Writers of Pro Football Prospectus 2008

Most Recent FO Features

OsemeleKel12.jpg

» Word of Muth: Osemele's Breakout Season

Ben Muth says that Baltimore's third-year guard Kelechi Osemele might be the NFL's next great interior lineman.

05 Oct 2004

Week 4 VOA Ratings

by Aaron Schatz

Here are the Football Outsiders team efficiency ratings through Week 4 of 2004, 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.)

Moving Up: SD (23 to 14), STL (24 to 15), HOU (30 to 21), ARI (27 to 19)

Moving Down: TEN (20 to 30), CAR (16 to 25), BAL (7 to 12), OAK (15 to 20), BUF (19 to 24), CHI (21 to 26)

I said before the season that special teams ratings are inconsistent from year-to-year, but they are also extremely volatile from week-to-week -- especially early in the season.  The Patriots, who are in the top five in offense and defense, drop out of the top spot in the rankings solely because the kickoff returned for a touchdown by Buffalo's Terrence McGee.  New England special teams plunge to the bottom of the NFL, but they will rebound.  A bigger worry for Patriots fans should be punt returns -- the Pats are also at the bottom of the league there, and that isn't just one play.  And the Patriots, as a great team, do return a lot of punts.

Another comment about the Buffalo-New England game: It is in the best interest of both the Buffalo Bills and Drew Bledsoe to get him the heck out of there as soon as J.P. Losman is healthy.  Early in the game, the Bills put on max protect, and Bledsoe completed a long bomb to rookie Lee Evans.  It demonstrated once again that Bledsoe can throw the ball with accuracy and length when you let him stand there long enough.  But for the most part they just can't keep him upright, and he's way too expensive to keep around when he is playing like this.  Yesterday, Steve DeOssie actually made a very good suggestion for Bledsoe on WEEI radio here in Boston.  Next year, he's going to have to go somewhere as a backup for a lot less money, and the perfect situation would be Oakland.  They won't be bringing Gannon back, they have receivers who love to go long, and that offensive line would give Bledsoe an amazingly stable pocket if he had to come in due to a Kerry Collins injury.

* * * * *

Four weeks into the season it is time to ask: Why were my NFC predictions so darn wrong?

In the AFC, I actually did a very good job on projecting things.  The Jets are living up to my expectations, and the Jaguars and Chiefs actually have VOA ratings in line with our projections even if their records don't match.  Miami is way below the statistical projection -- they have an offensive VOA of -62.3%, which is nearly three times worse than any other below-average team -- but given the turmoil it isn't a shock.  The one team that I don't understand is the Titans, who look horrible.  Their rating will drop even further next week when we start adding opponent-strength adjustments into the formula, because their one win came against the Dolphins.  Their defense is in tatters, and I'm not sure it can all be blamed on departed players, unsure rookies, and preseason injuries.  I'd like to email Jim Schwartz to ask him what's going on but I have a feeling his time is better spent doing something other them exchanging emails with me.  At least the defense has an excuse, I'm not sure what we can blame the poor offense and special teams on.  At least their terrible game Sunday makes DVOA's strangest statistical prediction actually reasonable: right now the San Diego Chargers have a top ten offense.

The correlation between actual VOA and projected DVOA for the AFC teams so far is .49, which is pretty good.  The correlation between actual VOA and projected DVOA for the NFC teams so far is .01, though, which means that my NFC projections might as well have been pulled out of a hat.  Even worse, the one team that I picked for purely non-statistical reasons, Washington, happens to be one of the few teams playing in line with statistical projections.  The Packers have been destroyed by injuries, meaning for the second straight year you can toss my Super Bowl pick out the window before we even hit the second week of October.  The Giants, Falcons, and Cardinals -- Cardinals? -- all discovered defense this year.  I think everyone now understands why we've said that defense is less predictable than offense.

* * * * *

As long as we're talking predictions, what about predicting the rest of the season?  With the league's best 0-3 team winning and the league's worst 3-0 team losing, my mind wandered to the future this week.  I can't remember where I heard this, on one of the pregame shows this week or in somebody's column, but there is a theory that teams are so afraid to play their starters in the preseason for a significant amount of time that the first few games have almost become the "new preseason."  Coaches are still getting used to their personnel, and with all the moves due to free agency, teammates are still getting used to each other.  This explains why, when you compare the last few years to most of NFL history, more teams have started poorly and turned around to make the playoffs, or started strong and then faded over the course of the season.

I went back to figure the VOA ratings for each team over the past four seasons to see if there were any indicators of which teams in 2004 will turn it around or fall apart.  I figured the VOA ratings the same way this week's ratings are figured -- after four weeks, without any opponent-strength adjustments.  Teams had played three or four games, depending on bye weeks.  There are two questions here.  First, what happened when a team's record after four weeks was out of line with its VOA rating?  Second, was there any way to tell which teams with terrible VOA ratings would suddenly improve, and which teams with great VOA ratings would nosedive?

The answer to the first question gives the Kansas City Chiefs reason to feel optimistic.  Over the past four years, eight different teams were in the VOA top ten after Week 4 despite records of 2-2 or worse.  (The 2003 Bucs are not one of these teams, by the way; they were 2-1 after Week 4.)  Six of these eight teams made the playoffs -- but five of the eventual playoff teams were a game ahead of Kansas City at 2-2.  They include the 2000 Eagles and Broncos, 2001 Eagles and Jets, and 2002 Browns.  The sixth eventual playoff team was the 2002 Atlanta Falcons, who were only 1-2 but fifth in VOA after Week 4.  The two teams that missed the playoffs were the 2000 Redskins, who started 2-2 and finished 8-8, and the 2003 49ers, who started 1-3 and finished 7-9.

The other side of this would be teams with winning records and poor VOA ratings after Week 4, but they don't really provide a good guide to this season because there aren't any teams like that this year.  The lowest-rated winning team by VOA is the Vikings, but at #18 they aren't really particularly low.  There are two teams that stand out for winning early despite poor performance.  The first is, you no doubt guessed, last year's Carolina squad.  The second team, well, that's a little strange.

The second team brings us to question two: are there any indications of which teams will turn around not just their wins and losses, but their VOA ratings over the rest of the year?  I haven't gotten to posting the 2000 ratings and commentary on the website yet, but I can tell you that the 2000 Tennessee Titans ranked #22 in VOA after Week 4 despite a 2-1 record.  Had Football Outsiders existed at the time, I might have written: "Despite two three-point wins, Tennessee has not played to the level of a winning team so far."  Boy, did that change.  By the end of the season the Titans were 13-3 and had the DVOA rating to match.

But looking at the teams that started badly and turned things around, I can't really find a connecting factor.  The one thing that is true of many of these teams is that they were blown out in the very first game of the season: the 2003 Patriots, Eagles, and Ravens, the 2002 Steelers, the 2001 Redskins, the 2000 Steelers.  (That might be a good indicator for this season's Panthers.)  Other teams that started with horrible VOA ratings and turned things around include the 2000 Titans, 2002 Jets and Titans, and 2003 Rams.  I'd be curious if anyone can come up with something that most of these teams have in common that might tell us which of the poor teams this season is about to get much better -- feel free to post ideas in the comments.

How about teams that started out with strong VOAs and fell apart?  There really weren't any in 2000.  I wrote about the 2001 Chargers a few weeks ago, and the 2002 Chargers had a similar fast start and collapse.  Also in the VOA top ten after four weeks before ending with negative ratings by season's end: 2001 Saints, 2001 Browns, 2002 Panthers, 2003 Giants, and 2003 Redskins.  Again, I'm not sure what the common thread is here and I'm taking suggestions.

* * * * *

Speaking of predictions, there's been a running line of conversation in the open thread each week about using the VOA ratings to predict point spreads.  Reader Sean McCormick has taken it upon himself to track how well VOA can predict winners each week.  I don't really mind, and I think it is kind of funny that VOA has done such a good job so far (if I remember correctly, it was horrible early last season).  The problem is that this has led a lot of other readers to request more analysis related to gambling lines.  Actually, for many of these emails and comments, "request" is the wrong word.  I'm getting "you should do this" and "you should do that."  I can't tell you, the readers, how enjoyable it has been to build this site and my new career as a sportswriter, and I'm really glad that people are enjoying what we're doing on Football Outsiders.  But as anyone who owns a 1988 Bill James Baseball Abstract could probably guess, I get rubbed the wrong way by comments that tell me what research we should or should not do.  

I've said this many times, but I'll say it again: I have always aimed to keep this site from becoming yet another gambling advice or fantasy football advice site.  That's why, for example, you won't find some unresearched, subjective "start and bench" list on our site each week.  Yes, I enjoy talking about gambling and fantasy football, and I know how important these things are to the popularity of the NFL.  I'm even participating in TwoMinuteWarning's "Head-to-Head" contest in a couple weeks.  But I started Football Outsiders because all I could find on the Web were sites devoted to predicting point spreads or discussing fantasy football, and nothing devoted to discussing how real NFL teams win and lose games, how teams can improve themselves to make the Super Bowl, which nuggets of conventional wisdom are true and which are nonsense -- in short, all the topics that, applied to baseball, made Bill James so much fun.  That's where my interest lies.  A point spread prediction system might be in the cards for next season -- heck, a fantasy football projection system might as well -- but it isn't on the top of my list.  If something like that happens, it would likely be premium content.  And telling me how I "should" do something, over and over again in the discussion threads, is not going to make me do it sooner.  Unless your request accompanies, say, a rather large donation to the Framingham Winter Heating Oil Fund.

In all seriousness, the best thing you can do if you enjoy the site is give a small donation (or, if you have something worth advertising, purchase an advertisement).  I'm making my living at this now, and there's been a lot of progress over the past couple months.  But despite my recent breakthroughs writing for other websites and publications, the family is just scraping by -- and I like to toss some money to the other guys for their hard work as well, especially the programmers.








PAYPAL DONATIONS

Amazon Honor System

Click Here to Pay

Learn More

This week we introduce schedule strength ratings, as as well as VARIANCE, which measures how consistent teams have been from week to week.  Gee, do you notice how the top-rated team in VOA, Seattle, has also had the easiest schedule?  The schedule ratings would seem to suggest that the Patriots and Eagles are actually the two best teams in football, and we should get a better idea of that next week when we add in the opponent-strength adjustments for the first time.  And if Kansas City wants to make the playoffs despite the bad start, they'll probably need a wild card -- notice that Denver is listed with the easiest remaining schedule in the NFL.

Since the season is still young, there are fewer statistics listed than usual.  Early season ratings are based on very little data and are unreliable.  In addition, these ratings will not be adjusted for strength of schedule until after next week.  OFFENSE and DEFENSE VOA are still adjusted to consider all fumbles, kept or lost, as equal value.  SPECIAL DVOA is adjusted for type of stadium (warm, cold, dome, Denver) 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. Player stats pages will be updated later today.


TEAM TOTAL
VOA
LAST
RANK
NON-ADJ
TOTAL VOA
W-L OFFENSE
VOA
OFF.
RANK
DEFENSE
VOA
DEF.
RANK
SPECIAL
DVOA
S.T.
RANK
1 SEA 52.5% 2 66.4% 3-0 4.6% 14 -48.3% 1 -0.4% 15
2 NE 47.4% 1 50.2% 3-0 39.2% 2 -15.5% 5 -7.2% 32
3 PHI 44.5% 3 39.4% 4-0 37.7% 3 -0.2% 16 6.7% 6
4 NYJ 42.4% 4 59.3% 3-0 29.5% 6 -9.8% 8 3.1% 12
5 ATL 28.4% 9 42.1% 4-0 5.5% 12 -16.2% 4 6.7% 5
6 NYG 28.0% 5 40.8% 3-1 21.5% 7 -13.1% 6 -6.7% 30
7 PIT 26.8% 6 23.8% 3-1 -6.0% 21 -29.6% 2 3.2% 11
8 IND 18.3% 12 16.3% 3-1 43.5% 1 24.8% 29 -0.3% 14
9 KC 18.1% 10 8.4% 1-3 19.3% 9 -5.0% 13 -6.2% 29
10 DAL 15.7% 11 1.3% 2-1 20.6% 8 3.1% 17 -1.8% 20
11 DEN 15.4% 8 16.4% 3-1 4.7% 13 -6.9% 10 3.8% 9
12 BAL 11.0% 7 17.2% 2-2 -4.7% 20 -4.5% 14 11.1% 2
13 DET 9.7% 13 14.7% 2-1 -0.7% 17 -6.8% 11 3.5% 10
14 SD 8.6% 23 16.2% 2-2 14.6% 10 10.7% 23 4.6% 7
15 STL -1.8% 24 -7.8% 2-2 30.9% 4 27.9% 30 -4.8% 27
16 JAC -4.1% 17 2.5% 3-1 -10.5% 24 -2.5% 15 3.8% 8
TEAM TOTAL
VOA
LAST
RANK
NON-ADJ
TOTAL VOA
W-L OFFENSE
VOA
OFF.
RANK
DEFENSE
VOA
DEF.
RANK
SPECIAL
DVOA
S.T.
RANK
17 NO -4.7% 14 -0.8% 2-2 1.4% 15 19.9% 28 13.7% 1
18 MIN -6.5% 18 -0.2% 2-1 30.5% 5 29.9% 31 -7.2% 31
19 ARI -9.6% 27 -7.2% 1-3 -14.2% 26 -5.1% 12 -0.4% 16
20 OAK -13.4% 15 -16.4% 2-2 -1.8% 18 8.9% 21 -2.7% 22
21 HOU -14.2% 30 -12.2% 2-2 1.2% 16 13.5% 25 -1.9% 21
22 WAS -14.5% 22 -19.2% 1-3 -23.0% 30 -10.1% 7 -1.7% 19
23 TB -17.0% 26 -18.3% 0-4 -24.2% 31 -8.0% 9 -0.7% 17
24 BUF -17.0% 19 -16.7% 0-3 -18.1% 27 6.5% 19 7.6% 3
25 CAR -17.4% 16 -19.9% 1-2 -2.6% 19 11.2% 24 -3.6% 23
26 CHI -17.7% 21 -14.0% 1-3 -6.2% 22 7.8% 20 -3.7% 24
27 CLE -18.3% 28 -16.9% 2-2 -22.7% 29 3.2% 18 7.5% 4
28 GB -24.8% 25 -23.6% 1-3 7.2% 11 33.9% 32 2.0% 13
29 CIN -28.2% 29 -34.0% 1-3 -13.8% 25 10.4% 22 -4.0% 25
30 TEN -28.4% 20 -29.1% 1-3 -7.5% 23 16.0% 26 -5.0% 28
31 MIA -42.4% 32 -47.5% 0-4 -62.3% 32 -21.4% 3 -1.5% 18
32 SF -43.5% 31 -57.5% 0-4 -20.6% 28 18.3% 27 -4.5% 26

  • PAST SCHEDULE lists average VOA of opponents played so far, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative).
  • FUTURE SCHEDULE lists average VOA of opponents remaining on the schedule, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative).
  • VARIANCE measures the statistical variance of the team's weekly VOA performance.  Teams are ranked from least consistent (#1, highest variance) to most consistent (#32, smallest variance).


TEAM TOTAL
VOA
W-L PAST
SCHED
RANK FUTURE
SCHED
RANK VARIANCE RANK
1 SEA 52.5% 3-0 -21.7% 32 -1.2% 23 22.7% 14
2 NE 47.4% 3-0 -0.8% 19 2.0% 15 15.6% 17
3 PHI 44.5% 4-0 3.3% 13 -1.9% 24 3.1% 31
4 NYJ 42.4% 3-0 -20.7% 31 1.6% 16 4.8% 29
5 ATL 28.4% 4-0 -18.1% 29 4.8% 11 23.1% 13
6 NYG 28.0% 3-1 -3.3% 24 6.3% 8 25.7% 7
7 PIT 26.8% 3-1 -18.2% 30 7.4% 5 38.7% 2
8 IND 18.3% 3-1 -2.5% 22 -3.0% 26 14.5% 18
9 KC 18.1% 1-3 -1.3% 20 3.8% 12 5.7% 27
10 DAL 15.7% 2-1 -13.1% 28 11.9% 1 24.5% 10
11 DEN 15.4% 3-1 1.4% 17 -7.4% 32 4.2% 30
12 BAL 11.0% 2-2 -0.4% 18 8.6% 3 29.9% 5
13 DET 9.7% 2-1 4.2% 11 -3.6% 29 11.3% 19
14 SD 8.6% 2-2 3.8% 12 0.8% 17 52.7% 1
15 STL -1.8% 2-2 -7.4% 27 5.6% 9 24.7% 9
16 JAC -4.1% 3-1 -2.9% 23 -3.1% 27 6.3% 24
TEAM TOTAL
DVOA
W-L PAST
SCHED
RANK FUTURE
SCHED
RANK VARIANCE RANK
17 NO -4.7% 2-2 -2.1% 21 0.0% 19 7.3% 23
18 MIN -6.5% 2-1 14.1% 3 -1.2% 21 19.7% 16
19 ARI -9.6% 1-3 17.3% 2 0.2% 18 29.2% 6
20 OAK -13.4% 2-2 -5.3% 26 6.4% 7 23.7% 11
21 HOU -14.2% 2-2 5.7% 9 -3.2% 28 21.9% 15
22 WAS -14.5% 1-3 2.1% 16 5.0% 10 23.5% 12
23 TB -17.0% 0-4 10.0% 5 -2.8% 25 5.9% 25
24 BUF -17.0% 0-3 10.0% 6 2.8% 14 5.7% 26
25 CAR -17.4% 1-2 7.2% 8 2.9% 13 7.4% 22
26 CHI -17.7% 1-3 5.7% 10 -6.8% 31 2.1% 32
27 CLE -18.3% 2-2 10.0% 4 6.5% 6 37.5% 3
28 GB -24.8% 1-3 2.8% 15 -1.2% 22 5.0% 28
29 CIN -28.2% 1-3 9.5% 7 7.7% 4 9.7% 20
30 TEN -28.4% 1-3 -4.9% 25 -5.1% 30 33.3% 4
31 MIA -42.4% 0-4 3.1% 14 9.1% 2 8.8% 21
32 SF -43.5% 0-4 18.6% 1 -0.4% 20 25.3% 8

Posted by: Aaron Schatz on 05 Oct 2004

comments