The rise and fall of prediction markets

Yesterday I saw an announcement for Microsoft Prediction Lab which allows users to predict upcoming events such as elections. While prediction markets are nothing new—the Iowa Electronics Market has been operational since 1988 and, over time, has routinely outperformed traditional polls—I have not heard about them as much recently. Google Trends seems to validate the idea that their popularity has waned, with a slow decline in interest, punctuated by the electoral cycle:

On paper prediction markets are appealing because of their flexibility and simplicity. The idea that a group of average persons can, when taken in aggregate, be more predictive than domain experts is attractive as well. For example, the Good Judgment Program is working to predict world events more accurately than the intelligence community and companies continue to experiment with prediction markets for sales forecasts, project timelines, and other distinct events.

It will be interesting to see if Microsoft Prediction Lab gains adoption or fizzles. My sense is that, if prediction markets are to become popularized beyond an academic setting, it will have more to do with larger trends like mobile and gaming than with the market itself. At the least, we can expect another bump in the 2016 US election cycle.