I’ve really enjoyed the ongoing empirical experiments at the Iowa Electronic Markets, where people can use real money to trade futures on political (and other) events. As can be expected, political polls tell a very different story than markets where real money is at stake.
Tonight, with the marginal victory for Clinton in Indiana, and the large win for Obama in North Carolina, a major shift happened in the future markets at IEM. See below for the graph:
As pointed out on IdeoBlog, the likelihood of a Clinton victory, according to these futures, dropped today from 28% down to 12%… a 55% change. That’s an incredibly significant shift, and it puts Obama back in the probability range that he spiked to after Super Tuesday.
Interestingly, there are still arbitrage dollars to be made on John McCain, whose futures are only trading at around 94.5%. You can make 5.5% or so just by buying the obvious winner. Of course, I believe IEM still limits your total funds to $500, so I’m not sure you can really make that much money here.
It’ll be very interesting to see the presidential futures for the final candidates once they are out – I’m expecting a very different story than the polls to date have been indicating.
Psychohistory 
After Pennsylvania, I issued a challenge to any Clinton backers–I’d take bets of any size at 2:1 odds on Hillary becoming the Democratic nominee. No takers.
Of course, I was that confident because I knew I could arbitrage against the real probabilities!