Below please find the 2015 week five NFL results based on the model we built last year. This model uses a learning algorithm fed by data from just the last four weeks. To date, success rates this season and last are as follows:
This week, the model was 64% correct (9 games correct / 5 games incorrect.) The crowd faired better at 79% correct (11 games correct / 3 games incorrect.)
The one game that the model got correct and the crowd incorrect was PIT 24 over SD 20 on Monday night. The other two misses on the part of the crowd were also missed by the model. These were CHI 18 over KC 17 and CLE 33 over BAL 30.
To help make the details below clear, keep in mind that we have slightly altered the 2014 model in various ways to yield three sets of predictions. Model 1 (M1) is the procedure from last year and is based on normal averaging. M2 is based on moving averages, and M3 is based on weighted percentages (thereby allowing games later in the season to count more than earlier games.)
We're setting up for week six. If last year is any indication of the strength of past data in determining the future within our model, we should see a big jump. I'm curious to see if this will indeed be the case.