Where BPI and perception differ

Whenever a new team rating is put out there, fans tend to examine the list and say "Team X is ranked too high" or "Team Y is ranked too low" based on how the ratings match their general perception. One commonality we've found with these controversial teams when it comes to BPI is that people focus on the highlights or lowlights of a team's résumé rather than how it's performed on the whole.

For a thorough way to look at the consistency of college basketball teams, you can turn to the BPI for help. A team's season-long BPI rating is built on its individual game ratings on the same 0-100 scale. Looking at how far, on average, each game rating is from the team's season rating gives a measure of how variable a team's performance has been. We like to call this metric "variation," where lower numbers indicate more consistency and higher numbers mean more variability.

For example, if you had a team that had a Game BPI of 75 in every game -- and therefore a season-long BPI of 75 -- it would have a variation of 0 as a perfectly consistent team. But if you imagine another team that also had a season-long BPI of 75 but had all sorts of game BPI scores -- say, 30, 46, 79, 85, 91, 95 and 99 -- that would be a much more inconsistent team with a variation greater than 20.

Using this metric, the list of the most consistent (lowest variation) teams in Division I does not provide any surprises: Consistently strong teams such as Kentucky and Syracuse and consistently awful teams such as Grambling State and Fairleigh Dickinson. On the other hand, looking at the most inconsistent teams among the top 68 in BPI reveals some interesting case studies.