Prediction market boosters often tout the platforms as barometers of public opinion, or even “truth machines” offering a clearer picture of where things are headed than polling or other data. In 2024, for example, Kalshi beat many pollsters in forecasting the outcome of the US presidential race. But their trading activity can also tell more perplexing stories.

Two global election markets on Polymarket’s offshore platform, for instance, generated over $600 million in trading volume, far more than most political markets typically attract. Most of the activity took place buying and selling contracts tied to long-shot candidates who ultimately did not win.

This summer, Ethiopian prime minister Abiy Ahmed’s Prosperity Party cruised to victory in federal elections, winning 438 parliamentary seats and paving the way for him to serve another five-year term. The closest competitor won 13 seats. On prediction markets, trading volume—the total value of contracts bought and sold—typically tracks roughly with the odds of a given outcome; frontrunners attract far more volume than long shots. But in this case, the trading patterns deviated sharply from the norm. Just $173,000 was wagered on Ahmed winning, compared to more than $41 million traded on his chief of staff, Shimelis Abdisa, becoming the country’s next leader. Altogether, the volume traded added up to over $285 million, nearly all on losing candidates.

Even odder? “Nearly all of it—about $284 million—came from pairs of accounts repeatedly buying and selling with each other,” says former Polymarket trader Chris Park, who now runs the prediction-market forensic analysis firm MSR Decode.

Such a high volume of seemingly pointless back-and-forth transactions has raised eyebrows among experts. “It’s obviously some form of volume manipulation,” says Rajiv Sethi, a Barnard College economics professor who researches prediction markets.

When asked about these markets, Shana Bautista, Polymarket’s head of global intelligence and investigations, told WIRED that the team “immediately flagged this activity, investigated, and blocked the accounts from trading.” Polymarket did not respond to follow-up questions about why the accounts were able to trade such a high amount of volume before it took action.

The timing of these trades made things even weirder. Polymarket’s rules stipulated that the Ethiopia market would be resolved in October after the prime minister was formally appointed and sworn in, which meant that it stayed open for several months after Ahmed’s victory was all but a foregone conclusion. For 123 days after the election, bursts of seemingly irrational trading continued, prompting Polymarket users to leave comments like “THIS NOT RESOLVED?” and “fuckkkk.”