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 10 was a let down. While the model picked 11 of 14 winners, which is great, Vegas managed to pick 12, which is greater. In addition, I went 6 - 7 - 1 against the spread (DET was a push), marking only the 2nd week I have been below 0.500.
On the season, my model is pretty much on the money with point spreads on average ( +0.1 per game), which is over a full point better than Vegas. That is mostly why I am 54.5% against the spread for the year.
In Week 11, I disagree with the Vegas predictions in 3 games, which ties for the most this year. I need at least 2 of these to stay ahead.
My model is probably picking ARI over SEA because Russell Wilson had a terrible game in week 10 and so the weighting I use is probably under-rating him. We’ll see.
The impressive week 10 performance by the Colts increased their win probabilities the rest of the year. I have a 50.8% win probability against Green Bay which is basically a coin flip. Once we make it past TEN, the game probabilities are more certain, with the exception of the Steelers.
For the year, the expected win total increased to 9.8.