As discussed in Part I, Pythagorean Projection is a technique for projecting a team's future wins based on their points scored and allowed the previous season.
In Part I, I looked at how it performed over 14 seasons of data (projections from 1993 to 2007, corresponding resuls from 1994 to 2008). In Part II, I'm going to look at the 2007 projections as compared to the 2008 results.
As with Part I, the results are ranked according to the difference between the projected and actual wins, as an absolute value, sorted from smallest to largest difference.
Team | Proj. | Actual | Diff. | Comment |
Raiders | 4.9 | 5.0 | -0.1 | |
Texans | 7.9 | 8.0 | -0.1 | |
Saints | 7.8 | 8.0 | -0.2 | |
Eagles | 9.1 | 9.5 | -0.4 | |
Vikings | 9.5 | 10.0 | -0.5 | |
Colts | 12.5 | 12.0 | 0.5 | |
Steelers | 11.4 | 12.0 | -0.6 | |
Redskins | 8.7 | 8.0 | 0.7 | |
Cardinals | 8.1 | 9.0 | -0.9 | |
Buccaneers | 10.0 | 9.0 | 1.0 | |
Bears | 7.6 | 9.0 | -1.4 | |
Rams | 3.7 | 2.0 | 1.7 | |
Cowboys | 11.0 | 9.0 | 2.0 | Tony Romo hurt |
Bills | 4.9 | 7.0 | -2.1 | |
Broncos | 5.7 | 8.0 | -2.3 | |
Chiefs | 4.5 | 2.0 | 2.5 | |
Patriots | 13.8 | 11.0 | 2.8 | Brady hurt week 1 |
49ers | 3.7 | 7.0 | -3.3 | Singletary take over week 8 |
Chargers | 11.3 | 8.0 | 3.3 | |
Giants | 8.6 | 12.0 | -3.4 | |
Jets | 5.4 | 9.0 | -3.6 | Favre comes to town |
Bengals | 7.9 | 4.0 | 3.9 | Carson Palmer hurt |
Browns | 8.5 | 4.0 | 4.5 | Derek Anderson, Brady Quinn hurt |
Titans | 8.1 | 13.0 | -4.9 | Kerry Colins comes in, week 1 |
Packers | 11.5 | 6.0 | 5.5 | New QB; Defense collapsed |
Lions | 5.7 | 0.0 | 5.7 | Was 2007 a fluke season? |
Jaguars | 10.7 | 5.0 | 5.7 | |
Ravens | 5.0 | 11.0 | -6.0 | New QB, HC |
Panthers | 5.6 | 12.0 | -6.4 | Jake Delhomme stays healthy |
Seahawks | 10.7 | 4.0 | 6.7 | Injuries, HC issues |
Falcons | 4.0 | 11.0 | -7.0 | New QB, HC |
Dolphins | 3.8 | 11.0 | -7.2 | Parcells; new QB, HC |
Broken down into the ranges used in part 1, we have:
Range # of Teams % of 32 Classification 0 ≤δ≤ 2 13 40.6% Reasonably Close 2 <δ≤ 4 9 28.1% Moderately Close 4 <δ≤ 8 10 31.3% Wrong 8 <δ 0 0.0% Yikes!
(Note that in Part I, the difference was expressed in winning percentage; here it is expressed in games won in a 16-game season.)
Comments:
The Lions had a terrible pass defense in 2007. 4 of their first 6 wins were against teams with weak pass offenses (only the Vikings and Broncos were ranked over #18).
This also shows a weakness of Pythagorean Projection: it doesn't easily produce extreme values. Only 3 teams (Rams, 49ers, and Dolphins) were projected to have fewer than 4 wins, and only 2 teams (Colts and Patriots) were projected to have more than 12.
Perhaps most interesting is that in the cases where Pythagorean Projection missed by the most, there are some typical issues: coaching, QB changes, and key injuries most often.
The 1994-2008 data set (detailed in Part I) had 21.9% of the projections fall at least 4 games (or 25%) away from the actual winning record. The 31.3% this range in 2008 is a notable increase over that. For a 32-team field, the difference is about three teams, which means the remarkable turn-arounds of the Ravens, Falcons, and Dolphins can account for the difference. It will be interesting to see if we see similar turn-arounds (or collapses?) in 2009.
Coming up
In part 3, I'll take a look at the projections for the 2009 season.