New Orleans is 2-0, yet they rank 31st in the current DVOA ratings. What gives?

I've had a lot of e-mail about this, and questions in the discussion thread over on Football Outsiders, so I thought I would explain how the current ratings work, and what that says about the Saints.

First of all, based on 2006 play only, the Saints are 11th. That's pretty clear by looking in the "2006 rank" column. However, the current ratings are not based on 2006 only. They're based on a formula that combines 2006 play with our preseason projections. Basically, I've done a ton of regression analysis on data from the last five years trying to get the most accurate prediction of where a team will rank in total DVOA for the year, based on the first two games and the projection.

And the Saints come out 31st.

I should point out, in their defense, that I still have not had the time to add a home-field advantage adjustment to these ratings, and the Saints have played both games so far on the road. If I adjusted for that, the Saints would rise a couple of spots.

Anyway, the Saints are low right now because they had the lowest preseason projection out of all 32 teams. And while we've seen some evidence that projection might be wrong, we have not seen enough evidence to toss it out entirely.

Why was the projection so low? Start with the fact that the Saints were one of the worst teams in the league last year. They have a new coaching staff, and in general teams with first-year coordinators suffer a little as they learn the new system (ask the Redskins). Sure, they replaced a terrible quarterback with a good quarterback, but remember that nobody two weeks ago knew if Drew Brees' shoulder would hold up. Frankly, we still don't know -- it's only been two weeks. The projection system has a variable for offensive line experience, and the Saints don't have much. Only two of the linemen started for the team last year, and only one started for the Saints two years ago. On defense, the Saints were much better against the pass on third down than on first and second down, and historically that tends to indicate a defense that will decline the next year.

Of course, the projection system isn't perfect, and there are tons of variables that it doesn't consider, and every year there will be teams that defy it. Sometimes, frankly, we know before the season even starts which teams the system is wrong about (Oakland, this year). Is it wrong about New Orleans too?

I went back to look for every team from the last five years that matches the following characteristics, like New Orleans does:

8-8 or worse prior year
Negative DVOA projection
Wins first two games with DVOA at least 25% higher than projection was.

Nine teams match those qualifications and here is what happened to each one, ranked in order of the difference between their performance in Weeks 1-2 and the preseason projection.

2002 Chargers: Coming off a 5-11 season, beat CIN 34-6 and expansion HOU 24-3. Finished 8-8.

2001 Chargers: Coming off a 1-15 season, beat WAS 30-3 and DAL 32-21. Finished 5-11. Same thing, both years -- and the weird part is, those fast starts came with different head coaches.

2003 Bills: Coming off an 8-8 season, demolished NE 31-0 and beat JAC 38-17. Finished 6-10.

2002 Panthers: Coming off a 1-15 season, beat BAL 10-7 and DET 31-7. Finished 7-9.

2004 Falcons: Coming off a 5-11 season, beat SF 21-19 and STL 34-17. Finished 11-5.

2006 Saints would go here.

2001 Bengals: Coming off a 4-12 season, beat Eventual Super Bowl Champs NE 23-17 and Defending Super Bowl Champs BAL 21-10. Finished 6-10.

2004 Lions: Coming off a 5-11 season, beat CHI 20-16 and HOU 28-16. Finished 6-10.

2003 Vikings: Coming off a 6-10 season, beat GB 30-25 and CHI 24-13. Finished 9-7.

2003 Redskins: Coming off a 7-9 season, beat NYJ 16-13 and ATL 33-31. Finished 5-11.

Nine teams. Average record: 7-9. Only one made the playoffs after their surprising start. Are the Saints going to be as bad as we projected? Probably not. But are the Saints good enough to follow in the footsteps of the 2004 Falcons? Probably not.

Post by Aaron Schatz