As of 7:20 EST Friday, on Intrade, Obama has moved up to $0.48 on the dollar vs. $0.50 on the dollar for Clinton. Clinton is showing as a close third to Edwards in Iowa dropped her from a high of $0.70 last week.

On the Republican side, McCain is trading at about $0.31 on the dollar for Republican nomination now ahead of Giuliani at about $0.27 with Huckabee at $0.17 and Romney at $0.14. Giuliani was trading at $0.45 at the beginning of December.

A political stock market, a class of info markets, is the best known device for aggregating poll data and all other available data, public and private about the chances of the candidates.

Obama-NH last trade at $0.65 out of $1.00 vs. $0.44 for Clinton. For SC, it's Obama $0.62 with Clinton's last trade at $0.42. Some arbitrage opportunities here to lock in more than $1 by selling short each candidate. McCain is trading at $0.75 for NH and in SC, The Rest of the Field (There's no Huckabee security here for some odd reason) is trading at $0.40 vs. McCain's $0.30.

On the Republican side, McCain is trading at about $0.31 on the dollar for Republican nomination now ahead of Giuliani at about $0.27 with Huckabee at $0.17 and Romney at $0.14. Giuliani was trading at $0.45 at the beginning of December.

Iowa must be a massive statistical outlier.

OTOH, winning even in such a state can translate into a funding boost. As I see it, there's a large crowd that wants to be on record as having donated to the eventual nominee, but doesn't want to spend a lot of money doing it. Donating to the early leaders reduces costs.

Or maybe these market-based predictors are really just reactive.

Iowa must be a massive statistical outlier.

Historically, yes. The media focus on itb because it's the first, and then they make a lot of noise about the outcome -- most of which usually goes right down the memory hole sooner or later because in most cases the candidate who wins Iowa doesn't go all the way unless he goes on to win New Hampshire also.

There have been exceptions (just ask any reporter who was in Iowa this week, about 1976), but the weight of the evidence simply doesn't support the attention focused on Iowa as a predictor of the ultimate outcome. But the media can't simply admit that, by this time in the process they're so starved for results, any results, that they'd hype a system that awarded delegates based on patterns read in pigeon droppings.

"Or maybe these market-based predictors are really just reactive"

I agree with Jane.

What I don't understand is why McCain has such high numbers. What does he have apart from a strong showing in NH?

A ballot is also a high-accuracy poll. The markets had better react to news. Whether they are more reactive than other predictions requires comparing them to other sources of predictions. I'd say they were the most predictive of any public source.

A political stock market, a class of info markets, is the best known device for aggregating poll data and all other available data, public and private about the chances of the candidates.

Sam, can you offer any proof that this is the case?