Brazil’s presidential runoff on Sunday delivered a sharp contrast between real-time trading and prediction market expectations, as Luiz Inácio Lula da Silva defeated Flávio Bolsonaro to secure a fourth term. The election triggered a dramatic surge in EWZ perpetual contracts, with hourly volume jumping from around $1 million to $52.8 million as results rolled in, according to Pyth Network. Between 19:00 and 00:00 UTC on election night, traders executed roughly $218 million in EWZ trades, and Pyth’s pricing data accounted for 99% of those transactions.
The frenzy underscored how geopolitical events can rapidly amplify trading volumes and reliance on real-time data. Yet while EWZ perps were pricing in the unfolding result, Polymarket was caught badly offside. Going into election day, the platform put Lula’s chances at only about 20%, with Bolsonaro near 80%. Even as results began arriving, one pseudonymous whale, FhantomBets, said the market was slow to reprice. He argued that retail traders were so opposed to Lula that they could be sold unlimited size at 60 to 80 cents when Lula had essentially already won.
Others pushed back on that explanation. Luke Nolan, senior researcher at CoinShares, said the more likely culprit was Brazil’s ban on prediction market platforms introduced in April, citing investor protection and gambling addiction concerns. Finance Minister Dario Durigan said the platforms violated betting regulations and lacked proper oversight. Nolan noted that without local participants, prices were set mostly by outsiders. Einar Knobel, CEO of TxODDS, agreed, adding that the Brazilian market attracted only 4.5% of the volume seen in the 2024 U.S. presidential election, leaving too little local knowledge to correct the mispricing.
Nolan cautioned against declaring prediction markets broken, pointing out that an outcome priced at 80% should still fail about 20% of the time across enough events. He said prediction markets continue to price better than traditional betting platforms, citing Betfair giving Hillary Clinton an 83% chance the day before the 2016 U.S. election. The Brazil episode, he argued, warrants scrutiny for persistent overconfidence, but not a wholesale dismissal of prediction markets.
