Monday, November 26, 2007

Prediction Markets Appear in the Economist - The Future of Futurology

We're just catching up on our reading after taking a few hours off this past Thanksgiving weekend and saw prediction markets mentioned in an article in the Economist.

From the article:

"The most heeded futurists these days are not individuals, but prediction markets, where the informed guesswork of many is consolidated into hard probability. Will Osama bin Laden be caught in 2008? Only a 15% chance, said Newsfutures in mid-October 2007. Would Iran have nuclear weapons by January 1st 2008? Only a 6.6% chance, said Inkling Markets. Will George Bush pardon Lewis “Scooter” Libby? A better-than-40% chance, said Intrade. There may even be a prediction market somewhere taking bets on immortality. But beware: long- and short-sellers alike will find it hard to collect."

Read the whole thing here. It's a pretty good read.

(thanks newsfutures and midas oracle for the reference)

Monday, November 19, 2007

Could a Political Campaign Use Prediction Markets?

There are several prediction marketplaces out there for the upcoming election season and probably more to come. All the ones we know of are intended for participation by the general public. But what if a Presidential campaign ran an internal marketplace? How could prediction markets be used to give a campaign a competitive advantage?

We put our political operative hats on for a few minutes and came up with some scenarios:

Resource Allocations
Speaking to veterans of previous presidential campaigns, one of the biggest issues mentioned was building consensus internally on resource allocation across the primaries, then for the general election. Conflicting polling data and infighting among advisors often led to the abandonment of several states where post-mortem analysis of actual voting patterns showed the candidate would have had a chance.

Using prediction markets as input to resource allocation decisions, questions could be asked that compare performance metrics across different states, i.e. levels of support among certain voter blocs, predicted endorsements, outcomes of local elections that could impact the general election, etc. This type of information is hard to gather through traditional polling mechanisms but could easily be captured across participants from individual states, locales, and the general campaign.

Fundraising Forecasts
We assume existing forecasting methods used by campaigns are fairly accurate at anticipating how much money will be raised on a quarterly basis from a defined donor list. What may not be as well defined, however, is the impact of various campaign maneuvers on donation levels.

For example, a campaign could internally test various scenarios with national campaign staff, field workers, even undecided voters to see if certain activities drive increased fundraising. If the campaign goes through with the activity, the campaign could evaluate the market and pay it out. If not, the market could simply be refunded.

Of course a campaign could also use prediction markets as further input to official forecasts across the different fundraising channels, allowing a more diverse group of people who may have additional insight beyond the “MBA types” at campaign headquarters crunching numbers.

Risk Management
(Using Inkling,) questions in a prediction market could be generated by the national campaign and staff at the local level. This "web of questions" would be especially useful when trying to anticipate risks to the campaign. The prediction market could be a clearing house of the whispers, rumors, and self-perceived weaknesses of the campaign to continuously test their merit or impact on the campaign.

For example, someone from the local staff may be aware of a negative perception the candidate suffers from in a particular voting district. They could run a prediction market about its impacts in an upcoming primary, i.e. “Will the candidate be perceived as weak on X in analysis of post-appearance local media coverage?” If the stock price remains low, that issue probably doesn’t need to be dealt with specifically. If it's high, it may be an issue the campaign chooses to address proactively ahead of the predicted negative coverage.

Policy Predictions
Given the interest shown to any major candidate, a prediction market gives a campaign an outlet for those supporters wishing to participate in a more meaningful way than simply donating money. A market geared towards public policy across a wide range of issues, both national and local, would be an excellent resource to send people to. Currently most candidate's online presence is focused largely on networking, information dissemination, event notifications, and fundraising. A broadly available prediction market would allow people to provide input on what they think will happen from a policy perspective, i.e. will a particular bill pass? How much funding will an initiative receive, etc.? The campaign could then take these predictions as input to shape policy.

A similar marketplace could also be set up with a more limited audience of dedicated national and local campaign staff. This marketplace could be augmented with policy experts from around the world to provide additional perspective.

Competitive Analysis
Intended for a tightly controlled group of trusted participants, prediction markets could be run on the performance of the other candidates related to their fundraising levels, endorsements, primary performance, etc. to see how one candidate compares to another. This information would be very useful for strategy formulation.

Four years ago some of the candidates tapped the blogosphere to drive early campaign participation and fundraising success. This year most candidates are trying to build up and leverage their online social networks a la Facebook. Will this campaign season also be the year we see a candidate tapping the collective wisdom of his/her sprawling campaign apparatus?