Prediction Markets may Revolutionise Business Forecasting

Prediction Markets may Revolutionise Business Forecasting

Prediction markets are a type of financial market (also known as information markets or event futures), in which participants trade in contracts whose payoffs depend on unknown future events. Of course, no prediction is ever perfect—there will always be some uncertainty involved. But a number of successes in predicting presidential elections and events within firms have generated enthusiasm. This mechanism exhibits a surprisingly high level of accuracy; in fact, prediction markets are known to outperform expert panels and large-scale polling organisations, as long as beneficial conditions are present.

Some of the big players, such as Google and Microsoft, are known to use prediction markets—not to mention the Hollywood Stock Exchange (HSX). Google’s employees bid on the likelihood of events affecting the company, such as launch dates, and Google then uses this information to plan and forecast. It has proved to be a highly accurate way to do forecasting — the consensus price on the prediction market is a very good approximation of the probability that a given event occurs. Aggregating information and knowledge of its workers helps Google make decisions based on future data.

Hewlett Packard has also been pioneering this method in sales forecasting and now uses prediction markets in several business units. It has launched a commercial product, BRAIN (Behaviorally Robust Aggregation of Information Networks), which was piloted by the European telecom company Swisscom and yielded impressive results. BRAIN produced forecasts that were, on average, “27 pct. closer to the actual result than the polling group and 17 pct. closer to the actual result than the top five experts.”

Even in smaller samples, such as districts in  elections, there is a very high degree of accuracy. This is obtained by people buying low and selling high and being rewarded for improving the market prediction. Meanwhile, those who buy high and sell low are punished for degrading the market prediction. As such, all participants in prediction markets have a strong incentive to provide their honest, best estimate and—as importantly—to research and collect information on the prediction in question.

The main drawbacks of prediction markets arise in contexts that are largely irrelevant to business forecasting. Prediction markets do seem to display some of the deviations from rationality that appear in other financial markets. The well-known ‘favourite long-shot bias’ appears in prediction markets in the form of the so-called ‘volatility smile’ in options, involving overpricing of strongly out-of-the-money options and underpricing of strongly in-the-money options. These results suggest that prediction markets may be poor at predicting small probability events.

A further possible limitation of prediction markets arises if speculative bubbles drive prices away from likely outcomes. This scope is more limited, however, seeing as prediction markets do not impose restrictions on short selling and tend to be small-scale, such that it is unlikely that informed investors would be capital-constrained. As long as the prediction market is of a decent size and (unbiased) information is readily available, prediction markets are probably the optimal method of forecasting.

The wisdom of crowds has been used extensively on the web already, on Wikipedia, on the HSX and in Twitter-based algorithmic trading. For businesses, it is an obvious step to take: harness insider knowledge and induce incentivized interest and information aggregation in their most prominent stakeholders, the employees.

This will mean that forecasting and the future of the business will be based on the most accurate, unbiased probability weighting available. Arguably, prediction markets should only be used as a supplement to standard forecasting. But this could form part of a shift in business away from basing decision-making on expertise and intuition toward prioritising scientific and data-driven methods.

Categories: Finance, International

About Author

Mikala Sorenson

Mikala Sorensen is an Economist with regional expertise in Europe. She holds a first class honours degree in Philosophy, Politics and Economics from the University of York and a Masters in Economics from the University of Copenhagen. Having interned at the Danish OECD-delegation in Paris and currently working at the Danish Ministry of Finance, she specialises in politics and macroeconomics. Analysis for GRI is an expression of her own views.