A day does not go by without viral stories about prediction markets, whether it’s about regulatory news or speculators getting rich with prediction markets. The hype almost rivals that AI:
Prediction markets saw a sudden surge of popularity during and after Covid. Although Polymarket is often cited as pioneer of prediction markets, the defunct cryptocurrency exchange FTX offered markets on the outcome of the 2020 US presidential election. I foresee prediction markets are here to stay, and will continue to gain popularity as a sort of middle ground between the stock market and sports betting. They complement each other, as there is a sizable user overlap between sports bettors and prediction market users. But prediction markets have the advantage of being much more broad, with the number of possible markets limited only by one’s imagination and a willing counterparty.
There are prediction markets now for seemingly everything, including even ‘mention markets’, where people bet if certain words will be spoken or omitted in a public speech.
There are two notable problems with prediction markets: Obviously, insider trading. Second, the predictions themselves affecting the outcome. If there is a lot of money riding on a certain outcome, someone who is in a position to affect said outcome could deliberately modify his or her behavior to affect the market, realizing a profit. For example, a speechwriter could conceivably plant or omit specific words, knowing in advance exactly which phrases would be spoken, and covertly placing bets accordingly. From the Atlantic, “America Is Slow-Walking Into a Polymarket Disaster:”
…The suspiciously well-timed bets that one Polymarket user placed right before the capture of Nicolás Maduro may have been just a stroke of phenomenal luck that netted a roughly $400,000 payout. Or maybe someone with inside information was looking for easy money. Last week, when White House Press Secretary Karoline Leavitt abruptly ended her briefing after 64 minutes and 30 seconds, many traders were outraged, because they had predicted (with 98 percent odds) that the briefing would run past 65 minutes. Some suspected, with no evidence, that Leavitt had deliberately stopped before the 65-minute mark to turn a profit…
There was renewed media focus on insider trading after $400,000 of profits were suspiciously realized on the night American forces arrested Venezuelan president Nicolás Maduro, prompting speculation someone was in the know. But it’s also interesting how so many people in the comments are dismissive or blithe to insider trading on prediction markets, or even support it. In the Hacker News thread of the aforenoted Atlantic article, user “smeej” in the comments endorses insider trading :
From a moralistic perspective, insider trading is possibly a victimless crime or even beneficial if we assume the purpose of markets is to determine the the value or worth of something (or for prediction markets, a binary outcome), in a process called ‘price discovery’. So insider trading only helps to expedite this process. If companies reported their finances daily or weekly instead of quarterly, this would mean fewer jumps or other discontinuities in the market. The withholding of information creates instability in the market, when this information is suddenly divulged compared to gradually leaked.
So case closed? Not so fast. But on the other hand, for prediction markets to work, in which there is a binary outcome and ‘all or nothing’ payout (in technical parlance, a binary option), requires that participants believe such markets are fair, or that insider trades are too small to move the market. Otherwise, from a game theoretic perspective, the entire thing falls apart.
For example, let’s consider a hypothetical contract in which there is thin volume. An insider, who shortly before the contract’s expiration, buys up all the liquidity on the ‘ask’ side, effectively settling the contract, and realizing a large profit. If this continued long enough, market participants would eventually learn that the optimal strategy is to wait until placing bets, anticipating insider’s arrival.
The insider would subsequently adjust her strategy to wait even longer, repeating, until asymptotically approaching the settlement date, with insiders and outsiders in a sort of standoff. The whole thing fails, as there are no participants, and hence no market. The insider would have to look elsewhere to profit from the information. For highly liquid markets, some insider trading would be undetectable, but for contracts with thin volume, this would be a major problem.
There are a few ways to mitigate this to some extent. For example, limiting trade sizes to prevent insiders from dominating a market. Or offering some sort of bonus or rebate for traders who bet early, which diminishes as the contract nears expiration. If history is any guide, I also predict that the wild west days of prediction markets will come to an end, and like sports betting and the stock market, prediction markets will become highly regulated.