Wakeforest22890
Snowpom
They also lost to TX A&M twice, Vandy and UTEP.
And Duke lost to Wake and Notre Dame, both outside the top 100 of KenPom.
Is there a statistically significant difference in losing two bad games versus losing four bad games over the course of a 33 game season? I would suspect there isn't since there's such a high variance at this sample size.
It boils down to this: are we going to use sample size 33 where we weight wins and losses as equal and don't consider margin very much, or are we going to use sample size 33 times 65 (for possessions) as the base unit, which increases the sample size, thus lowering variance? Just from a statistical vantage point it makes more sense to use points per possession.