The model is a standard statistical regression model. Independent variable is the game margins of all games played. Dependent variables are the team strengths of all teams, plus one home-court advantage variable.
As of 1/11/2003, there the variable sizes were 1784 games, 467 teams (394 Iowa teams, 73 non-Iowa teams).
Each game is considered an equation involving 2 team variables, a home-court advantage variable, and the independent variable of the actual game margin. Thus there are 1784 equations and 468 variables (467 teams + 1 home-court advantage variable). Since there are more equations than variables, a least-squares regression is performed to obtain the team strengths and home-court advantage. Find a standard college-level statistics book for more details.
In a real sense the only criteria is game margin. The rest of the computation is strictly statistics.
I have been ranking Iowa high school basketball and football teams since the Fall of 2001.
You can reprint or reference these rankings. Please let me know if you publish an article.
I experimented with other models, but the simple model described above has proved to provide the best predictions available anywhere. There are many examples I can describe, but the most recent one is this. I read about the Pomeroy-Palmer versus Newell-Fonda game in the Des Moines Register this morning. The Register described this game as an upset of #2 Pomeroy-Palmer by #4 Newell-Fonda. Before the game the computer model predicted #2 Newell-Fonda beating #10 Pomeroy-Palmer by 14 points. The actual game margin was 11.
Thank you for your interest!