Catherine Rampell noted earlier Monday afternoon that the Intrade odds for the "Democratic candidate" winning the 2012 election and for Obama winning the 2012 election don't match up perfectly. I downloaded the data and charted the spread.
The clearest interpretation I can give the spread is that it reflects the odds that Obama will die in office, which has happened to eight of his predecessors. But unfortunately for data lovers, it's pretty hard to credit this as a meaningful forecast. The spread has actually gone negative on many days, implying that people thought—as recently as Sept. 19—that there was some chance Obama could win the election but not as the Democratic nominee.
The real issue here is simply that the Intrade markets don't have enough trading in them to process information efficiently. Arbitrage opportunities arise all the time without being instantly exploited by highly leveraged hedge funds, and consequently weird stuff like this keeps happening. Somewhat ironically more "noise" traders buying and selling for no real reason might help eliminate these things by creating a deeper market that it would make more sense for arbitrageurs to dive into.