The "problem" with Ken Pom is that it measures how good teams are at scoring, and defending scoring, possession by possession. It doesn't measure how good they are at winning instead of losing. To take a simple example, if one week UConn and 'Nova both play at Creighton and Xavier at home, and UConn wins both games by 1 and Nova beats Xavier by 30 but loses to Creighton by 1, KenPom says Nova, just using data from that week, is "better," notwithstanding that UConn accomplished more.
It may be that Ken Pom excels as a prediction tool, and is therefor useful for setting odds and point spreads. But when the selection committee meets, 2-0 needs to trump 1-1 against the same schedule even if the margins indicate that Nova would be a favorite if it played UConn on a neutral court the next day.
Said another way, teams need to be rewarded and ranked in a manner where the difference between winning and losing is critical. Not in a manner that defines good based on who would be favored if they played. When Ken Pom says Providence ranks high on "luck," what it's really saying is we don't think pulling games out is a skill in and of itself, but just deviations from a standard mean based on how you play over 40 minutes. That, obviously, is nonsense.