You tend to get some fairly significant deviations from KP to AP rankings, once you get to about #8 for example since it is margin based rather than W/L based. Like all good analytics they should still never be taken as gospel, since they have known limitations.
If I gave you 7 to 1 odds for the following group to win the National Title would you take it? Based on KenPom rankings in Mid-December, I took 10 teams that had strong KP rankings but were still lagging in odds to win the National Title, with a targeted fixed payout. **
Baylor 50-1
Florida St 80-1
Butler 125-1
Purdue 80-1
Ohio St 28-1
Iowa 125-1
Penn St 80-1
Seton Hall 80-1
West Virginia 70-1
San Diego St 70-1
If I had avoided Ohio St, which cost me the most, it would be 9-1. For the most part the KP rankings projected fairly well as all those teams with the exception of Ohio St.
** I am a pansy when it comes to betting so the payout is low.