The problem I see with offshore lines as an accurate predictor is that those that those that bet offshore are not an accurate sampling of the american public.
Demographics have shown that offshore gamblers are about 90%+ male and 90%+ white. Moreover, what about the religious right zealots that make up about 1/3 of the republican party? Think they are betting online? Despite a few hypocrites, most probably believe making an offshore bet on the presidential election will sentence them to eternal damnation.
So what are we lacking in our sampling? Women, minorities, and the right wing zealots. Hmmm. come to think of it, those cancel each other out, so maybe this is an accurate sampling :toothless