A New Electoral Prediction Market

On October 2, a federal appeals court lifted a temporary stay and allowed the regulated prediction market Kalshi to list contracts referencing election outcomes. It didn’t take long for the exchange to do so—a market for contracts that reference the winner of the presidency was launched this morning.

The contract is very carefully worded. The referenced outcome is the party affiliation of the person inaugurated for the term beginning January 20, as verified by the office of the presidency. The market will therefore remain open until at least that date, and if there is significant volume and volatility long after the election has been called (as was the case in 2020) we will know that turbulent times lie ahead.

Close to $400,000 worth of contracts were traded within ten hours of opening, with prices confined to a very narrow range:

This forecast is roughly in line with those of other markets—the race is dead even on Polymarket and tilts towards Harris on PredictIt. Meanwhile, three major statistical models seem to have converged, all placing the likelihood of a Harris victory in the 54-56 percent range.

I’ve been following this case closely, not least because the CFTC and the court have both mentioned a paper of mine with David Rothschild from about a decade ago. In its latest ruling, the court states:

Ensuring the integrity of elections and avoiding improper interference and misinformation are undoubtedly paramount public interests, and a substantiated risk of distorting the electoral process would amount to irreparable harm. The problem is that the Commission has given this court no concrete basis to conclude that event contracts would likely be a vehicle for such harms. The Commission cites to only one example where a trader on an unregulated (and now suspended) exchange placed large bids on Mitt Romney to win the 2012 presidential election… It is “conceivable” the trader did so to “manipulate beliefs about the odds of victory in an attempt to influence fundraising, campaign morale, voter preferences, and turnout.” But the trader fell short because the attempted manipulation was easily detected by market investors… And the Commission’s speculation about that trader’s “conceivable” motivations is, at bottom, an “unsubstantiated and speculative” theory.

There are several quotes here from my paper with David, which Judge Millett seems to have read closely and interpreted correctly.

Not everyone was pleased with the court’s decision. Here is a response to the lifting of the stay by the legal director of Better Markets:

The court’s order will allow betting on an incredibly close and contentious race just weeks before the election. That makes this a sad and ominous day for election integrity in the United States… Gambling on elections will create powerful new incentives for bad actors to interfere with our elections and sway voters outside of the democratic process. The use of AI, ‘deepfakes’ and social media to manipulate voters and influence election outcomes has already become all too real. Ready access to an election gambling contract such as Kalshi’s will intensify that danger with the promise of quick profits.

The point here is that there is no way to undo the potential damage to the public interest of allowing bets in the final weeks of an election year… we have yet another reason to be concerned about the upcoming elections.

Consider, however, the possibility that the existence of this market might make election interference less rather than more effective, for three reasons.

To begin with, it makes market manipulation harder—one would have to operate on multiple markets simultaneously, or trade so heavily on one that arbitrage across markets brings the others into alignment. The court cannot shut down crypto-based Polymarket or UK-based Betfair, it can only determine whether or not they are to have a regulated US competitor.

Second, attempts to manipulate the beliefs of the public through the release of fake, biased, or partisan polls will be less effective in the presence of this market. Such polls will move prices less than those from highly rated pollsters with strong track records. They may move poll aggregates that don’t downweight or filter them out, but the divergence between such aggregates and market prices will then be diagnostic of poor poll quality. Similar effects arise with the selective and strategic release of information by campaigns, or the pronouncements of partisan pundits and prognosticators. If markets ignore them it will be easier for the general public to identify cheap talk that lacks credibility.

Third, and most importantly, the most significant threats to our democracy come not from AI, deepfakes, and social media manipulation, but from a source that has little to do with modern technology—the unwillingness to concede an election, congratulate the legitimate victor, facilitate the transfer of power, and abide by democratic norms. These are the concerns that have brought about the remarkable image of Liz Cheney and Kamala Harris side-by-side on a campaign stage.

Prediction markets neither amplify nor mitigate these threats. But they do give us a glimpse into information that might not otherwise make it into our echo-chambers or penetrate our filter bubbles.

One of the biggest challenges we face as a society is intractable disagreement about basic statements of fact. As I noted in an earlier post:

Prediction markets can serve as a modest antidote to this, by inflicting the cold, hard punishment of monetary loss on those who insist on believing and betting on falsehoods… Let those who absorb and disseminate fake news put money where their mouths are, and let them be publicly and punitively corrected. They may not change their minds, but at least their ability to change the minds of others will be diminished.

I don’t mean to suggest that the decision of the court in lifting the stay is highly consequential for the health of our democracy. It is not. But I do think that the “paramount public interests” mentioned by Judge Millett in her ruling are better served by allowing this market to operate than by shutting it down.