Pre-Week 1 NFL Rankings (total team performance)
The rankings shown in this post display the total GE number for each team. The GE number is calculated by looking at each team's offense and defense separately. For offense three key football stats are used: points scored, total yardage gained, and turnovers allowed. Likewise for defense, the three stats used are points allowed, total yardage allowed, and turnovers forced. The data is normalized and combined in a fashion that provides some valuable insight as to how good each team's offense and defense really are compared to the rest of the league. A GE number for offense and defense are generated separately and those two numbers are added together to provide the total GE number.
Here are the pre-Week 1 GE number rankings.
1 Seattle 2.473981149
2 New England 2.459384972
3 Kansas City 1.776640836
4 Cincinnati 1.753429164
5 Green Bay 1.230587891
6 Carolina 1.185332884
7 Arizona 1.092807727
8 Houston 0.925577087
9 Pittsburgh 0.888967724
10 NY Jets 0.75405583
11 Minnesota 0.596521933
12 Buffalo 0.400603374
13 Dallas 0.317694567
14 Denver 0.152216983
15 Washington 0.087661419
16 NY Giants 0.078667621
17 Oakland -0.088347102
18 Tampa Bay -0.419411519
19 Detroit -0.428524774
20 Los Angeles -0.490613571
21 Baltimore -0.609871911
22 New Orleans -0.616936584
23 Chicago -0.674629602
24 Jacksonville -0.723836076
25 San Diego -1.035090852
26 Tennessee -1.053310586
27 Indianapolis -1.068685441
28 Philadelphia -1.192641256
29 Atlanta -1.687423716
30 San Francisco -1.809569988
31 Cleveland -1.977599596
32 Miami -2.297638589
How I came up with this? Well, I was bored one day while living in Portland in 2000 and figure I try to use some of the stats I learned in college to design a way to rank NFL teams that can hopefully find value in betting lines. This algorithm has been refined over the years and I am proud to say that the correlation between the GE number and NFL regular season records is over 90%. This is superior compared to most other ways of ranking teams like just using yardage. Football Outsiders has a good stat called DVOA (Defense-adjusted Value Over Average) and this number is used often on ESPN. Yet, the correlation with that number the NFL regular season records is about 85 - 90% and is rather complex.
By using a simple way of evaluating teams, I am confident that using three key stats like points, yards, and turnovers can explain why some teams will rise to the top and why some never get out of the cellar. It is, however, the combination of those numbers and then tracking how that combination changes throughout the course of the season that can really help in predicting NFL outcomes. A GE number of 2 is considered Super Bowl contender worthy. A GE number of 1 is playoff worthy. A GE number of 0 is average. A GE number of -1 is pretty bad and will probably garner a top 10 NFL draft pick the following year. A GE number of -2 pretty much means you are in for a long season!
As for sample size, yes it is small compared to other sports like baseball and basketball but if you consider that each NFL game is comprised of 50 - 60 offensive plays for each team, then throughout the whole season, you are considering nearly 1000 plays. Points, yards, and turnovers are a result of those 1000 plays and if you treat the data as if it were normalized and combine them, you get the GE number which has proven to be a very good predictive measure of NFL success. This could also help with daily fantasy leagues which will be explained in a later post.
For now, it's fair to say that Seattle should thump Miami. Don't let the 10.5 point spread scare you. Seattle should win by about 20 points according to the GE model. I will post predictions for this week's game very soon!