Perhaps Austrians are against making predictions because they’re so bad at them
Intrade’s Crystal Ball
David Leonhardt
WASHINGTON - Might Intrade know the winner of Iowa tonight before many other sources do?
On the night of the Iowa caucuses in 2008, before news organizations declared Barack Obama and Mike Huckabee the winners, the odds on the Intrade markets began to suggest that both would win. Between 8:30 p.m. and 9 p.m. Eastern, the odds began rising for both Mr. Obama, then a Democratic senator from Illinois, and Mr. Huckabee, a Republican and former Arkansas governor.
By 9 p.m., Intrade – an online market where traders can bet on real-world events – gave Mr. Obama and Mr. Huckabee each more than an 80 percent chance of winning, according to Eric Zitzewitz, a Dartmouth economist who has studied prediction markets.
That pattern suggests that information from the caucuses was filtering out to people buying contracts on Intrade and affecting the prices.
But Intrade has hardly had a perfect record on nights like tonight, Mr. Zitzewitz notes. On the day of the Democratic primary in New Hampshire in 2008, Mr. Obama’s odds of victory rose to about 99 percent – before eventually falling to zero. Mr. Zitzewitz wrote, via e-mail:
“I’m not aware of anyone having studied systematically whether the within-election-day market movements have predictive power in the absence of exit poll leaks. We probably do not have a large enough sample size to say much. In contrast, it’s fairly clear that the market movements in the days leading up to elections have plenty of information content.”
In the days leading up to the Iowa caucus, Rick Santorum’s odds of victory were rising – and continued to rise through parts of Tuesday – but Mitt Romney was still given the highest chance of winning (roughly 44 percent) at 6:30 p.m. Mr. Santorum’s odds were about 30 percent, and Ron Paul’s were about 25 percent.
In a 2008 column, I looked at the benefits and drawbacks of Intrade odds in more detail.
I’m watching the results come in from an app on my iPhone. Only a few votes have been tallied so far but it’s showing Paul in the lead with 49 percent.
The idea that InTrade (or any market) is the most reliable indicator of the outcome of the election is based on the Efficient-Market Hypothesis. EMH is crap.
Clayton -
That’s a pretty incredible leap.
Haha, Paul up to 50% in inTrade: http://www.intrade.com/v4/markets/?eventId=90932