It could have been so much simpler.
When the NCAA went searching for an RPI replacement it came up with its own NCAA Evaluation Tool (NET): a major upgrade from the archaic previous system but a metric that also comes with flaws. And needless complications.
Earlier this season we wrote about NET's bumpy debut and some of its problems -- including the fact that it combines predictive and résumé metrics into one number and that it incorporates wins and losses unadjusted for opponent or location, to some degree. As we approach NET's first Selection Sunday, we wanted to bring a solution for what the NCAA could have done (and could do in the future) to save itself from headaches and institute a fairer selection process at the same time.
Just use strength of record.
For the at-large selection. For the seeding. For the whole thing.
(For the unfamiliar, strength of record determines the chance that a top team could achieve a certain record given a particular schedule. If we think of strength of schedule as a means of separating teams with identical records, SOR can separate teams with different records. It is more difficult, for example, for Purdue to have gone 23-8 this season with its schedule than Nevada to have gone 28-3 with its schedule.)