Predictive Markets
Dean
I wrote yesterday about predictions markets and why they work so well. Well (via The Pamphleteer) I found a somewhat better description of how and why they work right here. Although my less technical explanation still seems to cover it: when you people put even a small wager on a proposition, they're more likely to tell you what they honestly think vs. what they want to believe, and, while the intuition of one person may be suspect, the intuition of hundreds or thousands of people, taken as a gestalt, is usually pretty on the money. At least when it comes to predicting certain events.
The most respected of the predictions markets is the University of Iowa's Iowa Electronic Market. There's also Intrade, which, unlike the non-profit, university-funded Iowa market, is a private business. There's also the Strategypage Market, which asks a surprisingly diverse (and sometimes silly but interesting) set of questions.
Here's how it works: Let's say you want to see if John Kerry will win the election in November. That is what you create a bidding war on: Kerry wins.
The question of whether Bush will win is separate. So if you test whether Bush will win, you create a contract that says "Bush wins." Then people can bid on that.
You can also create contracts on "Nader wins" and "Badnarik wins" and so on.
Standard practice is that a winning contract is worth $100 at closing. So if you think Bush will win, you buy shares in "Bush wins," on the theory that you'll collect $100/share when he wins. But if Bush loses, you get zero. So how much are you willing to gamble on Bush winning?
At the moment, the Iowa market is selling shares in "Bush Wins" at about $62 each. Shares in "Kerry wins" are selling at about $39 each. On Intrade, shares in "Bush wins" are selling at about $69, and Kerry's selling at around $30.
It's important to note that these are not predictions of vote totals. They're just predictions on the "Yes/No" of whether the candidate will win or not.
What those prices indicate at the moment is that both markets are predicting a Bush win. However, these contracts are still trading. It's very possible that something could change to cause people to start selling their shares in Bush and trying to buy shares in Kerry, causing the prices to shift. Trading won't stop until election day.
Mind you, these tools are not used merely to predict elections. If you look at these markets they ask dozens, sometimes even hundreds, of questions. They run bids on sports teams, on economic events, and many other things. Historically, these predictions markets are right far more often than they are wrong.
It appears from all this that human beings in the aggregate are far more intelligent than most elitists want to give them credit for. But then, that's an observation I've made many times.
Related Posts (on one page):
;-)
So the only people who'd know the predictions being tested would be our government officials. Although I'm sure they'd be declassified over time.
Sad, although I think it would helpful. Too bad the PC crowd got what they wanted crybabying their desires into policy.
Actually, Jay, you pointed out one of the very non-PC problems with this idea: as soon as the predictive power of the market became apparent, allowing the government to put preventive measures in place, this would have to be "priced into" the market. In other words, people would no longer be betting just on the likelihood of a terrorist attack, but also on the probability that if the market agrees with their assessment, then the attack would be prevented. Whatever the morality of valuing money over the lives that would be lost in a terrorist attack, the simple fact is that people would be much less willing to risk money if they could be fairly sure that the government would act to prevent the attacks they were predicting. So really, as soon as the market started working, it would also start to lose its effectiveness. This is true even if money were not involved, since prestige or credibility would simply replace dollars as the currency of the market.
Really, the same is true in the real stock market, which is exactly why people argue for minimal government interference: if people are aware of a party with a disproportionate ability to affect the outcomes of their bets, then they have to factor this knowledge into the decision. If this party's behavior is easily predictable given a certain set of conditions, then it almost becomes its own constant in the risk/return equation, particularly if those conditions have to do specifically with the behavior of the market.