Senator Flávio Bolsonaro finished ahead of President Luiz Inácio Lula da Silva in the opening round of Brazil’s 2026 presidential election, an ordering that several late national polls had not projected and that the prediction market Polymarket had priced for weeks.
The split between survey averages and contract prices recalls the period before the 2024 US presidential election, when conventional polling and event markets also pointed in different directions before the votes were counted.
Official figures released by the Superior Electoral Court place Bolsonaro, the Liberal Party nominee and son of former president Jair Bolsonaro, at 47.03 percent of valid votes, more than 56 million ballots.
Lula, seeking another term for the Workers’ Party, received 45.16 percent, about 53.8 million votes.
Neither candidate reached the absolute majority required to win outright, so the race advances to a runoff scheduled for October 25.
The remaining share of the electorate was divided among smaller tickets that will not proceed.
In the final stretch before election day, a number of registered surveys still showed Lula level with or slightly ahead of Bolsonaro in the first round.
Datafolha’s closing reading, among others, put the incumbent near 42 percent and the senator near 38 percent, a gap that sat inside or near the margin of error once the full list of candidates was taken into account.
Simulations of a head-to-head second round were widely described as technical ties.
Those numbers left open the possibility that Lula would lead the crowded first ballot even if a two-candidate race remained close.
Polymarket’s contracts on the eventual winner described a different ranking. On that platform, participants buy and sell positions that pay out if a stated outcome occurs, and the trading price is read as an implied probability.
Those prices crossed in Bolsonaro’s favor around mid-September and remained there through the close of the campaign.
By late September and the first days of October, the senator’s contract commonly traded in the high 50s to low 60s, against roughly 40 percent for Lula, on a book that had accumulated well over $150 million in volume.
After the first-round count confirmed Bolsonaro’s plurality, the same market moved further in his direction for the presidency as a whole.The pattern is familiar from the United States in November 2024.
National polls in the closing weeks generally framed the contest between Donald Trump and Kamala Harris as tight, and some averages assigned Harris a narrow edge.
Polymarket and the regulated exchange Kalshi both priced Trump ahead by roughly 10 percentage points.
Trump carried the principal battleground states, won the Electoral College, and took nearly half the popular vote.
Commentators afterward argued that markets pooling paid stakes can absorb information that telephone and online samples sometimes miss, including differential turnout and late movement, while noting that the platforms are not immune to thin liquidity or concentrated positions.
Brazilian authorities have restricted domestic access to political prediction contracts, so the Polymarket prices largely reflect trading from outside the country.
Even so, the figures circulated widely during the campaign, and allies of Bolsonaro treated them as a signal of momentum distinct from the survey averages.
The first-round result does not settle the presidency.
It does add another case in which market-implied probabilities ranked the leading candidates differently from several widely cited polls, and in which that ranking matched the order of the official count. The runoff on October 25 will supply the next comparison between the two methods.