Prediction Markets Outperform Traditional Polls in Brazilian Election, Forecasting Bolsonaro Advance

Flávio Bolsonaro captured 47.03% of votes in Brazil's first round against incumbent Lula's 45.16%, contradicting traditional polls that favored the incumbent and triggering a second round runoff on October 25. Prediction market platforms like Polymarket had accurately reflected Bolsonaro's rising chances since September, while conventional surveys remained bullish on Lula. The result reignites debate over prediction market reliability even as Brazil maintains legal restrictions on such trading platforms.
Prediction markets tracked a significant shift in Brazilian electoral sentiment beginning in September. While traditional polling organizations like Datafolha maintained Lula's advantage through the campaign period, betting platforms reflected growing support for the challenger. The gap between methodologies became especially pronounced in the final weeks before voting, suggesting market participants detected momentum shifts that surveys did not capture.
However, a critical structural limitation undermines direct comparisons between the two forecasting approaches. Brazilian financial regulators prohibited political derivatives trading in May, subsequently blocking international prediction platforms and restricting existing Brazilian accounts. This regulatory barrier means the market signals emerged primarily from international participants rather than the Brazilian electorate itself, potentially introducing skews unrelated to domestic voter sentiment.
The Brazilian election outcome may intensify global debate about prediction market credibility versus conventional polling as election forecasting tools. If markets continue outperforming polls across multiple democracies, policymakers could face pressure to reconsider restrictions on political betting platforms, despite concerns about market manipulation and speculation. Conversely, the role of non-domestic participants in shaping these markets raises questions about whose preferences they truly represent, complicating their use as reliable public sentiment indicators.