So as a reader of this ongoing conversation, you may need some background on the analysis. Here is a taste.
In 2005, I started developing a simple system to track my office pool bets in a less "manual" way. Over the course of the year, I started adding tidbits and statistics to help me improve my picks. It was pretty straightforward picks against the spread (ATS), and I was using the statistics at FootballOutsiders.com, the MASTERS of football statistics, as my base.
After I while, I decided to stop wading and take the plunge. Using a combination of analysis from Football Outsiders, historical NFL statistics, and historical spread information, I started the overall analysis. The basic premise is very easy: measure teams against a moving average (week to week, the "average" team changes), find the average score for any given team on any given Sunday, combine the two to determine a spread. That was the starting premise - since then, I've realized the real art is in measuring HOW to bet ATS, not WHO to bet ATS. The "Who", over the long run, ends up being more or less a coin flip. Understanding the How and When is where the money is.
This gets us to the type of bets, how to pick the right bets, and developing a system of "Conviction". The system now ranks in the following ways:
Win Percentage (as in, which number ranked picks win the most)
Low Risk Conviction
Conviction and Win Percentage Averager
"The Juicer" (takes into account the lay and whether a bet is worth the money)
... and then I have subroutines that aggregate the results and average out the ranks.
More to come at a later date, but that's a start.