Two new polls put Flávio Bolsonaro ahead. But a lead in the polls and a near-90% prediction-market probability are not the same thing.
October 10, 2026 | ProbGap Featured Forecast Review
Brazil's presidential election is entering its final two weeks.
After Flávio Bolsonaro finished ahead of incumbent President Luiz Inácio Lula da Silva in the October 4 first round, prediction markets sharply reassessed the race.
Our latest market reference puts Bolsonaro's probability of winning near 89%, although this is a third-party snapshot rather than a confirmed live executable price.
That is a striking level of confidence for an election that still has more than two weeks to run.
Two new polls now provide an opportunity to examine the evidence behind that confidence.
The First Round Changed the Picture
The October 4 results were:
Candidate | First-round vote share |
|---|---|
Flávio Bolsonaro | 47.03% |
Luiz Inácio Lula da Silva | 45.16% |
Other candidates | 7.81% |
Bolsonaro finished ahead by 1.87 percentage points.
The result challenged pre-election polling expectations and prompted an immediate reassessment of the runoff.
But the first-round margin is not itself a forecast of the second-round outcome.
The remaining electorate, turnout differences and campaign developments all matter.
Two New Polls, One Important Distinction
The latest surveys offer two separate measurements of voter intentions.
Datafolha, October 6–8
Bolsonaro: 52% of valid votes
Lula: 48% of valid votes
Sample: 2,520 respondents
AtlasIntel/Bloomberg, October 3–8
Bolsonaro: 52.8% of valid votes
Lula: 47.2% of valid votes
Sample: 5,026 respondents
Both surveys measure voting intentions, not the probability of an election victory.
A candidate polling at 52% does not necessarily have a 52% probability of winning. Translating a polling lead into an outcome probability requires assumptions about sampling uncertainty, systematic polling error, turnout, remaining campaign time and other risks.
That distinction is central to the ProbGap approach.
What the Market Is Actually Pricing
The reported market price near 89% implies a high level of confidence in one outcome.
Such a price may reflect more than the published polls alone: traders can incorporate expectations about campaign developments, voter behavior and information not captured by any individual survey.
But a market price is still an estimate.
It should be assessed against the evidence, not treated as an independent confirmation of that evidence.
The relevant question is not simply whether the polls show a lead.
It is how much uncertainty remains between today's polling and the October 25 result.