Methodology

My NFL model, which performed very well for many years, tailed off in 2014 and then had a bad year in 2015.  As a result, I have taken time this off-season to develop several new predictive models which have tested well and I’m encouraged by their out of sample results.  One of my new models incorporates elements of DVOA (Defense-adjusted Value Over Average) to provide micro-level context and is based on the great work the folks at Football Outsiders have done.  DVOA measures a team’s efficiency by comparing success on every single play to a league average based on situation and opponent.  The other model is a re-engineered compensated stats model.  I’ll be using them both this season along with only the best situations out of the hundreds of NFL systems that are in my database.  The combination of new predictive models and strong technicals along with subjective factors, and without the encumbrance of the write-ups, should get me back near my historical average (@56%) after my first ever below .500 season last year.