Until recently at Hewlett-Packard, a group of managers would sit through unending meetings to predict future prices of key computer supplies – information the company would then use to plan its purchasing strategy. Today, a prediction market at HP accomplishes the same task, but with better accuracy in far less time.

A prediction market is a stock market for ideas, where people can buy and sell “stock” in different outcomes, which impacts their relative prices. Between 1868 and 1940, prediction markets forecasted elections before advanced polling. They made a comeback in 1988, when the University of Iowa started the first electronic prediction market focused on economic and political events, such as elections.

The idea behind prediction markets is that the aggregated knowledge of diverse individuals produces a prediction with greater accuracy than a small number of experts. Diversity is critical here – if everyone starts out with the same knowledge, predictions will be the same and betting will therefore be negligible.

Though they are not infallible, prediction markets are strikingly accurate. In the 2006 Senate elections, for example, no public opinion poll predicted all 33 races correctly, but bettors at Tradesports did. Today, people use prediction markets to forecast everything from the popularity of movies to influenza outbreaks. In the corporate world, ArcelorMittal, Best Buy, General Electric, HP, Nokia, and Samsung have been using prediction markets to divine public reaction to new products, next quarter’s sales revenues, whether products will be ready on time, and future commodity prices. A comprehensive study at Intel concluded that prediction markets are at least as accurate as forecasts by Intel’s management, and often as much as 20% better.

Small companies have been slower to embrace prediction markets, but Entrepreneur.com points out that there is no reason they should miss out on the benefits. Markets can work well with relatively small groups of traders, and a small employee base can be supplemented with suppliers, vendors, and other outsiders. Small companies often seek the same answers as big corporations and can save precious time and money by replacing complicated market research with prediction markets when appropriate.

Despite all their benefits, prediction markets are only appropriate in cases where predictions are separable into neat categories: Will Microsoft release its software before November, in November, in December, in January, in February, or later than February? Will Amazon.com sell more books, iPods, or 500-thread-count sheets next month? This condition limits the prediction market’s usefulness when companies want to aggregate opinions without influencing people with a preconceived framework of possible responses or when nuanced answers would be more appropriate.

MixedInk sees prediction market software as a kindred tool. MixedInk steps in where prediction markets leave off by allowing people to respond to open-ended questions and provide more complicated answers. In addition to asking which of five products will be most popular using a prediction market, a company could use MixedInk to ask its employees to help conceive of its next innovative product or design a marketing strategy.

Using democratic, meritocratic aggregation tools like prediction markets and MixedInk, companies can truly benefit from the vast knowledge base of their employees quickly and easily. Of course, deciding to use these tools requires leaders to recognize employees’ wisdom, which can mean a big shift for some companies…but that’s a subject for another post.

This entry was posted on Tuesday, October 23rd, 2007 at 6:34 am and is filed under PredictionMarkets, collaboration, politics, prediction markets, software. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.