Tuesday, September 2, 2008

Political Prediction Markets

I came across a site called intrade on my general web surfing the other day. Intrade is a prediction market. You register for the site and can actually 'bet' money on the outcome of certain events.

There is a long history of prediction markets for all sorts of things from sports to Hollywood movies - 'long' in the Internet sense of the word, which is 'short' in historical terms. Wikipedia has a page giving some good background details.

I navigated my way to intrade as it was mentioned on a political news site. On intrade, McCain is currently siting at about a 40% chance to win the White House while Obama is in the low 60% range. This was interesting as poll after poll puts them in a dead heat.

I looked into the intrade system a bit and it looks fine (I am no expert here but at least I understand it - there are probably some pretty smart people behind it). You buy and sell 'contracts' with other traders and the price of a contract varies between $0 and $10. Each 'contract' has an unambiguous binary outcome and is ultimately worth (at the conclusion of the event) either $0 for it not happening or $10 for it happening.

So if you buy Obama contracts at $6.10 and he wins the election, you get a payout of $10 - $6.10 = $3.90 (minus a commission - finally a Web 2.0 site with a business model!). If he loses the election you lose all your money as your contracts are worth $0.

This is the Obama chart on intrade:

So why does Obama look like a shoe-in on intrade but a lame duck in the polls? Is Obama mania getting into the heads of intrade traders? Do they long for change? Need hope? Feel higher taxes on the rich is the solution to their poor lot in life as traders? Likely none of these.

On the surface it's tempting to equate the prediction market to polling, but it's really very different. The intrade numbers aren't saying Obama is going to win in a landslide 60%/40%, all they are saying is he is most likely to beat McCain - margin unspecified (although you would think there would be a correlation between the strength of the prediction and the ultimate margin - we just don't know what that is).

So intrade traders think, given the current polls and events, Obama is still more likely to pull it off. However, if you look at the Republican v Democrat leanings (a vote that indicates preference for a party rather than an individual), support for a 'generic Democrat' is strong.

Or in other words, McCain is neck-and-neck with Obama despite strong support for a Democratic ticket, an unpopular president from the same party, an unpopular war and an economic downturn.

You would think a logical trader trading in presidential picks would give McCain better odds considering what he as overcome to be even at this late stage. Of course balancing this is Obama's huge war chest - money for political advertising - that will be unleashed in the coming weeks. Obama media saturation here we come.

Although even with that war chest, I don't think I would give Obama much over a 50% chance. He still seems over priced.

Ultimately though, no one really knows who is going to win. Prediction markets for political outcomes are just a stab in the dark as there is no set of logical sequences or historical precedents that point to one outcome or another. There is just a whole lot of future uncertainty.

It's like trying to predict the price of oil. Demand and supply can be forecast somewhat accurately, but Israel bombing Iran's nuclear facilities with no UN backing can not.

I'd like to see the political prediction markets on intrade react when Obama reveals he is the illegitimate child of a certain elderly Arizona senator. It could be true...

Digg this
Sphere: Related Content

2 comments:

Anonymous said...

I don't get it - you get $3.90 if he wins and 0 if he loses and it costs you $10? Something's missing there...

Anonymous said...

sorry ignore