Most CFB models have to use preseason data that bleeds into the mid season to get good results.
I built EVO as a different model. Zero bias. Zero history. Solely on the field performance.
Here’s how this disagreed with the committee last year:
@_liamwalton @JoshNorris You have to make these thresholds:
Complete at least 77.5% of passes
Average at least 12.5 yards per attempt
Throw a touchdown on at least 11.875% of attempts