A test for political prediction markets

Ron Rosenbaum writes:

"So I was down in DC this past weekend and happened to run into a well-connected media person, who told me flatly, unequivocally that ‘everyone knows’ The LA Times was sitting on a story, all wrapped up and ready to go about what is a potentially devastating sexual scandal involving a leading Presidential candidate. ‘Everyone knows’ meaning everyone in the DC mainstream media political reporting world….By the way, it’s not the Edwards rumor, it’s something else."

Slate’s Mickey Kaus thinks this speculation is worth repeating.

If there really is some “devastating sexual scandal involving a leading Presidential candidate” that the general public is not aware of but everyone in the “DC mainstream media political reporting world” knows then we have a perfect test for political prediction markets.

The price of candidates in prediction markets should take into account information known only to DC mainstream political reporters.  The prediction market Intrade gives Obama a 12.6% chance of capturing the Democratic nomination.  I follow politics fairly closely and this 12.6% seems low to me based on the information I have.  Obama has lots of cash, is doing well in the Iowa polls and lots of Democrats are worried about Clinton’s chance of winning in a general election.  So it seems very possible to me that Obama’s prediction market price is being negatively influenced by something that the general public is unaware of.

Intrade gives Clinton a 71.3% chance of winning the Presidential election and a 47.7% chance of becoming president.  Some people are speculating that Rosenbaum’s devastating sex scandal involves Clinton.  If this is true then her Intrade prices should be much lower than they currently are. 

In sum, if many media members have private information about an Obama sex scandal then the Intrade prediction markets are working as prediction market enthusiasts would hope and thus the Obama scandal represents a victory for prediction markets.  If, however, many members of the media have private information about a devastating Hillary Clinton scandal then prediction markets, in this instance, have failed us and supporters of prediction markets will have to rethink our enthusiasm towards them.

GD Star Rating
Tagged as:
Trackback URL:
  • This might be a test of whether these markets contain all info available to anyone, but that isn’t a very interesting thing to test. What we most want to test is how they compare to other available long-lasting institutions to which we might learn to defer. What other institution might reflect this scandal sooner? This makes a stronger prediction re Clinton than Obama, as there another obvious reason to expect him to do poorly.

  • Recovering irrationalist

    A possible experiment:

    Post that to a few blogs and forums popular with DC political reporters. Do them all at once. Include info on how Intrade works and how to use it.

    Track carefully what happens to the prices.

    If rumours are true, prices should quickly change further through opportunists who hadn’t thought of that before, or people anticipating the same.

    Things will get very interesting if someone with an agenda decides it’s worth losing a little money to affect the price and it’s interpretation.

  • It seems really hard to isolate what effect a rumor we can’t even articulate is having.

  • http://www.byrneseyeview.com/quicklinks/fact_imitates_science_fiction.html

    I’d encourage you to consider what happens when you use information from a market to determine fundamentals in that market. Given what you wrote, your natural reaction should be to short Obama, which lowers his price. And if Obama’s price is even lower, people should be even more likely to think that he’s doing something awful, causing more sales and an even lower price.

  • g

    Robin, why consider only Clinton and Obama? (One possible reason: the rumour comes from a Pajamas Media site, and they’re less likely to report anything dangerous to the Republicans :-).)

  • Nate

    “I follow politics fairly closely and this 12.6% seems low to me based on the information I have.”

    The fact that a market is trading at prices you don’t agree with doesn’t seem to be a good test for the efficiency of a market.

  • I think this is an interesting falsification test. It rests on three premises:

    1) There actually is a large number of insiders with access to the information. This is not necessarily true–Rosenbaum’s source could be playing him, trying to get him to release a rumor that really hasn’t been packaged by the LA Times.

    2) The insiders believe the information. This seems to be the weakest premise–Rosenbaum himself states that he does not believe it. And so even though he knows the story, he is not going to bet money on it and create an observable change in the prediction markets. Perhaps all the other insiders who are aware of the story are similarly skeptical. And while the fact that a major newspaper is ready to run with it lends credibility to the story, I don’t think that’s enough to stake money on. Major newspapers have run scandal stories before that have turned out to be false.

    3) Given that the insiders do know the story and believe it, they further need to know how to profit from the information. This is also a difficult premise, not simply because they are likely to be unfamiliar with Intrade and IEM, but also because they have to savvy enough to arrange a short-selling of the candidate. Unless if you arrange to borrow shares from another investor, you essentially have to buy a bundle of shares (one for each candidate), sell off the overvalued asset, and then wait on the all the undervalued assets. And so its not an easy, one-time transaction nor can you make a decision about the value of one candidate without being force to evaluate the others.

    I am not convinced that enough investors to tip the market can make through all three.

  • Michael Sullivan

    I am not convinced that enough investors to tip the market can make through all three.

    It seems to me that the last barrier is trivial if the first two are satisfied by a significant number of insiders. What are the chances that those insiders won’t talk? If “everybody” really knows the story, then they learned about this “story” from their insider friends, and most of them probably have outsider friends. Some of those outsider friends will know about Intrade/IEM, or know somebody who trades there.

    It only takes one true believer with a lot of money to move the market in some of these cases, that’s actually a *weakness* of markets as a predictor, at least markets as small and inefficient as most of Intrade’s. But it means that you don’t need lots of people to do this, you just need one who’s willing to make a very big bet.

  • We have a winner! And it’s not Obama…


  • Kennedy,

    But the Hillary rumor you link to has not influenced her Intrade prices.

  • Rob123

    You said “Intrade gives Clinton a 71.3% chance of winning the Presidential election and a 47.7% chance of becoming president.”

    I think you may have meant “Intrade gives Clinton a 71.3% chance of winning the Democratic nomination . . .”?

  • Rob123,

    You are of course correct.