"The ranking system is based upon a hypothetical vote by each college basketball team of who is best in college basketball. The “vote” is derived from an analysis of what each college basketball team knows about its opponents. Each team can only vote for teams that it has played. And it knows nothing about the results of any other games."
Can you give me the detailed explanation on this? Would love to know how you are weighting discreet factors.
For each team's "vote" a list of teams they have played is created. Two factors, game location and margin of victory, are considered to rank order the teams. But pure margin of victory is not absolute. Each victory is placed into 9 categories with a linear weighting for each category. This forms the "opinion" of the opponent. This factor is modified to consider game location. Once an "opinion" is done for each team, the ranking can be done. It's important to note that this ranking by this team does NOT consider where it falls on this list (cannot "vote" for itself). The top 25 opponents (which can include same team multiple times) are given 1 point for each place on the list (1 point for last team on list, 2 for 2nd to last team, etc.). All ties (based upon the weighting described earlier) are averaged, so that every tied team gets the same number of points on the "vote".
This is done for every team in Division I. Each vote's points are summed and the teams are then ranked based upon the points they received in the poll.
The category weightings are: 9 - a 25+ pt loss; 8 - a 15-24 pt loss; 7 - a 10-14 pt loss; 6 - a 4-9 pt loss; 5 - any OT game or margin of fewer than 4 pts; 4 - a 4-9 pt win; 3 - a 10-14 pt win; 2 - a 15-24 pt win; 1 - a 25+ pt win. So a larger weighting is given to an opponent that beat the team - larger weighting = better team. Each home game is modified by +0.5, each road game is modified by -0.5.