Polymarket’s Ethiopia election market saw more than $41 million traded on a candidate who did not become prime minister, while just $173,000 was wagered on the eventual winner. The pattern forms part of more than $600 million in combined volume across the platform’s Ethiopia and Romania election markets, where trading continued after major electoral events.
The activity raises questions about how long election contracts should remain open and whether headline volume reflects genuine interest. Much of the trading involved long-shot or losing candidates, according to WIRED’s investigation.
Trading continued after the results
Ethiopia’s Prosperity Party won 438 parliamentary seats in the summer election, compared with 13 for its closest competitor, giving Prime Minister Abiy Ahmed a path to another five-year term. Yet the market recorded more than $41 million in trading on Ahmed’s chief of staff, Shimelis Abdisa, becoming the next leader.
The Ethiopian market’s total volume exceeded $285 million, nearly all of it tied to losing candidates. MSR Decode, a prediction-market analysis firm, estimated that about $284 million came from pairs of accounts repeatedly trading with one another, WIRED reported.
Polymarket’s rules tied resolution to Ahmed’s formal appointment and oath of office, so the market remained open for 123 days after the vote. Polymarket’s head of global intelligence and investigations, Shana Bautista, said the company flagged the activity, investigated and blocked the accounts. The company did not answer follow-up questions about why they could trade so much before the intervention.
Romania’s 2024 presidential election process was annulled amid concerns about foreign interference. Polymarket kept the original election market open while a separate market ran for the subsequent vote, creating another period in which contracts linked to a disrupted process remained available.
Volume is not proof of market demand
Trading volume often rises with the perceived odds of an outcome, with frontrunners typically drawing more activity than long shots. But repeated buying and selling can inflate volume without representing comparable levels of distinct capital or genuine interest.
Rajiv Sethi, a Barnard College economics professor who studies prediction markets, and colleagues developed an algorithm that flagged nearly all of the Romanian market’s volume as suspicious; during certain weeks, it classified 99.6% as wash trading. That practice, which creates the appearance of heavy activity through repeated trades, is prohibited by Polymarket’s terms and banned on regulated US financial exchanges.
MSR Decode examined more than 38,000 post-election trade loops in the Ethiopian market. More than 98% left neither side with a gain or loss, and the median time accounts held contracts before reselling them was nine seconds. The firm identified activity farming-artificially increasing recorded volume-as the likeliest explanation, but the analysis did not establish who controlled the accounts or their intent.
Oversight questions remain
Sethi has described the Romanian activity as more dramatic than the Ethiopian case, but cautioned that apparent wash trading is not always wash trading. An earlier study by Sethi and colleagues also found evidence of widespread wash trading on Polymarket’s offshore platform.
Polymarket has raised more than $1 billion at an estimated $21 billion valuation and reported annualized revenue above $1 billion earlier in the summer. Volume is a prominent measure of platform activity, making the distinction between real trading and repetitive turnover consequential. The CFTC declined to say whether it is examining these election trades; it has previously investigated suspected manipulation on Polymarket, generally involving possible insider trading. The debate over prediction-market safeguards and contract controls now includes the question of how platforms monitor markets after the underlying event has effectively been decided.
