How to predict the future

Betting to Improve the Odds (New York Times)

In my opinion, one of the greatest developments to come out of the Internet is the ability to effectively harness the knowledge of a wide variety of people (also known as "wisdom of the crowds") for projects such as Wikipedia. With Wikipedia, the goal is to organize and share information, but the wisdom of the crowds can be used for many other purposes.

This New York Times article discusses "prediction markets," in which individuals can essentially place bets on the outcome of a particular event (such as "Will Hillary Clinton be the Democratic nominee for President?"). Placing bets ("trades") in a prediction market is kind of like stock trading, except that the "share price" stays between 0 and 100 and reflects the current probability (in the minds of the traders) that the event will occur. In the Hillary Clinton example, her market at Intrade is currently trading at 15.4, meaning that traders believe her current probability of clinching the nomination to be 15.4%. If a trader thinks the likelihood is greater than 15.4%, they can buy shares (hoping the market will go up), and if it does they will make money. Eventually, all markets resolve at 100 (if the event happens) or 0 (if it does not happen), but in the meantime the share price is determined purely by trading activity.

I'm a big fan of prediction markets because the traders have an incentive (i.e. making money) to research the most likely outcome of events, and thus as long as the markets are being actively traded, the markets should demonstrate the probability of an outcome occurring much more accurately than polling, individual research studies, or any other predictive tool. Furthermore, those individuals with insider information and/or expert knowledge about these events have the greatest incentive to participate in prediction markets, which should further increase the accuracy of the markets.

The Times article describes how prediction markets are now being used internally by corporations as a way to improve their forecasting, such as predicting whether a new product will launch on time. As long as there are incentives (such as money or prizes) for predicting outcomes correctly, the "wisdom of the crowds" should indeed shine through in the market activity.

Legitimate concerns do exist about prediction markets, such as the possibility that someone could profit from participating in a market in which they can determine the outcome, or other types of market manipulation. As well, there are markets that could be considered tasteless, such as the Pentagon's failed Policy Analysis Market (in which bets could be placed on the likelihood of terrorist attacks) or the Celebrity Dead Pool (in which bets are placed on who will die this year).

Regardless, I have no doubt that prediction markets are here to stay, and will increasingly be used to tap our collective wisdom. And as long as participation increases, incentives remain, and systems are put in place to prevent fraud, prediction markets should only become more and more accurate.