Lula's upset in Brazil prompts debate over prediction market accuracy, though experts urge caution

Brazilian President Luiz Inácio Lula da Silva won the weekend's national runoff against Flávio Bolsonaro, becoming the first candidate since 1989 to take the presidency after finishing second in the opening round. Bolsonaro had led the first round on 4 October with roughly 47% of the vote, and that result pushed Lula's odds on Polymarket down to 13.5%.

Lula's price recovered by a few points on the eve of the vote. According to the report, one trader made a profit of $47,595 by capitalising on the mispricing, while a trader known as "timetowander" lost around $27,100. The outcome may be among the clearest cases of mispriced odds on a prediction market ahead of a major global election.

Experts cautioned against reading too much into it. David Lawant of Anchorage Digital wrote on X that markets repriced to the new reality before the expert consensus did, adding that prediction markets show where consensus stands in real time, while deciding what to do with that information remains the user's job. Three major Brazilian polls, including Datafolha's, had also pointed to a Bolsonaro win.

The article notes earlier cases where markets missed. Andrew Cuomo's odds reached 93% weeks before last year's New York City mayoral primary and stood at 57% when voting closed, yet Zohran Mamdani won decisively. Markets also misjudged primaries in Michigan and Wisconsin, where traditional polls were off as well. In the Netherlands last October, D66 beat Geert Wilders' party despite low odds, and its chances jumped from around 5% to near certainty within minutes of exit polls being released.

Other factors may make Brazil an anomaly. Prediction markets are relatively sparse in the country of more than 200 million people. Political analyst Bruno Centurion of Fatto Inteligência Política said the main public consumption of forecasts is inside financial markets, which also held some wishful thinking after the first round. He correctly predicted Lula's upset.

The debate comes less than two weeks before the US midterms, which Kalshi's Tarek Mansour has called truly the first prediction market election. Commentators such as Eric Wallerstein argued the lesson is that prediction markets are a function of increasingly unreliable polls, while Nic Carter said the markets were meant to give sharps a financial motive to research alternative data.