Stephen Marcus wanted to predict the success of a new business book his company was producing. So his company set up an online trading market where thousands of people purchased imaginary securities based on how well they thought the book would sell. When more than 85 percent of the market participants predicted that the book would rank among Amazon.com's top 1,000 sellers, Marcus breathed a sigh of relief.
Based on the accuracy of previous prediction market forecasts, the president and co-founder of 35-person Shared Insights likes the chances for the new book, to be titled We Are Smarter Than Me. Standard forecasting tools, such as focus groups, involve much smaller numbers of opinions, while the prediction market aggregates thousands of opinions, notes Marcus, 42, whose Woburn, Massachusetts, company generates annual sales of more than $10 million helping corporations build online communities.
Prediction markets aggregate the opinions of many people by asking them to trade shares or place bets reflecting their answers to questions about future developments. Since prediction markets emerged about 20 years ago, they have usually beat expert forecasters in predicting a wide range of events, including election outcomes, natural gas demand levels and even the weather.
Now businesses are using them to foretell sales and profits, identify the most appealing product features, anticipate likely project completion dates and perform other prognostications. Eli Lilly, General Electric, Google, Microsoft, Pfizer and Yahoo! have all used prediction markets.
Small companies have been slower to jump on the bandwagon, notes Robin Hanson, a George Mason University professor who has studied prediction markets for nearly two decades. However, he adds, there is no reason why they should stay away. Groups of as few as a dozen traders can constitute an effective market, and traders can include suppliers, vendors and other outsiders as well as employees. The monetary and time costs of prediction markets are also moderate, especially compared to the high-quality market research they can sometimes replace.
Small companies and large ones have similar questions about future sales, ideal feature configurations and product development completion dates, says Hanson, who even suggests companies elect to fire or retain their current CEOs based on prediction markets. "You could ask about any major corporate decision and how it would affect sales or profits," he says.
Prediction markets gain their effectiveness by harnessing the expertise of many individuals instead of relying on only one or two experts. The idea is that lots of people have a little or sometimes a lot of information about the likely outcome of a situation or event, so a decision that taps more of that knowledge will be a better one. Experience with prediction markets in a wide variety of arenas has shown that to be true in general, but it has also revealed some problems.
For one thing, the idea that tallying bets is better than asking an expert is hard for many people to grasp. Employees need to be trained in why prediction markets work and how to trade in the markets. They also need to be given time to do the trading; however, David Perry, president of Nashville-based prediction market software seller Consensus Point, says that spending just five minutes each on three or so trading sessions per week should do it.
Cost is also a concern. Consensus Point charges $30,000 to $50,000 for its Foresight Server prediction market software. The company will also operate turnkey markets for clients for as little as $500 per month. Hanson says in the beginning, he created prediction markets using paper and pencil--a solution that would also work for cash-strapped entrepreneurs who only want to use prediction markets occasionally. But Perry still recommends that prediction markets be used to answer high-value questions such as sales forecasts, ideal product features and future demand levels.
Entrepreneurs also have to decide whether they will have traders use real money or play money. Hanson prefers real money because it motivates traders to do their best. Most markets, however, use play money and encourage participation by awarding prizes. Hanson recommends that companies using play money include the employees' prediction market performance in annual evaluations and reviews. Perry, however, says motivation isn't a problem because trading is fun. "The reason these work so well in generating good predictive behavior is that people actually want to participate," he says.
Some prediction markets have run into controversy, notably the Pentagon's efforts to set up a market intended to forecast political developments in the Middle East. But Hanson and others think that is only a blip on the shining future of prediction markets. "Over the long term," he says, "it's golden."