Luiz Inácio Lula da Silva’s victory over Flávio Bolsonaro in Brazil has exposed a sharp pre-election mispricing on Polymarket, but analysts say one unexpected result is not enough to show that prediction markets are systemically flawed.
Lula became the first candidate since 1989 to win the presidency after finishing second in the first round. Bolsonaro had taken approximately 47% of the vote on 4 October, pushing Lula’s Polymarket odds down to 13.5%. On the eve of the election, his chances had recovered by a few points.
Some traders profited from the shift: one made $47,595, while a trader using the name “timetowander” lost about $27,100. At the close of voting, prediction markets gave Lula roughly a 20% chance of winning, according to Smarkets CEO Jason Trost.
A price, not a verdict
Trost said the odds should be understood as a real-time price rather than a determination of which outcome is “right”. He compared a 20% chance of a Lula win to a weather forecast giving a 20% chance of rain: the event remains possible, even if it does not happen in most cases. “If the weather report says 20% chance of rain, nobody says the report was ‘wrong,’” he said.
David Lawant of Anchorage Digital argued that markets adjusted to the changing situation before expert consensus did. “Prediction markets show you where consensus is, in real time. What to do with that data, to predict and price the future, is still your job,” he wrote on X.
Brazil’s prediction market activity is relatively limited, despite the country having more than 200 million residents. Political analyst Bruno Centurion of Fatto Inteligência Política, who correctly anticipated Lula’s upset, said public forecasting was concentrated in financial markets and included “some wishful thinking” after the first round. Three major Brazilian polls had forecast a Bolsonaro win, including Datafolha, which projected a two-point margin.
A test ahead of US midterms
Prediction markets gained wider attention during the 2024 US presidential election. Polymarket put Donald Trump’s chances of beating Kamala Harris at about 58%, while total activity at Polymarket and Kalshi exceeded $4.2 billion, according to Pew Research Center. Kalshi’s Tarek Mansour has described the coming US midterms as “truly the first prediction market election”.
There have also been notable misses, including in the New York City mayoral primary, where Andrew Cuomo’s odds reached 93% weeks before voting and 57% as polls closed. Zohran Mamdani won the first round and finished with 56.4% in the final ranked-choice round.
Prediction market activity for the US midterms is widely expected to exceed $1 billion in notional volume. POPVOX co-founder Marci Harris said the Brazilian result showed the limits of trying to make people’s choices predictable: “The people are in charge, and people are wonderfully unpredictable.”
