Is your formulation too complicated for me to understand? Because if it's not, I'd like to hear how you come up with these numbers (and what exactly they mean). I'd say twice a week is probably excessive, maybe once a week or every 2 weeks.
What they mean would be a lot easier than how. It's a mathematical assessment of the accomplishments of teams to date. It's based on a 60-dimension fixed point iterative technique that takes both opponents and margins into account. It's all done in a simple spreadsheet. While I don't claim any predictive characteristics, it often turns out that after each team has played 6-8 games, it starts to pick winners of games fairly consistently.
So, for example, tonight Atlanta plays at Cleveland. Hawks away are at 21, Cleveland at home is at approximately 19, the Hawks should win by 2. If they don't, the system will make adjustments tomorrow. Wizards should beat the 76ers by 8; Pistons should beat the Celtics by 3; etc. Not enough games have been played so far for the system to be dependable, but it tends to get more dependable as the season goes on.
