This is an exercise in building a simple blind predictive model based on a team’s historical stats in 4 areas: pass and rush efficiency, and avg EPA for special teams and penalties. The season-long experiment will track how well future games are determined by past statistical performance and compare that record against Vegas betting lines.
All data used originates from Pro Football Reference and nflFastR.
Last weekend, the model went 2 for 4 straight up and 2 for 4 against the spread. That’s not good, but with only 4 games the variance was high and so I’ll live with it.
Vegas hit on 3 of 4 straight up, so the model falls back to 3 games behind on the season with an overall 66.4% prediction rate. Against the spread, I’ve gone 56.3%, which is much, much higher than I had hoped for.
The model agrees with Vegas on a Green Bay victory, but it continues to throw shade at Kansas City, favoring Buffalo. I don’t agree with that pick, but I’m not the boss.