| Rank | Team | off | def | total |
| 1 | San Francisco 49ers |
3.23 | -8.87 | 12.10 |
| 2 | Houston Texans | 6.12 | -5.11 | 11.23 |
| 3 | Denver Broncos | 6.15 | -2.76 | 8.91 |
| 4 | New England Patriots | 10.55 | 2.10 | 8.45 |
| 5 | Chicago Bears | 0.96 | -6.40 | 7.36 |
| 6 | New York Giants | 5.30 | -0.43 | 5.72 |
| 7 | Seattle Seahawks | -1.95 | -7.49 | 5.54 |
| 8 | Tampa Bay Buccaneers | 4.61 | -0.59 | 5.20 |
| 9 | Atlanta Falcons | 2.11 | -2.49 | 4.60 |
| 10 | Baltimore Ravens | 0.98 | -3.05 | 4.03 |
| 11 | Green Bay Packers | 2.60 | -1.08 | 3.68 |
| 12 | Washington Redskins | 4.97 | 2.15 | 2.82 |
| 13 | New Orleans Saints | 6.00 | 3.52 | 2.48 |
| 14 | Pittsburgh Steelers | -3.25 | -4.03 | 0.78 |
| 15 | Detroit Lions | 3.76 | 3.25 | 0.51 |
| 16 | Dallas Cowboys | 0.55 | 1.21 | -0.66 |
| 17 | Cincinnati Bengals | 1.26 | 1.99 | -0.72 |
| 18 | Minnesota Vikings | -1.08 | 0.22 | -1.29 |
| 19 | San Diego Chargers | -1.37 | 0.16 | -1.54 |
| 20 | St. Louis Rams | -2.76 | -1.17 | -1.60 |
| 21 | Arizona Cardinals | -4.61 | -2.95 | -1.66 |
| 22 | Carolina Panthers | -3.57 | -1.50 | -2.06 |
| 23 | Cleveland Browns | -3.72 | -1.02 | -2.69 |
| 24 | Buffalo Bills | 0.04 | 4.16 | -4.12 |
| 25 | Indianapolis Colts | -2.75 | 2.12 | -4.87 |
| 26 | Miami Dolphins | -5.68 | -0.58 | -5.10 |
| 27 | New York Jets | -1.74 | 3.39 | -5.13 |
| 28 | Philadelphia Eagles | -4.47 | 0.85 | -5.32 |
| 29 | Tennessee Titans | -2.33 | 8.03 | -10.36 |
| 30 | Jacksonville Jaguars | -7.90 | 3.24 | -11.14 |
| 31 | Oakland Raiders | -2.36 | 8.89 | -11.25 |
| 32 | Kansas City Chiefs | -9.66 | 4.24 | -13.90 |
November 27, 2012
NFL ratings november 27
5 Comments »
RSS feed for comments on this post. TrackBack URL
I was wondering exactly how you’re weighting the games? I’ve read your book “Mathletics” cover to cover twice. When I used your methods described to rate NFL teams, I have New England at the top and ahead of San Fran by nearly two points, regardless of how I weighted recent games. I was just curious about the discrepancy. Thanks in advance.
Comment by Billy Smith — November 27, 2012 @ 8:58 pm
i am using a slightly different method than least squares that I would like t okeep confidential.
Comment by wwinston — November 27, 2012 @ 11:22 pm
That’s completely understandable and I can appreciate that. Thank you for the response. If I could then, I would like to ask one more question regarding the weighting of the least squares method. You mentioned in your book that a value of ƛ = 0.95 works well for a weight in professional football games, but that it could also be optimized to yield more accurate results. My question is how one would go about optimizing that value? Would you use Excel’s Data Tables in the same manner as with Pythagorean or could you add it to the Solver model as a changing cell? Maybe there’s a completely different way to go about it? Once again, thank you in advance.
Comment by Billy Smith — November 28, 2012 @ 8:42 pm
you would need to figure out the ratings with a given lambda at each point in time and figure out the lambda for each week that results (over many seasons) in best forecasts
Comment by wwinston — November 29, 2012 @ 2:14 pm
i DO NOT KNOW. this would require research using data off pro football reference
Comment by wwinston — December 29, 2012 @ 4:12 pm