By pulling apart the statistical models used by services like KenPom and Sagarin, I am learning how subjective these systems really are. Just because you assign a numerical value to an input does not make the input objective.
For example, Commissioner will point out how many points and other stats lost each 2018-19 HL team is experiencing. That shows up in these stats models in the form of pre-season numbers like Adjusted Offense, Adjusted Defense, Offensive Tempo, etc.
But the stats models do not account for players like Loudon Love on Wright State last year. His contributions to Wright State's success were not even factored into last season's KenPom pre-season model because he was not one of the 30 top freshmen recruited in his class of 2016. Then, he redshirted the 2016-17 season.
Basically, these models simply try and quantify known factors, but cannot account for unknown factors at the mid-major level because rarely does a mid-major ever recruit a top 30 player to their rosters. In other words, they are simply educated guesses with numbers assigned to them.
If you review the Adjusted Tempo numbers of most teams, they are much higher than their actual numbers from the prior season. Since the rule changes were implemented for the 2015-16 season, Nagy's teams Adjusted Tempo have been 68.3 at SDS in 2015-16, 69.0 at WSU in 2016-17, and 68.3 at WSU in 2017-18. Yet, KenPom projects this year's Nagy coached team to average 73.5 possessions per game.
The Titans' Adjusted Tempo since the rule changes was: 73.7 under Ray in 2015-16; 72.6 in 2016-17; and, 72.4 in 2017-18.
Mike Davis at TXSO put up 67.8 in 2015-16, 68.6 in 2016-17, and 71.7 in 2017-18. Last year's team at TXSO had a losing record, but the teams playing more to Wright St's tempo had winning records because they had much better offensive efficiency.
None of this should be surprising because if your team is losing in most games, they will try and claw their way back in each game with faster possessions in order to cut the deficit. But teams with winning records will lengthen late game possessions in order to preserve their lead and run the clock.
It's just the long way of saying that these pre-season stats don't mean much especially when a team has a large turnover in personnel. If Nagy is winning or close in most games, I don't think that Wright State will speed up to 73.5 possessions per game. Likewise, if our Titans are close in most games, they will not run 75 possessions per game.
But it's important to understand these stats because this year's NCAA tournament selection committee will replace the old RPI with a new model that includes some of the KenPom type stats in order to award at large bids and seed the teams.
The fact that there is no difference in the pre-season HL rankings at KenPom from the HL pre-season poll of coaches, SIDs and two media people from each team's market really shows that Commissioner's more straightforward observations have equal validity with the pre-season stats models - especially in mid-major conferences.