Below please find the 2015 week nine NFL results based on the model we built last year. This model uses a learning algorithm fed by data from the last eight weeks. To date, success rates this season are as follows: Week 2, 50%: Week 3, 62.5%: Week 4, 60%: Week 5, 64%; Week 6, 57%; Week 7, 50%; Week 8, 71%. This week the model fell to 53.8%. The crowd hit the same percent, (assuming that the record below is right about the crowd prediction on the STL @ MIN game. I seem to remember that the crown picked MIN, but I don't have a way to check that now.)
Of the six games of thirteen missed by the model, one had a six point spread, four had a three point spread, and one had a one point spread. Any of these could have gone the other way, but they didn't, though this makes it clear that it was an interesting week for football.
The two games where my intuitions diverged from the model (the STL @ MIN game and the OAK @ PIT game) were both right according to my intuitions, but not the model. This raises some cognitive science questions about intuition. What am I seeing that the model is missing? It's hard to say.