As the election season kicks into full gear, I've become obsessed again with the concept of prediction markets. Wikipedia offers perhaps the most simple explanation for what a prediction market is. To wit, "prediction markets are speculative markets created for the purpose of making predictions. Assets are created whose final cash value is tied to a particular event (e.g., will the next US president be a Republican) or parameter (e.g., total sales next quarter). The current market prices can then be interpreted as predictions of the probability of the event or the expected value of the parameter. Prediction markets are thus structured as betting exchanges, without any risk for the bookmaker … People who buy low and sell high are rewarded for improving the market prediction, while those who buy high and sell low are punished for degrading the market prediction."
Another way of looking at these markets is that they reflect perceived probability at a given moment. Some types of prediction markets are structured so that one bets in favor of an algorithm that simulates the other side of a transaction (which is becoming more common in "thin" prediction markets, according to a friend). As I write this post, Obama has a lead in the Iowa polls, yet he is behind in the more liquid and heavily capitalized intrade.com market (this is the market that predicted the 2004 election with a correlation coefficient per state of .87)
But the purpose of this post is not to head down a path that offers a best "market-driven" guess at the Presidential election. Rather, I’d like to introduce the concept of prediction markets into the spend management world. After digging into prediction markets for some time -- and after watching a close friend and colleague work with them quite successfully on a number of projects -- I’ve come to believe that they offer tremendous promise to help procurement and operations organizations predict and manage through uncertainty.
Just as they can help predict our next President-elect, they can also help companies predict the probability of a specific event in a defined time period. But unlike nearly all other forms of polling, they have an uncanny way to capture the individual opinions of a group of participants, having proven time and time again to be more accurate than other surveying tools. In addition, perhaps their largest potential benefit is time. When working properly, a prediction can tell you within a few hours -- much less days or weeks -- of the probability of an event.
What would the best applications of prediction markets be in the Spend Management world? I reckon they could work equally well for both macro-level topics (e.g., chance of a supplier failure, currency movements, supplier quality, supply disruptions, etc.) and internal issues (e.g., chance of a delay for an e-Procurement implementation, chance of hitting supplier diversity targets or savings goals, etc.). The key is having enough liquidity (i.e., participants) in the market to provide an adequate gauge of voter/trader sentiment on each topic. A few dozen participants or less, in some cases, might be sufficient. In other cases, even hundreds can be wrong. In general, they are better than other methods at predicting the future but are far from prescient.
If anyone is interested in applying prediction market theory to Spend Management, I’d be happy to put you in touch with my above-linked colleague, Art Hutchinson, who has done quite a bit of work on the topic and would be more than happy to have a dialogue. He’s also found and negotiated with providers of prediction market software that have some surprisingly low rates. When I spoke to Art about this post, he pointed out to me that the real costs are organizational -- getting the right people to participate, getting them to pay attention over an extended period of time, etc. And perhaps most important, getting them to act on the information they discover.
- Jason Busch