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.
Week 9 was a good week. Well, not for the Colts.
The model went 11 for 14 and beat the spread in 9 of 14 games. It went 2 for 2 on its upsets (NYG over WAS, NO over TB) but Vegas changed the line on the Charger/Raider game favoring the Raiders at the 11th hour and so I had to give back one of those games.
Still, I am now up 2 games on sin city and am beating the spread with an impressive 55.7% accuracy rate.
In Week 10, Vegas and I are predicting the same winners in every game. So unless the lines change, neither of us is going to gain or lose ground on the other. It makes for a boring week, but I can still expand my margin over the spread.
There are only a couple of games, where the spreads significantly disagree. Vegas likes Pittsburgh, Green Bay and New Orleans much more than I do. Green Bay is probably because Jake Luton is starting for the Jaguars and my model does not know that. Similar issue with SF-NO. I’m not sure what is driving PIT-CIN.
Philip Rivers’ poor performance against the Ravens, dropped his efficiency numbers dramatically and shaved almost 1⁄2 a game off the predicted season win totals, which now stands at 9.6.