I rarely bet on the NFL given the unpredictable nature of it. I can count on both hands how many NFL bets I've placed in my life. Not counting Super Bowl props or something. But Regular season games, probably less than 10 ever.
So if you know me, you've seen where I've made excel models using regression analysis to predict future outcomes. Did this to find expected home runs prior to the 2022 season, Matt Carpenter was in line for a big year according to this, I was laughed at for wanting the O's to take a flier, and if he didn't get hurt he probably hits 90 homers for the Yankees last year (slight exaggeration).
Anyway, I used 2021 NFL stats to find out which ones correlate best to scoring margin. Net passing yards, net TD%, and net 3rd down efficiency. After all the games in 2021, the R^2 was .99+. So 99% of the scoring margins can be accounted for by those three stats.
Started this after a couple weeks into 2022. Kept track of the R^2, it was low after a few weeks, not enough samples. R^2 was up to .80 by week 6, and this past week was .90, highest of the season. I looked back at every week since week 6 and If I bet on games that my expected margin is 5 or 6 points off what Vegas thinks, no more, no less, I would have won 68.8% of those bets, and in no single week would I have done worse than .500.
Decided to place 5 bets this week. Went 1-4. The 1 was the Giants coming in the backdoor in garbage time.