I added projected wins this week so you can more closely see the differences between teams as indicated in the rankings are often statistically insignificant. The projected wins assumes that rankings points are distributed from top to bottom in a way that's perfectly analagous to win distribution. In other words, the standard deviation between rankings points equals the standard deviation between wins.

The stats that are the basis for these rankings are not necessarily the ones that correlate to winning the most this week or even this year. They are the ones that have correlated the best, on average, over the past decade or so. The strongest case for a new entry is Net Third-Down Percentage. Further research on that is coming after the new year.

The benchmark this week, for correlative purposes, is this: average top 10 in winning percentage, .700; bottom 10, .300. The closer you get to these, the better the single stat at this point in time for 2006. These are the winning percentages of the top and bottom 10 team in each of our categories: Net Yards Per Pass Attempt top 10, .650, bottom 10, .352; Net Points Per Pass Attempt top 10, .614, bottom 10, .336; INT-Net (per game) top 10, .664, bottom 10, .336; Move Net (runs plus completions) per game top 10, .643, bottom 10, .350; Net Red Zone possessions (per game) top 10, .664, bottom 10, .350. Note the records for some other categories not included: net third down percentage top 10, .693, bottom 10, .400; net yards per rush top 10, .536, bottom 10, .450; net red zone TD efficiency top 10, .643, bottom 10, .378; net average yards on first down top 10, .529, bottom 10, .443.

One good criticism of this system is that it doesn't adjust for quality of opposition. But changes made due to this are usually marginal. Plus, there's a case to be made that quality of opponent works both ways. And, technically, the Yerkes-Dodson arousal curve makes a good case for teams playing up or down to their level of competition.

The points teams earn in each category merely reflects where they rank among the 32 teams in each stat. For example, 1 point means the team is first in that category, 21 means they're 21st. The Power number is the sum of all these rankings. So, the lower the Power number, the better the team.

The point of this Power Index is to learn why teams win or lose. It's not to be used as a predictive tool. While we are confident that the good teams and bad teams as reflected here will play that way prospectively on balance, it's impossible to know which teams will play well or poorly in one game, as performance can deviate too far from average "on any given Sunday."

Key: YPA Net=Yards per Passing Attempt (sack yards included); Move Net=runs plus completions; PPA=Points per Passing Attempt (TD passes times 7 divided by passing attempts); RZ Net=Inside the 20 possessions. W2005=2005 final power ranking weighted 25 percent. Obviously, all Net categories are what's achieved by the offense minus what's allowed by the defense. They are calulated on a per-game basis to adjust for byes.

Click Here For The Previous Stat Power Index