The rationale in this thread for why Vegas might believe Duke is much better than last season is all wrong, since Vegas actually believes Duke is much worse (and/or Elon is much better) than last season.
The spread Vegas and the neutral betting public set last season is a much better benchmark for how smart people are assessing Duke and Elon than the final result of the game last season.
Duke was a 26-point favorite in a semi-road game in Greensboro last season (with +2 points for Elon in that arena). Duke is currently a 19.5-point favorite at home (with +3.5 points for Duke at home). This means Duke is believed to be about 12 points worse than last season relative to Elon. Much of this is due to Elon returning nearly their entire rotation and being better than last season, but not all 12 points can be attributed to that.
Another more concrete indication that Duke is believed to be worse than last season are the national title odds. Duke was 3-1 to 4-1 (20-25%) to win the title at this time last season. Duke is now 6-1 to 7-1 (12-14%) to win the title this season.
The expectations for Duke 2018 are much lower than for Duke 2017 at the same point in the season, against Elon or against anyone else. This season's team is highly suspect despite its raw talent advantage relative to the field, and this is all indeed priced into the line. Vegas will be 9 points off one way or the other, on average, but they're not stupid.