Writers of Pro Football Prospectus 2008

29 Dec 2003

Final 2003 Team Efficiency Ratings

by Aaron Schatz

Here are the team efficiency ratings after Week 17, measured by our proprietary Value Over Average (VOA) 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.)  These are unofficial numbers, where a few plays may differ from the official NFL play-by-play.

This was going to be where I wrote a paragraph about how the New England Patriots should not be the prohibitive Super Bowl favorite, since the St. Louis Rams had been playing just as well and nearly as consistently.  Don't need to write that paragraph now, do I?

Last year, the top two teams in DVOA actually met in the Super Bowl; this year, I think the results of the playoffs will be a little different.  For one thing, the top four teams in full season DVOA are all from the same conference, the AFC, thanks to St. Louis' final week plunge.  In the AFC, the top six teams in DVOA all make the playoffs.  In the NFC, DVOA says that Tampa (#8), San Francisco (#10), and Minnesota (#11) were all better than Carolina (#18), Dallas (#14), and Philadelphia (#13).  This shows you the value of consistency, not to mention a schedule that includes two games against the Giants.

Kansas City and Seattle would meet in the Super Bowl if the top teams in DVOA were to make it all the way, but both of those teams peaked earlier in the season and have been fading a bit of late.  In addition, the road to Houston goes through Foxboro for any of the three teams which stand above New England in our ratings, and two of those teams have to play three games instead of two to make it to the Super Bowl.

I was very excited that the Patriots finally had a huge win, because they would finally climb to the top of the WEIGHTED DVOA ratings, the numbers that count recent performance stronger than early season performance.  However, the Pats only make it to #2 in WEIGHTED DVOA; the top team is Tennessee, which also struggled in the first couple weeks of the season and then demolished Tampa Bay on the last day of the season.  The Patriots also climb from #8 to #4 in total DVOA thanks to the win over Buffalo, but they don't get any higher because over the past two weeks their previous opponents have taken a beating, which has affected the opponent adjustments, which has affected the Pats' rating.  Part of this can be solved in the offseason by creating a "rolling opponent adjustment" that judges each game by how the opponent is playing at that time, rather than over the entire season -- Washington was playing better when they faced New England than they were by the end of the season, for example, as were the New York Giants, while Jacksonville was playing better when they faced New England than they were at the beginning of the season.

I've added a couple of columns to these (unofficial) final ratings.  The first is 2004 SCHEDULE.  This represents the average 2003 DVOA rating of next year's opponents.  Obviously teams will change between now and next September, but this is a good early indicator of which teams will face easier or harder schedules next year.  The numbers show bad news for the Miami Dolphins, as well as yet another reason why Tampa Bay should be back in the 2004 playoffs, and some possible hope for those looking to see the return of quality football in Detroit.

The other new column represents the projected wins according to football's version of the Pythagorean theorem that predicts wins and losses based on points scored and allowed.  Back at midseason, I wrote at length about how Carolina and Tampa Bay were both on a historic pace to either win more games (CAR) or less games (TAM) than their points scored and allowed would indicate.   Tampa did lead the league in "underperformance," but ended up tied with San Francisco, winning 2.2 fewer games than the Pythagorean projection, and neither is among the 20 worst underperforming teams of the past 20 years.  The Bucs apparently decided that if they could play well and lose, they might as well play like horse pucky and lose, and so they did.

On the flipside, Carolina ended up losing some close games at season's end, so while they still won a record-tying seven games by three points or less (Hello, 1998 Arizona Cardinals!) they did not end up as the "luckiest" team in the NFL this year.  No, the team that leads the league in outperforming its Pythagorean projection is the New England Patriots, who won 14 games when projected to win only 11.4.

There is not a very good history for teams which win a high number of games and are significantly "lucky" in winning more than their point differential would indicate.  The 1999 Titans and Jaguars each won 13 games instead of a projected 10, and neither won the Super Bowl.  Instead, the Super Bowl went to the team that actually led the league in Pythagorean wins, St. Louis.  The 1991 Lions won 12 games instead of the projected 9, and the Super Bowl went to the team that actually led the league in Pythagorean wins, Washington.  The 1985 Raiders won 12 games instead of the projected 9, and the Super Bowl went to the team that actually led the league in Pythagorean wins, Chicago.  The 1990 Raiders won 14 games instead of the projected 11, and the Super Bowl went to the team that finished second in Pythagorean wins, the New York Giants (12.0), and would have gone to the team that actually led the league in Pythagorean wins, Buffalo (12.2), if not for "wide right."

So, history would seem to indicate that the New England Patriots will not win the Super Bowl, and that the Super Bowl champion will instead be the team that led the league in Pythagorean wins... the New England Patriots.  Yes, that's right, the Patriots not only outperformed their Pythagorean projection by the largest amount in the NFL this year, they also had the highest Pythagorean projection in the NFL this year.  Their projection of 11.38 wins barely beat Kansas City's projection of 11.35 wins.  This marks the first time since the 1982 strike that this has happened. For all I know, this is the first time in NFL history that this has happened -- my spreadsheet of Pythagorean records only goes back to 1983, so I can't be sure until I crunch more numbers.  This literally means that only the Patriots can beat themselves.

All year long, the argument has raged over whether the Patriots are lucky or good.  It turns out that they are both.  They are the luckiest team in the NFL, and the best team in the NFL, at least according to points and wins.

That was a lot of Patriots talk, I know, but I wanted to save any discussion of the eight teams playing in the first week of the playoffs for some long playoff preview articles coming later this week.  And now, the final (unofficial) DVOA ratings:

  • As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.
  • All numbers are adjusted for opponent quality except for NON-ADJ TOTAL VOA.

Here are the ratings through Week 17:
 


TEAM
TOTAL
DVOA
LAST
WEEK
NON-ADJ
TOTAL VOA
W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
SPECIAL
VOA
S.T.
RANK
1 KAN 27.5% 1 33.3% 13-3 27.1% 1 4.0% 21 4.4% 1
2 TEN 23.3% 7 26.2% 12-4 17.1% 5 -6.5% 11 -0.3% 21
3 IND 21.0% 3 19.5% 12-4 21.0% 2 0.2% 17 0.1% 13
4 NWE 20.6% 8 23.0% 14-2 -0.2% 12 -20.0% 3 0.9% 8
5 SEA 20.6% 6 22.1% 10-6 18.7% 3 -1.9% 15 0.0% 17
6 STL 20.6% 2 24.7% 12-4 -0.6% 14 -21.3% 2 -0.1% 19
7 DEN 16.0% 5 16.9% 10-6 7.6% 10 -8.3% 9 0.1% 14
8 TAM 15.8% 4 15.8% 7-9 -1.4% 16 -19.2% 4 -2.0% 26
9 GNB 14.9% 11 16.3% 10-6 5.6% 11 -8.8% 8 0.4% 12
10 SFO 14.7% 10 15.8% 7-9 13.0% 6 -4.2% 13 -2.6% 29
11 MIN 13.4% 9 18.8% 9-7 18.2% 4 1.8% 19 -3.0% 30
12 BAL 12.5% 13 13.4% 10-6 -23.0% 32 -32.7% 1 2.9% 2
13 PHI 8.2% 14 10.5% 12-4 12.6% 7 6.5% 23 2.1% 4
14 DAL 7.5% 12 15.4% 10-6 -9.3% 21 -17.2% 5 -0.5% 23
15 MIA 0.7% 17 0.9% 10-6 -14.0% 27 -15.1% 6 -0.4% 22
16 JAC -0.1% 16 -0.3% 5-11 -0.6% 15 -4.9% 12 -4.4% 32
TEAM TOTAL
DVOA
LAST
WEEK
NON-ADJ
TOTAL VOA
W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
SPECIAL
VOA
S.T.
RANK
17 PIT -0.4% 15 0.8% 6-10 -8.7% 19 -6.9% 10 1.3% 7
18 CAR -5.8% 20 -2.3% 11-5 -9.1% 20 -2.5% 14 0.8% 9
19 BUF -7.0% 18 -12.3% 6-10 -16.6% 29 -9.8% 7 -0.2% 20
20 NYJ -8.8% 21 -9.8% 6-10 10.1% 8 18.9% 32 0.0% 16
21 NOR -9.3% 23 -7.9% 8-8 -0.4% 13 9.6% 26 0.7% 11
22 CIN -10.1% 19 -6.3% 8-8 8.4% 9 16.4% 30 -2.1% 27
23 CLE -10.1% 24 -18.3% 5-11 -10.8% 24 -0.8% 16 0.0% 18
24 OAK -13.1% 22 -17.3% 4-12 -9.7% 22 6.0% 22 2.7% 3
25 SDG -14.8% 25 -19.7% 4-12 -2.1% 17 10.6% 27 -2.1% 28
26 CHI -17.5% 28 -20.3% 7-9 -16.0% 28 3.5% 20 2.0% 5
27 WAS -18.1% 26 -16.1% 5-11 -5.6% 18 11.9% 28 -0.7% 24
28 ATL -18.7% 29 -22.8% 5-11 -10.8% 23 7.9% 25 0.1% 15
29 DET -19.2% 30 -23.6% 5-11 -19.3% 30 0.6% 18 0.8% 10
30 NYG -21.1% 27 -22.0% 4-12 -13.1% 26 6.8% 24 -1.2% 25
31 HOU -25.1% 31 -32.4% 5-11 -12.6% 25 14.4% 29 1.9% 6
32 ARI -41.1% 32 -46.6% 4-12 -19.8% 31 18.0% 31 -3.4% 31

  • ESTIMATED WINS uses a statistic known as "Forest Index" that takes into account offense, defense, and special teams, as well as consistency, red zone performance, and performance in the second half when the score is close, and then projects a number of wins adjusted to a league-average schedule.  The Forest Index is described here.
  • PYTHAGOREAN WINS represent the number of wins projected from the team's points scored and allowed, as described in this article.
  • WEIGHTED DVOA combines the team's DVOA performance from each game.  The past four weeks are each weighted at 100%, while each week before that is weighted progressively lower, beginning with Week 1 at 9%.
  • 2003 SCHEDULE lists average DVOA of past opponents, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative).  This number will differ from the difference between DVOA and (non-adjusted) VOA because schedule strength is based on the opponent's total efficiency rating, while opponent adjustments to VOA take into account situations faced within each specific game.
  • 2004 SCHEDULE lists average DVOA of next year's opponents, based on their 2003 performance.
  • 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
ESTIM.
WINS
PYTH.
WINS
WEIGHTED
DVOA
RANK 2003
SCHED.
RANK 2004
SCHED.
RANK VARIANCE RANK
1 KAN 27.5% 13-3 12.1 11.4 23.6% 3 -4.5% 30 0.7% 14 20.1% 21
2 TEN 23.3% 12-4 11.0 10.7 31.9% 1 -2.2% 23 -0.7% 21 23.6% 17
3 IND 21.0% 12-4 11.4 10.6 15.9% 8 -0.7% 20 2.3% 8 19.9% 23
4 NWE 20.6% 14-2 12.3 11.4 29.3% 2 -1.8% 22 1.6% 12 17.6% 27
5 SEA 20.6% 10-6 11.5 10.0 12.0% 11 -3.5% 28 -0.2% 17 18.4% 26
6 STL 20.6% 12-4 11.0 10.8 21.1% 4 -4.3% 29 -0.1% 16 31.8% 7
7 DEN 16.0% 10-6 10.1 10.2 16.9% 7 1.5% 10 -0.6% 20 24.3% 15
8 TAM 15.8% 7-9 10.9 9.2 6.6% 12 -2.6% 24 -4.5% 31 29.4% 9
9 GNB 14.9% 10-6 9.9 11.3 19.8% 5 0.5% 15 -2.3% 23 26.7% 12
10 SFO 14.7% 7-9 7.3 9.2 12.2% 10 0.5% 16 -3.0% 27 35.7% 3
11 MIN 13.4% 9-7 8.2 9.5 5.7% 13 -3.3% 27 -2.3% 24 32.3% 5
12 BAL 12.5% 10-6 10.8 11.0 19.4% 6 -0.6% 19 -0.7% 22 22.2% 19
13 PHI 8.2% 12-4 11.2 10.4 15.8% 9 -2.9% 26 -4.1% 29 20.0% 22
14 DAL 7.5% 10-6 7.8 9.0 5.1% 14 -8.4% 32 -4.2% 30 34.8% 4
15 MIA 0.7% 10-6 8.8 9.6 3.7% 16 0.2% 17 3.5% 3 26.2% 13
16 JAC -0.1% 5-11 6.4 6.3 4.0% 15 1.5% 11 2.4% 7 16.4% 29
TEAM TOTAL
DVOA
W-L ESTIM.
WINS
PYTH.
WINS
WEIGHTED
DVOA
RANK 2003
SCHED.
RANK 2004
SCHED.
RANK VARIANCE RANK
17 PIT -0.4% 6-10 8.9 7.2 -1.3% 17 1.8% 9 -2.9% 26 24.0% 16
18 CAR -5.8% 11-5 6.9 8.6 -2.7% 18 -5.6% 31 1.8% 10 12.4% 32
19 BUF -7.0% 6-10 6.2 6.7 -13.8% 23 2.4% 7 1.2% 13 39.1% 2
20 NYJ -8.8% 6-10 7.7 7.5 -10.2% 22 0.7% 14 -0.3% 18 12.4% 31
21 NOR -9.3% 8-8 8.8 8.4 -4.7% 19 -2.8% 25 2.1% 9 20.2% 20
22 CIN -10.1% 8-8 6.1 7.0 -7.0% 20 0.1% 18 1.6% 11 15.3% 30
23 CLE -10.1% 5-11 6.5 5.8 -18.5% 27 4.7% 3 -3.4% 28 32.0% 6
24 OAK -13.1% 4-12 6.6 4.9 -19.9% 28 3.5% 4 3.2% 4 23.4% 18
25 SDG -14.8% 4-12 5.6 4.9 -14.6% 24 2.8% 6 2.7% 6 18.4% 25
26 CHI -17.5% 7-9 7.0 6.1 -9.4% 21 1.3% 12 2.8% 5 16.7% 28
27 WAS -18.1% 5-11 4.2 5.6 -23.6% 29 -1.3% 21 0.2% 15 28.5% 10
28 ATL -18.7% 5-11 5.3 4.9 -16.6% 26 1.9% 8 -0.5% 19 28.0% 11
29 DET -19.2% 5-11 7.2 4.9 -15.4% 25 3.4% 5 -2.7% 25 25.3% 14
30 NYG -21.1% 4-12 4.2 4.0 -29.5% 30 1.0% 13 -5.1% 32 29.7% 8
31 HOU -25.1% 5-11 5.8 4.5 -32.3% 31 5.8% 2 4.8% 1 18.8% 24
32 ARI -41.1% 4-12 2.7 2.6 -39.7% 32 6.1% 1 3.7% 2 43.3% 1

PREVIOUS WEEKS:

Posted by: Aaron Schatz on 29 Dec 2003

comments