Econometric Power Analysis for NFL Week 6

Econometric Power Analysis for NFL Week 6 Here are power ratings based on season-to-date lines. Just take the home team rating, subtract the visitor rating and subtract a home field edge of 2.5. Let me illustrate with Arizona's schedule: Visitor Home HomeLine Arizona Jackson -3 should be 0.5 - (-1.5) - 2.5 = -.5. Arizona Seattle -3 should be 1.0 - (-1.5) - 2.5 = 0. Houston Arizona -6 should be (-1.5) - 0.75 - 2.5 = -4.75. Indiana Arizona -3 should be (-1.5) - (-4.5) - 2.5 = 0.5. SanFran Arizona -5 should be (-1.5) - .25 - 2.5 = -4.25. The average line for Arizona was (3+3-6-3-5)/5 = -1.6. This is exactly equal to the average power rating prediction (.5+0-4.75+.5-4.25)/5 = -1.6. It works for all teams. For NFL Week 6, there are two major discrepancies: Philadelphia at Oakland is 14, should be 8 - (-4.25) -2.5 = 9.75. Arizona at Seattle is -3, should be 1 - (-1.5) - 2.5 = 0. Have fun! Arizona -1.5 Atlanta -2.25 Baltimore -6 Buffalo 3 Carolina 0.25 Chicago -2 Cincinnati -1.25 Cleveland 4.5 Dallas -4.5 Denver 1 Detroit 6.75 Green Bay -4 Houston 0.75 Indianapolis -4.5 Jacksonville 0.5 Kansas City 4.75 Miami 1.75 Minnesota -5 New England -6.25 New Orleans -6.25 NY Giants -4.75 NY Jets -1.25 Oakland 8 Philadelphia -4.25 Pittsburgh -6.75 San Diego -2.75 San Francisco 0.25 Seattle 1 St. Louis 6.75 Tampa Bay 5.5 Tennessee -2 Washington 0

The methodology of using market pricing (i.e. actual lines) to measure strength is indeed interesting. But the biggest difference in results between this method and more conventional power ratings is that you use old information as though it were current. It seems that at the very least you need to weight the more recent data more heavily. Consider your most dramatic conclusions about this week's games. You rate Oakland too high because they have dramatically underperformed (very low) expectations. You also overrate Arizona, a team which has disappointed. And you underrate Denver, a team which has greatly exceeded expectations. I doubt you will have meaningful success betting this system since its primary feature is to overweight old data that others update with newer information.

I've done work with this for all major sports. Basically, it tells you what the line would be if it weren't for ... https://www.nutshellsports.com/predictions.html This guy's been doing it for years for football. If you think changes in the market's appraisals contain too much overreaction to recent performance, than you'd want something like this. I find them useful for judging how the market's reacting. It'll help put a price on something like a major injury. For ex. if you know (from such a system) that the line would normally be -7, but because of a QB injury you see the line is -4, then you know the market is valuing that injury at 3 points. As for Oakland, it allows you to put a number on how much worse than expected Oakland is. The market has dropped it's opinion of the Raiders by ~TD.