Model 1 (M1) is the procedure from last year and is based on normal averaging. So far, this has been the most accurate model. M2 is based on moving averages, and M3 is based on weighted percentages (thereby allowing games later in the season to count more than earlier games.) AVE is, of course, the average of the three.
The Average Power Ratings are based on the summation of M1, M2 and M3 and appear to be rather unintuitive on inspection. The 4-4 Seahawks, for instance, top the Broncos and the undefeated Panthers. The reason for this is that the Seahawks had two very high point spread games, won the other two by just a little, and also lost the other four by just a little. This suggests to me that they are a better team than their win/loss record shows. But, as we all know, the wins and losses are what matters, though perhaps not for prediction. The Seahawks are erratic and hard to predict. Factors like this must be taken account in future models.