- For predicting societal outcomes of all sorts, collective market intelligence packs a punch greater than any existing supercomputer.
- Recently, prediction markets have made use of collective intelligence to forecast a wide variety of outcomes.
- Prediction markets aren't panaceas, but they're the best forecasting tools we've got.
A fringe organization of astrophysicists called SETI (Search for Extra-Terrestrial Intelligence) was founded in 1984 to detect intelligent life beyond Earth. One SETI approach uses radio telescopes to listen for intelligent radio signals from space. Their directive is simple, but its scope is mind numbingly huge: SETI must sift billions upon billions of signals, tune out the avalanche of static, and locate a proverbial pin dropping ten thousand light years from Earth.
So, innovators that they are, SETI exploited the world's biggest supercomputer—the millions-strong network of idle PCs on desktops from Iowa to Hong Kong. Willing participants download a screensaver that analyzes fragments of information from SETI and sends back the results. In this way, SETI rapidly evaluates mountains of data that would otherwise require a thousand years to search.
Granted, most of us aren't looking for intelligent life in a galaxy far, far away. But SETI's groundbreaking exploitation of unused computing power is analogous to markets. To get this, substitute a network of plodding PCs with a network of human brains, the greatest supercomputers created to date. Then replace the incentive to contact Jar Jar Binks with the incentive to make money, and you're packing an analytical tool of mammoth proportions.
Recently, prediction markets have made use of collective intelligence to forecast a wide variety of social outcomes. Intrade (www.intrade.com), the most well-known of these markets, creates contracts that pay $10 if a given outcome is realized or $0 if it never comes to pass. Participants either buy contracts for events they believe will happen, or sell contracts (like shorting a stock) if they believe the event won't occur. As the contract is traded, the price rises or falls in line with the payoff's probability. If at season's start, a contract the Knicks win the NBA Championship is priced $.01, we can assume all currently available information indicates their chances of winning in June are slim to none. On the other hand, if the Celtics-to-win contract trades at $9.50, it's a sure sign the market believes they're favorites.
Over the years these markets, though not perfectly efficient, have returned remarkably accurate forecasts. However, there are many who don't want to believe human networks are that smart. It offends their hearts that cold, cash-driven analysis could ever trump more "high-minded" alternatives. Naturally they leap at any chance to disprove market theory. In the process, they usually reveal a fundamental misunderstanding of markets and probability instead.
Traders' Calls Just as Bad on Elections
By Mark Gongloff, The Wall Street Journal
This article posits one prediction market's forecast that John McCain has a 38% chance of winning the presidency throws the whole system into doubt—McCain clearly has a better chance than that, right? But a 38% probability of a favorable outcome for McCain is actually pretty good. Keep in mind there are still three candidates, so in reality, a 33% outcome would be an even shot that McCain gets the nod. Additionally, one of his opponents is currently stronger than the other, so markets are likely beginning to discount the three-candidate scenario, skewing results toward a head-to-head matchup (where 50% would be an even shot at the presidency).
The article goes on to worry that prediction markets are thinly traded and prone to bubbles and manias just like other markets. This is often a valid claim. Many contracts at Intrade are very seldom traded. It's important to remember that although market-driven predictions are currently the most accurate forecasting tools we've got, they're in their infancy and have much further to evolve.
Also keep in mind such contracts aren't necessarily revealing the true outcome so much as they're revealing what the market thinks the outcome will be. It so happens the crowd—with its vastly superior aggregate information—is more often right than the individual. But not always. Markets are efficient discounters of all widely known information, but particularly in the short term information interpretation is subject to wide swings in sentiment (hence, bull market corrections and the like).
So, as skeptics keep on doing what they do best, remember this: The most powerful analytical tools efficiently make use of all available computing power. That makes markets' ability to unite brainiacs worldwide one of history's best methods of divination.