Gambling on Democracy

Intrade's irrational underbelly.

By Dara Lind,  November 11, 2008

If Longfellow had spent his “Rainy Day” surfing Internet forums, he might have written, “Into each board a little troll must fall.” Lately, the forums on Intrade.com, the most prominent online prediction market for American politics, have been no exception. By Election Day, some users were tying Obama to everyone from the Black Panthers to the Illuminati, while others disgustedly asked, “When will this forum again be readable?” “It would be nice if we could all try to steer the forum a bit back toward trading-related topics,” another wrote, while admitting: “I say this as someone who has gotten a bit partisan on here at times.” When the price of a prediction (an Intrade “contract”) for John McCain’s victory rose, Barack Obama’s supporters charged market manipulation by “rogue traders” — allegations the site verified in at least one high-profile case. When it fell, McCain buyers devised increasingly elaborate arguments against conventional wisdom, relying on theories of media and polling bias that occasionally shaded into outright conspiracy.

On discussion boards with a tangential relationship to politics, this sort of message-board sniping would be standard operating procedure. But during the 2008 election, Intrade became a political powerhouse. The site enjoyed a monumental growth spurt both in size — trading volume has increased 700% since 2006 — and in stature, not least because of its sterling reputation among both pundits and economists for the dispassionate accuracy of its markets’ predictions.

Like all markets, Intrade works by allowing individuals to share information. By aggregating what each member knows, Intrade employs more information than any individual trader could handle. The Intrade crowd trading in its election markets was massive — 3 million contracts were traded solely on the markets for McCain’s and Obama’s victories, and activity was so heavy in the days before the election that Intrade set up a separate site for VIP traders in case the main site crashed. (It didn’t.) The success was uncanny: sure, traders predicted (that is, bought at a higher value) that Indiana would go for McCain and Missouri for Obama, but with 11 electoral votes apiece they canceled each other out in the site’s aggregate prediction.

At this birds-eye view, Intrade could be a diagram out of an economics textbook. Markets work. Case closed. And it's true: Intrade's markets consistently provide rational results. But what about its users? Are they models of rationality too? Perhaps not. Despite the accurate predictions it generates, the wisdom of crowds relies on behavior that is often irrational — and certainly far more complex and personal than the data points that make up a day’s trading might suggest.

Intrade gets lots of attention for its political forecasting, but for the site's many traders, the crowd-sourced predictions are a sideshow. They're only relevant insofar as it means that they make money. Traders turn a profit by buying contracts for an outcome that eventually happens, or short-selling on one that doesn’t. The day after the final presidential debate, for example, Obama was trading at 84.6 — that is to say, an 84.6% chance to win the election. A buyer might take a block of 200 contracts (typical for a single trade) at a cost of $169.20 plus fees, eventually earning it back with a $30.80 profit once the contracts “settled” at 100 after John McCain’s concession. McCain’s contracts settled at 0, leaving any trader who bought them at a loss.

As with the activity of any futures market, this is basically gambling: a prediction with money behind it is a bet. But this time around, American participants on Intrade forums weren’t particularly concerned with the legal ramifications of betting on their own elections. The moral ramifications were something else again. After all, Americans see little wrongdoing when a baseball fan bets for his team to win. But the last manager to do it — Pete Rose — was banned from the game for life. When traders are both setting the prices and making the profits — not to mention voting for the eventual outcome and living under the victor’s administration — the line between spectator and participant becomes far blurrier.

Congressional Quarterly reported in October that the site was investigating an “institutional member” who was singlehandedly driving up McCain’s trading value; one observer interpreted it as an attempt to boost morale, saying, “It’s obviously someone who wants good news for McCain.” None of the forum’s posters endorsed this strategy; the “honest market” itself, in the words of one poster, was inviolable. But their commitment to the accuracy of Intrade’s predictions rarely extended to their own investments — at least when those predictions went against their own desires.

To the end, some McCain supporters argued that Intrade users had been hoodwinked by the mainstream media and polls. Two days before the election, one user started a thread called “How John McCain Won the 2008 Presidential Election” and included state-by-state expected electoral vote totals. Another announced that while he’d been able to turn a small profit on Obama, he sold his contracts on him the week before the election and invested again in McCain. He admitted he’d lose his shirt, but “it just doesn't sit right with me to have any financial gain if such a destructive candidate like Barack Obama wins the election,” he confessed. Most forum users weren’t this vocal, but all seemed to agree that traders couldn't reach across the aisle; one user crowed after the election that it's "an absolute pleasure taking money from conservatives!"

An economist might, under duress, call this sort of fringe benefit "value added", but it doesn't really fit with any economist's notion of a "rational actor." But who decided that once a man — much less an American voter — had some money in his hand, he'd become rational, anyway? Maybe it's wiser to think of an Intrade investment as a stake in the market's outcome — much like a vote in the presidential election the market predicted. By this theory, Intrade's election invisible-hand jujitsu simply reflected the economic interests of people who already had an emotional interest in one candidate or the other, and of course there happened to be more of them willing to make bets than there were of their opponents.

But that's only half the story — the weird alchemy of rational calculation of self-interest and political loyalty that went into a trader's decision compounded his desire to see his contract settle at 100. The scope of the election may have made it easier to notice the deluded losers and sore winners that make Intrade deals, but that doesn't mean they only come out in November. Loyalty to a particular outcome, emotional and financial, is the engine driving the predictive market — and what makes it attractive to the critical mass of users it needs to keep its crowdsourcing reliable to begin with.

Now, the drama's ended — nationally as well as on the forums, which since the election have been conciliatory and quiet. Throughout the site, some attention is being paid to unresolved races and Cabinet appointments, but activity has slowed down considerably. For the moment, Intrade is fueled neither by endless number-crunching, profit-maximizing speculation nor by the “rogue” tactics of the impassioned. Markets may already be up for the 2010 and 2012 elections, but the crowd — wise or otherwise — has dispersed.

Dara Lind is a contributing editor for Culture11.