My ratings are calculated using a modified least squares method in which games are weighted by their closeness. Closeness is measured by min(score difference, rating difference). Thus, the ratings affect the weights and an iterative procedure is necessary.

The algorithm minimizes the effects of big blowouts, which are unreliable evidence as to either teams strength due to bench clearing and merciful play calling (or lack of such) after the game is out of hand. In my opinion, a close game between two teams is hard-to-ignore evidence that they ought to be rated close to each other.  Home field or court advantage is taken into account.  Its magnitude is calculated by the algorithm to get the best fit to the data.

Since I don't use any information from previous seasons things are a little squirrelly until each team has played several games. The ratings are normalized so the average Div. I-A team (football) or Div. I team (basketball) has a 50 rating. Subtracting ratings gives you a predicted point spread.

I only do this for amusement and don’t take it too seriously.  Don’t get offended if your favorite team is ranked lower than you think they should be. 

Having spent some time ranking teams with computers, I’m strongly in favor of scrapping the Bc$ and determining the Div I-A football champ through a playoff.

Thanks for you interest.