Ethiopia’s ruling Prosperity Party won 438 parliamentary seats in this summer’s federal election. Its closest competitor won 13. Prime Minister Abiy Ahmed’s victory was, by any measure, settled. Yet on Polymarket’s offshore prediction market for who would lead Ethiopia next, just $173,000 was wagered on Ahmed, while more than $41 million traded on his chief of staff, Shimelis Abdisa — a man who won nothing. Total volume exceeded $285 million, nearly all of it on losing candidates, and nearly all of it after the outcome was known.
As WIRED’s analysis of the markets showed, the volume was not dispersed across thousands of opinionated traders. ‘Nearly all of it—about $284 million—came from pairs of accounts repeatedly buying and selling with each other,’ says Chris Park, a former Polymarket trader who now runs MSR Decode, a prediction-market forensic analysis firm that examined the market for WIRED. Together with a second market on Romania’s annulled 2024 election, the two generated over $600 million in volume — an order of magnitude beyond what political markets typically attract.
What 38,000 loops look like
MSR Decode’s analysis of the Ethiopian market identified more than 38,000 ‘trade loops’ after the election — instances where individual wallets bought and sold contracts to each other in quick succession. Over 98 percent of the loops ended with neither side gaining or losing money. The median holding time between buying a contract and selling it back was nine seconds. The firm concluded the pattern was highly unlikely to be sophisticated algorithmic traders extracting profits, because there were no profits to extract. The trades were, by construction, pointless — except as volume.
That is the signature of wash trading: accounts trading with themselves or each other to simulate activity while money moves in a circle. It is banned on US-regulated exchanges and prohibited in Polymarket’s own terms of service. Rajiv Sethi, a Barnard College economics professor who studies prediction markets, told WIRED the Ethiopian market was ‘obviously some form of volume manipulation,’ while cautioning that because the wallets are anonymous, ‘what appears to be wash trading isn’t always wash trading.’ An earlier study by Sethi and colleagues found evidence of widespread wash trading across Polymarket’s offshore platform.
Romania was worse. The original 2024 presidential election was annulled over foreign-interference fears and rerun months later; Polymarket kept the market on the voided race open alongside the new one. An algorithm Sethi and his colleagues built to flag wash trading classified 99.6 percent of the old market’s volume in certain weeks as suspicious. Sethi calls markets like these — kept open after the underlying event has already happened — ‘zombie markets.’ During the 123 days the Ethiopia market stayed open after the vote, because Polymarket’s rules tied resolution to the prime minister’s formal swearing-in in October, users left comments reading ‘THIS NOT RESOLVED?’ and ‘fuckkkk.’
Why farm a market nobody can win
The likeliest motive, per MSR Decode, is ‘activity farming’: inflating trading volume in hopes of qualifying for a future cryptocurrency token distribution. In 2025, Polymarket chief marketing officer Matthew Modabber said on a crypto podcast that there ‘would be a token,’ without specifics. Sethi finds the token-allocation hypothesis plausible, and Andre Guettler, director of Ulm University’s Institute of Strategic Management and Finance, told WIRED the trades are ‘textbook examples on how market participants farm a potential airdrop.’ No token exists; the traders were spending fees on the expectation of one. Under that reading, $284 million in circular volume is not evidence of demand for election contracts — it is the cost of a lottery ticket, capital recycled between two wallets the same operator controls.
Whether that distinction matters depends on what volume is used for. Polymarket’s growth is usually measured in volume traded, and the company’s recent numbers are large: a funding round of over $1 billion led by 1789 Capital, the venture firm where Donald Trump Jr. is a partner, at an estimated $21 billion valuation; announced annualized revenue above $1 billion this summer. Political markets are a small slice of that, and two anomalous ones do not reprice the company. But if the reported totals include markets where 98 percent of loops net to zero and a forensic firm attributes nearly all activity to paired wallets, volume is measuring something other than what the word suggests.
The pattern is not confined to these two elections. Barron’s reported earlier this year that much of the activity in Polymarket’s 2028 US presidential markets concentrated on Elon Musk and New York City mayor Zohran Mamdani — both born outside the US, neither eligible to run. Rival Kalshi, meanwhile, faced wash-trading allegations on X over its new crypto perpetual-futures market and published a blog post denying them; Sethi agrees with Kalshi there, arguing the volume pattern fits market-maker incentives that let other traders profit from small price discrepancies. Same chart shape, different mechanism, per his analysis.
The platform and the regulator
Polymarket’s head of global intelligence and investigations, Shana Bautista, told WIRED the team ‘immediately flagged this activity, investigated, and blocked the accounts from trading.’ The company did not answer follow-up questions about how paired accounts cycled through over $284 million before anyone intervened. Polymarket is already fighting on other fronts — New York sued it in September, calling the platform unlicensed gambling; Polymarket sued back, arguing only federal regulators have jurisdiction.
That federal regulator is the Commodity Futures Trading Commission, which declined to tell WIRED whether it is examining these trades. The CFTC has previously investigated suspected manipulation on Polymarket — several cases first reported by WIRED — but those involved potential insider trading, not wash trading. The agency has imposed large fines for wash trading in other markets. ‘It seems to me that the CFTC would want to know about weird activity in a very large market,’ a former CFTC employee said.
Two numbers remain unaccounted for. One is who controlled the wallets: anonymity is why the airdrop-farming reading rests on inference from 38,000 loops rather than on any identified trader. The other is how much volume across the platform matches this pattern — Sethi’s earlier study said widespread, MSR Decode’s here says nearly all of it, market by market. The Ethiopia market resolved this month. The token has not launched.