagles
9. Dolphins
10. Broncos
11. Cowboys
12. Chargers
13. Chiefs
14. Giants
15. Vikings
16. Saints
17. Panthers
18. Patriots
19. Steelers
20. Bears
21. Lions
22. 49ers
23. Rams
24. Titans
25. Bills
26. Browns
27. Raiders
28. Ravens
29. Cardinals
30. Packers
31. Texans

Wild: Jets

Surprisingly (or maybe not, since it's only week 3), only the Jets are impossible to place, due to how the Jets/Chiefs/Broncos/Dolphins games played out.

One should also note the meaningless nature of this kind of ranking, since many teams have lost to a team with a lower DVOA and beat a team with a better DVOA:

1) SD has beat NYG, but lost to DEN and DAL.
2) PHI lost to ATL.
3) NE lost to CAR, but beat PIT.
etc.

Anyway, my powers of prognostication predict:

1) OAK beats DAL at home.
2) SD beats NE.
3) KC beats PHI in a rebound game for them.
4) There is no way to predict what is going to happen in the NYJ/BAL game. Stay far, far away from this one. Has there ever been a half with no first downs?

16
by Drew (not verified) :: Tue, 09/27/2005 - 7:15pm

The Colt offense rates 9th in DVOA? That's higher than I would have thought. My eyes tell me that they have been somewhat stymied and struggling so far this season. My first thought was "defensive adjustment", but the Baltimore and Cleveland defenses are rated low. Maybe I'm just spoiled, and struggling for them is still better than average.

17
by DavidH (not verified) :: Tue, 09/27/2005 - 7:29pm

The past and future schedule numbers seem to be listed in the wrong columns. That would the confusion in comment #11, and change the analysis in comment #12 (The Bengals have actually played the hardest schedule of the top 6 teams)

Of course, at this point 1/2 to 1/3 of an team's opponents' schedules is made up of that team, so it's kind of a self-fulfilling prophecy type deal (the good teams have played worse schedules, because the people they beat performed badly against them, lowering their ratings). Did that make sense?

18
by rk (not verified) :: Tue, 09/27/2005 - 7:48pm

#11, check out Washington's estimated wins (0.6). They've apparently been quite lucky.

19
by Pat (not verified) :: Tue, 09/27/2005 - 7:50pm

The past and future schedule numbers seem to be listed in the wrong columns.

Yah, I agree. I think there's a mistake there - I don't understand how Philly's DVOA can be *higher* than their VOA when they've only played below-average (or nearly below average) teams.

20
by CaffeineMan (not verified) :: Tue, 09/27/2005 - 8:42pm

Just thought I'd have fun pointing out that The Average Team is Carolina (1.9%).

And The Replacements are:

QB - Mike Vick (1.1)
RB - Ahman Green (0.1)
TE - Kris Mangum (Mr. Blutarski, 0.0)
WR - Marvin Harrison (!), Amani Toomer, and Plaxico Burress, all together with Mr. Blutarski.

I just love looking at stuff like this. What a geek.

21
by CaffeineMan (not verified) :: Tue, 09/27/2005 - 8:52pm

Oh, and while I'm at it:

The Average Offense is KC (1.1%) or Tampa Bay (-1.1%).

The Average Defense is St. Louis or Minnesota (both 0.0%).

The Average Special Teams is KC or Houston (both 0.1%).

22
by Aaron (not verified) :: Tue, 09/27/2005 - 9:00pm

Heh heh. Whoops, you are right. The past and future are reversed. I'll fix that. Sorry.

23
by CaffeineMan (not verified) :: Tue, 09/27/2005 - 9:22pm

Oh, excuse me all to pieces. I forgot to include rushing totals in my Replacement calculations. That boosts Vick out the hole, and makes Jake Delhomme the Replacement Quarterback (-1.8).

But my favorite so far is Drew Bledsoe 5th in QB Rushing DPAR, at 2.1, just behind Culpepper at 2.3. Drew, Drew, the Scramblin' Foo'...

24
by TomC (not verified) :: Tue, 09/27/2005 - 9:24pm

Good catch, DavidH. I could tell something was fishy about the Bears past-schedule number, but I didn't catch that the two were just reversed. Just to be sure, I toted up the Bears future number, and indeed it's -20.2%. And in doing that calculation, it came to my attention that the Bears next six opponents average -45.6%, and if you drop the TB and PIT games from the future schedule, the remaining 11 opponents average -29.6%. So if early-season DVOA means anything at all, you might want to see what kind of over/under you can get in Vegas this week for Chicago wins.

25
by Johonny (not verified) :: Tue, 09/27/2005 - 9:33pm

Why should I believe TB is a fluke. They had a good DVOA in 2002 and 2003. They were middle of the pack last year. It doesn't seem out of the question that they are a legitimate top 10 team this year. now if Miami finishes ahead of the Pats I'll declare Saban a genius. Oh wait the media already has...

26
by Vince (not verified) :: Tue, 09/27/2005 - 9:46pm

Yay! I'm so happy to have this stuff to pore through. Seriously, the NFL season seemed incomplete without it.

Note ATL's offensive line ratings compated to MINN's defensive line ratings. The Vikings are doomed this weekend. BWAHAHAHA!!!

Also, ATL's special teams rating show just how much Allen Rossum is missed.

27
by Basilicus (not verified) :: Tue, 09/27/2005 - 10:01pm

My incredibly amazing and scientific rankings, because I'm on a long road trip and need some way to take advantage of the wi-fi at Drury Inn.

1. Cincinnati Bengals 3-0
2. Tampa Bay Buccaneers 3-0
3. New England Patriots 2-1
4. Pittsburgh Steelers 2-1
5. Indianapolis Colts 3-0
6. Philadelphia Eagles 2-1
7. San Diego Chargers 1-2
8. Denver Broncos 2-1
9. Kansas City Chiefs 2-1
10. Washington Redskins 2-0
11. Carolina Panthers 1-2
12. Atlanta Falcons 2-1
13. Jacksonville Jaguars 2-1
14. Seattle Seahawks 2-1
15. Oakland Raiders 0-3
16. Miami Dolphins 2-1
17. Cleveland Browns 1-2
18. Tennessee Titans 1-2
19. New York Giants 2-1
20. Chicago Bears 1-2
21. St. Louis Rams 2-1
22. Dallas Cowboys 2-1
23. Buffalo Bills 1-2
24. San Francisco 49ers 1-2
25. Minnesota Vikings 1-2
26. New York Jets 1-2
27. New Orleans Saints 1-2
28. Green Bay Packers 0-3
29. Baltimore Ravens 0-2
30. Houston Texans 0-2
31. Detroit Lions 1-1
32. Arizona Cardinals 0-3

28
by Goldbach (not verified) :: Tue, 09/27/2005 - 10:08pm

I'm confused.

Philly's DVOA is greater than their VOA.

Cincinatti and Pittsburgh each have DVOA's less than their VOA's.

Cincinatti has had a harder schedule than Philly, and Pittsburgh has had an easier schedule than Philly.

How is Philly's DVOA higher than their VOA?

Also, Oakland, Arizona, Baltimore, Washington and Detroit all have worse special teams than a team which has had a kicker with 1 good leg for 1.5 out of 3 games. That's pretty sad.

29
by Aaron (not verified) :: Tue, 09/27/2005 - 10:44pm

I'll answer the Philly question here rather than waiting until the mailbag.

Non-adjusted VOA is different from DVOA in two ways. One is opponent strength. The other is that non-adjusted VOA only penalizes lost fumbles, not all fumbles. The Eagles have fumbled four times on offense and lost all four.

30
by Goldbach (not verified) :: Tue, 09/27/2005 - 11:12pm

Ah, thanks. I had forgotten about the fumbles.

31
by Dan (not verified) :: Wed, 09/28/2005 - 12:12am

Do you have the DVOA rankings from week 2, and will they ever be posted? I see that the Bears were 2nd then, and I'm curious to see how much of their week 2 blowout was offense.

32
by Fnor (not verified) :: Wed, 09/28/2005 - 12:23am

I think I see the opposite that everyone else does as far as Tampa and Cincinatti goes. I see Tampa being around for a while due to prior history and the Bengals falling off a bit.

Why? Because the Bengals' defensive DVOA, I believe, is skewed because of the ridiculous number of turnovers they've had thus far in the season (and even bled into the offense, because of good field position, which lets you open up your playbook, making it more likely to make a good drive yadda yadda). I don't think they can keep up anywhere near the level of turnover production as the season goes on. I think it'll still be high, but not ridiculously high as to keep their defensive DVOA in the top 5.

Although, I may be wrong. Perhaps I should go back and re-read the description of DVOA to get a sense of how much a turnover helps. But that's my take on it, at least.

33
by Josh (not verified) :: Wed, 09/28/2005 - 1:18am

Any update on when we can expect to see the power rankings on FoxSports.com?

34
by Bruce Dickinson (not verified) :: Wed, 09/28/2005 - 1:37am

When you look at TB's schedule, fluke or not, they should have a realistic chance of winning 11 games. Assuming they go 3-3 in division games, they can pick up the following wins:
against DET
at NYJ (who now suck)
against Chicago
possibly against Miami
against the 49ers
against the Redskins.

they do play NE in week 16 (i think) and their own division is going to be very unpredictable, but they have a lot of winnable games the rest of the way.

35
by james (not verified) :: Wed, 09/28/2005 - 1:41am

Cinci falling off?

Someone please research how many teams have had a
1000 yd rushher
1000 yd receiver
3000 yd passer

and not made the playoffs.

Tampa Bay reminds me of Atlanta when they just gave the ball to Jamal Lewis every play.

They will be for real as long as he can stay healthy.

However how many backs can carry the ball 500 times? I think he is on pace for something close to that.

36
by Fnor (not verified) :: Wed, 09/28/2005 - 2:07am

Josh: Go to one of the two FO columns they have linked on the front page, in a box you'll see "power rankings" along with the other 3 outsiders content. Couldn't find another way to it.

The more I look at the bengals, the more I don't like them. Hosh-man-zah-day (yay CBS) has looked sub-par, and I'm not sold on the running game. It looks like the Palmer and Johnson "hey look we have great field position" show right now. A show that will probably end when they hit a team with a QB that doesn't just make random mistakes and running backs that don't cough up the ball.

I'm also not sold on Aaron's contention that their schedule is particularly easy. Perhaps overall it is easy, but there are still games against teams that the Bengals' magic formula, I think, won't work.

Wins: HOU, TEN, GB, BAL, CLE, DET
Losses: PIT, JAC, IND, BAL, PIT, KC
Bizzaro Land: BUF

either 10-6 or 9-7.
Why these? They're teams that don't turn the ball over a whole lot. The high is with Baltimore, and I put them down as split because I think their defense can overcome one Wright game, but not two.

The other teams simply don't cough the ball up much, except for Buffalo, which depends on whether or not Interceptaverde shows up or not. Aside from Wright, they all have efficient QBs that don't make too many mistakes. None of them fumble much (unless they're at their opponents' 5 after a 30 yard pass). PIT, BAL and BUF have strong defenses. KC and IND have good pass rushing, which is a problem when a QB only has 1 or maybe 2 every-play outlets.

So while the chumps assembled to play Cincinnati the rest of this year might not look great as compared to the league average, they all seem, at least to me, well suited to play cincinnati in particular, which is really the important bit. Counting Buffalo as a push, that's .500. Definitely good, but not top-5 good or superbowl good, given the schedule.

(I am so sorry this went on that long)

37
by DavidH (not verified) :: Wed, 09/28/2005 - 2:21am

#35:

Didn't do exactly that, but a quick google search found the article linked in my name. According to that article, 10 teams in 2002 had a 3000 yd passer, a 1000 yd receiver, and a 1000 yd rusher. Only FOUR of them made the playoffs.

So more than half of the good trios missed the playoffs. Apparently it is not very important or indicative of quality.

So my guess is that the answer to your actual question is "last year," but I haven't actually checked that out yet.

38
by DavidH (not verified) :: Wed, 09/28/2005 - 2:31am

Last year there were 6 teams with at least one of each:
3000 yd passer
1000 yd rusher
1000 yd receiver

4 made the playoffs.

Those that did:
Seattle (Hasselback, Alexander, Jackson)
Indianapolis (Manning, James, Wayne/Harrison/Stokely)
Denver (Plummer, Droughns, Smith/Lelie)
Green Bay (Favre, Green, Walker/Driver)

Those that did not:
Houston (Carr, Davis, Johnson)
New Orleans (Brooks, McAllister, Horn)

39
by Fnor (not verified) :: Wed, 09/28/2005 - 2:52am

In tandem with David's point:
you're making a bad causation argument about yardage and a good team. A good team (usually) gets good yardage on offense, because they are good, not because of the yardage itself. Not all yardage is created equal. Take Johnson's respectable 300 (exactly! Wow.) yard season in progress:

Qtr Rush. Yrd. Avg. Long
1 16 44 2.8 8
2 18 71 3.9 13
3 19 93 4.9 11
4 20 92 4.6 25

Tired defenses playing from behind ahoy! Johnson has 1 TD thus far in the year, in the second quarter vs. cleveland. In fact, the Bengals have scored 5/9 TDs and 5/6 FGs in the first half, the not-rudi-johnson part.

Against a good defense (CHI), on the other hand, most of the scores came late, because Chicago's defense practically died of exhaustion from being on the field so much. Even with all that, they only put together 2 drives greater than 50 yards, both in the second half, and one containing a 40 yard pass to Johnson.

So what does all that mean? It shows that the convex of the statement "Rudi Johnson is good, so he gains a lot of yards" isn't true, because he's gaining yards mostly due to the rest of the defense, much like the the recievers against CHI last week put up big numbers mostly due to the defense's ability to get Orton and co. off the field. In fact, Rudi Johnson, aside from eating clock, hasn't been much of a factor in why CIN has won at all.

This was all just a specific case, but it's good as a general rule to remember that while a good team/player will usually create yardage, good yardage will not create a good team/player.

(How the #%^@# do these keep getting so long?!)

40
by bobman (not verified) :: Wed, 09/28/2005 - 2:58am

Re: #35/37
The Colts in 2001 are an example of a team with a 1,000 yard rusher and receiver and 4,000 yard passer and they ended up not only missing the playoffs, but with a 6-10 record. Their only year out of the postseason since Manning's soph campaign.
The huge difference is quality of D, which Cincy seems to have so far, and in Indy that year, was probably their worst in the past 7-8 seasons (which is saying something). It was that miserable D and kick coverage performance that led to the amazing stats, and it was Manning's big passing days that opened up running lanes for an undrafted rookie (Rhodes) to run for 1,104 yards in 11.5 games. Hell, they had almost 1,800 yards and 12 TDs out of the starting tailback position adding in pre-injury Edge, plus Manning's usual 4,000 yards and Harrison's 1,500+ receiving yards, and still lost 10 games. Now if you do a search for THOSE stats, 4,000/1,500/1,500 and find a losing team, you can rest assured the D is in the bottom third of the league.
So not only do I agree that the "statistical triplet" indicator is not helpful in predicting wins or postseason appearances, but if you ignore the other side of the ball, you're just wasting your time.
To reiterate, though, Cincy appears to have pretty good coaching, scheme, and personnel on D. The fountain of TOs can't keep gushing, though, but I'd pencil them in for 10-11 wins. It would certainly be surreal if their Nov. 20th tilt with the Colts featured the season's last two undefeated teams!

41
by Fnor (not verified) :: Wed, 09/28/2005 - 3:17am

"Rudi Johnson is gaining yards due to the rest of the defense" should read "rest of the offense.

Sorry about that.

42
by Tom W (not verified) :: Wed, 09/28/2005 - 3:45am

G.B.: Ranked 24th in offense, 24th in defense, 26th in special teams. Good to see that my boys' offense has finally caught up to their D and ST's in terms of crappiness. Balance is important. Plus they should have already won a 1/2 a game!! I don't see what everybody here in cheeseland is complaining about.

43
by Aaron (not verified) :: Wed, 09/28/2005 - 8:41am

Hey folks. Ratings are now up on FOX. Click link on my name.

44
by charles (not verified) :: Wed, 09/28/2005 - 9:05am

Oh, Detroit is upsetting TB this week ratings be damned. The Kevin Jones experience begins now.

45
by Moe (not verified) :: Wed, 09/28/2005 - 10:48am

- I see that MN went into week three ranked last in the NFL. I still think they could end up being a good team. Perhaps they can set some sort of DVOA record in terms of the greatest positive increase over a season.

- If TB and CIN stay good that should eventually help MN's ranking yes?

- Seeing NE ranked so low (for them) should definitely set of the no respect alarms, but with Harrison out for the year it is hard to see who answer the call.

46
by Pat (not verified) :: Wed, 09/28/2005 - 2:29pm

OK, so I've just been playing around with a bit of an idea, so I'll throw it out there. Once I get it finished I might email it to Aaron just to make sure he sees it, but if he sees it here, cool. :)

Warning that this is really long, but the idea is pretty easy to understand.

One thing I realized is that the "strength of schedule" computation is a little simplistic. Right now it's just the average of all your opponent's DVOAs. The problem with this is that this allows one "really really bad" opponent to make a difficult schedule look easier than it is, and the reverse as well.

So I thought about a different approach. I used the ESTIMATED WINS, divided by the number of games played, to get a "winning percentage". I then used Bill James's log5 percentage to calculate the chance that a 0.500 team would beat them. (Note that since Cincinnati and Tampa has 1.000 "estimated winning percentage", they're automatic losses. Reverse for Baltimore & Houston). I then summed all of those chances, and subtracted it from them half the number of games. So if it were the full season, I'd subtract it from 8, because that's what you'd get if all the teams they faced had a 0.500 average. That's "wins above average for a league-normal schedule". Convert that to a percentage of 8, and that's your "strength of schedule."

I've done this for a couple of teams so far, and the results do look different than the PAST and FUTURE schedule results above. For one thing, they're a lot more spread out. The "average of all DVOA of opponents" constrains the schedules to be near zero, since the DVOAs are (nearly) zero sum.

Let me give an example of the strengths of this: Suppose that a team plays CIN, ARI, and MIN. An average of those 3 DVOAs is about zero - so it looks like an average schedule. But in truth, ARI and MIN are gimmes, and CIN is a loss. That's not average - that's easier than average - 2 wins in 3 games.

So here they are for the few teams I calculated it for (negative is weaker schedule, positive is stronger):

PHI: PAST SCHED: -20.0% FUTURE SCHED: 1.5%
CHI: PAST SCHED: 6.6% FUTURE SCHED: -22.6%
CIN: PAST SCHED: -17.7% FUTURE SCHED: -16.9%
TB: PAST SCHED: -57.8% FUTURE SCHED: -6.6%
PIT: PAST SCHED: -40.0% FUTURE SCHED: -1.0%
NYG: PAST SCHED: -28.8% FUTURE SCHED: -4.6%
TEN: PAST SCHED: -8.8% FUTURE SCHED: 3.07%

So of the top 4 teams, Tampa actually had the easiest early schedule, not Pittsburgh. Why? Because Minnesota and Green Bay were virtually guaranteed wins, and Buffalo was a pretty good bet too. So an average team would've won 2.4 games in those first three games. Looking at Pittsburgh, though, while Tennessee and Houston were very easy, New England was less so. An average team would've won 2.1 games there. Here, you're not biased by the fact that Houston is awful, because Houston could've been 10 times worse, and it doesn't change the schedule weighting. It's still only one game.

Also to be noted: Cincinnati's early schedule was about as easy as Philadelphia's. For an average team, Cleveland and Chicago were roughly 50/50 shots, and Minnesota was a lock. Pretty much the same for Philly - Atlanta and Oakland were 50/50 shots, and SF was a lock. Cincinnati's future schedule is as easy as their early schedule.

Anyway, what do people think? I like this because it prevents a really crappy team from making a team's schedule look easier than it is. It also has a lot of potential for improvement, too.

47
by Larry (not verified) :: Wed, 09/28/2005 - 3:32pm

Pat,

I like it. This is much better than average opponent rating. You could augment it by adding the wins you'd expect the team to get, given its schedule (use their expected wins, instead of .500). FO obviously has ways of calculating a win percentage based on two teams' rating, since that is the basis of the predicted win range in the book and the preseason playoff liklihood predicitons. So, they might want to use that instead of the log5 method, to be consistent, but this concept is clearly the direction to go.

48
by Jimmy Two Times (not verified) :: Wed, 09/28/2005 - 3:36pm

Great idea, Pat. I agree that the schedule strength ratings look fishy as is.

49
by Fnor (not verified) :: Wed, 09/28/2005 - 3:50pm

I like that idea, Pat. Good point about Houston and the hypothetical houston x 10. I still like the idea of comparing individual matchups by strengths, though. For the same reasons that ur-Houston isn't going to help the team win more, a team that has a high VOA mostly due to a strong run offense and pass defense aren't going to have nearly as easy a time against a team with strong run defense and run offense than they would against a team with a weak run defense and run offense, even if the two theoretical opponents had the exact same overal DVOA. If they had identical offense and defense DVOA, even.

I suppose that's what I was getting at with Cincinnati's strength of schedule... just looking at the average DVOA we can go "yay! Happy Cincinnati party time! Fear the bengal!" but that sort of ignores that, while the remaining schedule would look quite easy for, say, NE or PHI or ATL or STL, the teams on the docket are well-suited to playing against the bengals.

50
by Pat (not verified) :: Wed, 09/28/2005 - 7:20pm

You could augment it by adding the wins you’d expect the team to get, given its schedule (use their expected wins, instead of .500).

I've also done "given a team's expected win rating, this is the number of wins they're expected to get with this schedule. This is the number of wins they're expected to get with a league-average schedule (all 0.500 teams). This is the difference" although that doesn't quite intuitively give you "schedule strength". It tells you how many wins you can expect the team to exceed your projection, based on the schedule. I call it "WXNS", for wins in excess of normal schedule. Wax-nass.

Top WXNS of the ones I've looked at so far? Tennessee. Why? Because they play Baltimore and Houston (x2), and they aren't very good, so any "free wins" are huge to them. The fact that they play Indianapolis twice and Pittsburgh doesn't counter that - because well, they'd have a good shot at losing to a normal team, so it doesn't hurt much more to lose to a good team. Tennessee, playing a league normal schedule the way they're playing now, should end up with about 3-4 wins. With the schedule they have, they'll have 5-6 wins. Should make Titans fans happy.

51
by Aaron (not verified) :: Wed, 09/28/2005 - 11:14pm

I'd be happy to look at ideas, but I can explain to you why some of these ratings look fishy in two words: THREE GAMES

52
by CaffeineMan (not verified) :: Wed, 09/28/2005 - 11:33pm

So, Pat, is that pronounced "Waxin' a$$?"
:)

53
by Pat (not verified) :: Thu, 09/29/2005 - 12:48am

Aaron:

Oh, definitely. Heck, the WXNS for Cincinnati, Tampa, Houston, and Baltimore is zero, both past and present. Simply because both of them have either perfect (CIN/TB) or zero (HOU/BAL) records.

And, of course, log5 breaks when two zero-winpct or perfect-winpct teams play each other. :)

But something like WXNS, and the alternate "schedule strength" won't fix any of those problems, either. But they do offer additional information - like saying "if the season goes on, Tennessee will probably overperform its ESTIM. WINS by 2 or even 3 wins" - even though Tennessee has a roughly average schedule, it's far better for them to have BAL/HOUx2 on their schedule than a whole bunch of 0% DVOA teams which would kick their butt. WXNS tells you "this is how much of this team's overperformance was probably due to their schedule."

It's like college football. You could swap, say, USC for Ohio State in Penn State's schedule, and it's not going to hurt their overall record, even though it just got "harder" (they'll still lose). But if you even swapped, say, Notre Dame for Cincinnati, poof, one win goes completely away.

Incidentally: woo hoo! look at the estimated wins column. Note Detroit, with a lower DVOA, but higher ESTIM. WINS than Arizona? That's because Detroit has a VARIANCE of 132.7%, and Arizona has a VARIANCE of 6.1%, and a high variance is good for bad teams, and bad for good teams (Fluctuating up is ridiculously helpful, fluctuating down doesn't hurt you, as you already sucked to begin with). That was my suggestion - almost two years ago, I think. Woo hoo!

54
by pchase (not verified) :: Thu, 09/29/2005 - 4:32pm

could we get the fox percentages?

55
by Will Allen (not verified) :: Thu, 09/29/2005 - 5:19pm

I have a question about the sensitivity of the early season rankings. Suppose Jermaine wiggins of the Vikings does not draw a phantom intereference penalty against the Bucs, thus the Vikings get about 40 yards more passing, along with one more completion and a touchdown. Suppose the Bucs don't have a 70 yard td run by Cadillac Williams in the closing seconds, when the Vikings stack the line knwoing they have to force a fumble. The Vikings win the game 20-17. How would the rankings be affected?

56
by B (not verified) :: Thu, 09/29/2005 - 5:23pm

Will, since there are diminishing returns on long plays (IE a 70 yard run for a TD isn't worth that much more than a 30 yard run for a TD), the answer is not that much. Min would be a little higher and TB would be lower, but they wouldn't be switching places. The effect in standings would be a lot more dramatic. Minny at 2-1 would be tops in the NFC North and there'd be no mention of how Culpepper misses Moss.

57
by Will Allen (not verified) :: Thu, 09/29/2005 - 5:39pm

Well, I didn't think they would switch, but I am curious how much movement there would be. Actually, that 70 yard run was about as statistically meaningful as an interception thrown by a team down by ten, on a hail mary pass, on the last play of the game. I wonder if FOs' methods reflect that.

58
by Pat (not verified) :: Thu, 09/29/2005 - 5:49pm

Minny at 2-1 would be tops in the NFC North and there’d be no mention of how Culpepper misses Moss.

I think the 8 interceptions he's thrown so far might still give some people fodder. :)

59
by Jimmy Two Times (not verified) :: Fri, 09/30/2005 - 4:56am

Aaron--

I get that this is a small sample and things are going to look a bit more random than they should because of it. However, there's still some money to be made in wicking as much of that randomness away as possible. Regardless of whether Pat's idea is the best way to cut down on the randomness, it seems to me that he's conceptually on the right track, in that he's merely trying to limit how much one specific event can effect a ranking. The logic is built exactly on an understanding of the small sample sizes in which we're dealing, sample sizes small enough that each individual event can have a massively out-of-proportion impact on what we're trying to calculate unless we manage it.

Just my 2c.

60
by Dan (not verified) :: Sat, 10/01/2005 - 6:23pm

Pat, ESTIMATED WINS is based on performance in previous games. When you're calculating future schedule strength or WXNS, you're interested in how opponents will perform in future games. Estimated Winning Percentage (EWP = Estimated Wins / # of games played), on its own, is not a great predictor of future performance, at least early in the season. This is largely because Estimated Wins is dependent on luck (teams outplaying or underplaying their ability in a small sample size), so there will be regression to the mean over time.

What I think you want to do is to use a weighted average of the team's Estimated Winning Percentage and the league average EWP (which is .500), with the weight depending on the number of games that have been played. We can call this Future Estimated Winning Percentage (FEWP). With a weighting factor of k, FEWP = k x EWP + (1-k) x .5. Before playing any games, every team has a FEWP of .50. After one game, you might give a weight of .1 to their EWP from that game, so a team with a 1.0 EWP would have a FEWP of .55 (.1 x 1.0 + .9 x .5), and a team with a 0.0 EWP would have a FEWP of .45. After 3 games the weight might be up to .28, so an EWP of 1.0 would give you a FEWP of .64. After 15 games, the weight on EWP might be up to .96, so an EWP of 1.0 would give you a FEWP of .98. You could set the weighting factor for week w by comparing teams' EWP after week w with their EWP at the end of the season, and seeing what weight of their week w EWP gives you an unbiased estimate of their end of the season EWP (or, perhaps, of their EWP for the remainder of the season).

Then you'd just have to redo your future strength of schedule and WXNS calculations with FEWP instead of EWP.

You might also do the same thing with past schedules, to account for the impact of luck. So maybe Future Estimated Winning Percentage is not the best name. Perhaps: Estimated True Winning Percentage? Or: Expected Estimated Winning Percentage? Or: Adjusted Estimated Winning Percentage? Or ...?

(In case it isn't obvious, I like your idea, which is why I'm trying to improve on it.)

61
by Pat (not verified) :: Mon, 10/03/2005 - 2:27pm

Dan:

That's a great idea, although I think I'd do what FO does and use the "0.1 weighting per game until Week 10". Ideally, it'd be best if there was a "weighted estimated wins" after that.

Oh, and I'm switching "WXNS" to "SWXN" because the first "wins exceeding normal schedule" makes it sounds like the team did something to win - like, they're playing better than they would against a normal schedule. SWXN stands for schedule wins exceeding normal, which sounds more like what it is.

I'll probably have SWXN and adjusted sched. strength for all teams before this weekend, and maybe I'll post it in the rankings thread, and email the spreadsheet to Aaron if he wants it.

Tennessee's SWXN is still highest of all teams that I've calculated so far. Sucks to be a Titans fan, I guess, because they'll get a lower draft pick than they really deserve.

62
by Freddie Falcon (not verified) :: Mon, 10/03/2005 - 4:21pm

Falcons at 17??
This is just wrong.
The system just does not work.
Stop embarrasing yourselves.