It’s August 2004 and a development team in the testing organization at Microsoft is slated to deliver a software application--to be used by other Microsoft programmers--in November. The team leader is hearing feedback from the team that the date is achievable and, in fact, there is some thought that the application may be ready early.A researcher in Microsoft’s platform and services division decides to set up a prediction market, a sort of stock market for ideas, in which Microsoft employees can place bets on predictions. He creates a market with six possible bets: that the product will ship before November, in November, in December, in January, in February, or later than February. His bettors include not just the development team, but other developers, program managers from related teams, and programmers who will be “customers” of the product.
Each bettor receives $50 to wager. All six bets were started equally at 16 2/3 cents on the dollar, meaning a bettor only had to wager that amount to win $1 if correct.
What happened?
Within seconds, the pre-November market went to zero and never moved.
The November deadline date went to 1.2 cents in about three minutes.
Nobody believed the product would be completed on time, but nobody had said so until then.
Features were cut. Internal customers got angry. Features were added back in. The market predicted a February completion date.
The application was completed in February.
The use of prediction markets, according to the September 2007 IEEE Spectrum, is growing. Here are 8 take-aways:
1. In the past few years the “wisdom of crowds” technique has really taken off, with at least a dozen start-ups (like Consensus Point and Inkling) competing for business. Best Buy, GE, HP, Nokia and Samsung have all begun using the technique to help predict public reaction to new products, the future price of a commodity, or sales revenue in the next quarter.
2. The number of public markets is growing quickly and includes the ability to predict the popularity of websites, new movies, financial instruments and the success of a music CD. You can guess how many inches of snow will fall in NYC’s Central Park in December, or the outcome of a Senate race.
3. Groups can be remarkably intelligent, and are often smarter than the smartest people in them.
4. Prediction markets in business use real money because they reduce the chance that respondents will push their own agendas.
5. HP hosted the first experimental prediction markets in 1997, finding that the markets were considerably better at predicting the future sales of the company’s printer products than the HP official forecast.
6. Some people don’t know as much as they think they do, or make lousy bets. This is a feature, not a disadvantage of the technique; people with money and expertise will trump the bad betting or ill-informed with additional wagers of their own. A market with experts who were always right would see little action because bettors would win little. (Imagine a poker game where all the cards are face up and everyone is a good player. As soon as the player with the best hand makes a raise, the game ends.)
7. The University of Iowa has created a market to predict the outbreak of influenza which turned out to correctly predict the outbreak level (defined by the Centers for Disease Control) one week in advance 71% of the time.
8. Some corporations reject the use of prediction markets because they can reduce control over outcomes and “threaten the established order.” Programmers, for example, may be more productive if a manager drives for a software release date—despite the fact that a prediction market may indicate that the date is unlikely.
0 Response to "Prediction Markets: 8 Take-Aways"
Post a Comment