The Intrade prediction model hasn’t been wrong yet. It correctly predicted Obama’s win in 2008.
Its far more reliable than any poll since while people can lie to pollster about whom they’ll vote for, they don’t lie about where they put their money.
Material interest is a more powerful predictor of people’s intentions than sentiment, which is inherently changeable.
It's generally accurate at the time of the event. In this case on election day. It was dead wrong on the SCOTUS Obamacare decision.
Intrade is also a small enough market that a campaign can buy the desired result with pocket change, I prefer the economic model.
I’ve never invested any money there, but I do consider them as important, and valid as most polls.