I had a campaign consultant come speak to one of my classes today. He seemed to think that Obama will sweep Clinton on Tuesday, even though the polls have them pretty much tied or even give a small lead to Clinton (Disclosure: He works for Obama). His rationale behind this is that all the major polling firms are using their tried-and-true models of voter turnout to predict who will show up to vote. However, those models are substantially underestimating turnout by young voters and African Americans this year. Pollsters, he figures, have figured this out by now, and probably have much more Obama-friendly numbers in their private records, but they're not releasing these data publicly since it relies so much on hunches. They at least have scientific rationales for the numbers they're releasing now.
We'll see. Meanwhile, one of my students is urging me to buy some Clinton stock on Intrade and then selling right after her anticipated win in Ohio. She's selling at 19 right now, so it's not a bad buy. I'm tempted.