Will Flávio Bolsonaro win the 2026 Brazilian presidential election?

TL;DR

The saved forecast is 64.2% Yes for Flávio Bolsonaro to win the 2026 Brazilian presidential election and 35.8% No. Flávio begins the October 25 runoff with a verified first-round lead and a narrow advantage in most late polling. The result turns on whether that small edge survives turnout changes, voter transfers, and final-week movement.

Context

Brazil’s presidential contest went to a second round after neither finalist won an outright majority in the October 4 first round. Flávio Bolsonaro and Luiz Inácio Lula da Silva are the finalists, with the runoff scheduled for October 25 under the electoral calendar (TSE resolution).

The race is no longer about whether Flávio can reach the final contest. It is about a close two-candidate vote with only days left in the campaign. The market resolves on the candidate who wins, including the second round; a normal official result is expected shortly after voting rather than through the market’s June 2027 fallback.

Evidence

The historical backbone is the completed first round. The near-complete official count put Flávio at 47.03% and Lula at 45.16%, giving Flávio a 1.87-point advantage (TSE). This is the best direct evidence of the electorate’s revealed preference. It also matters more than one poll because it is an observed national vote rather than a sampled estimate.

But the first round is not a runoff result. Roughly 7.8% of valid first-round votes went to other candidates, while abstention, blank and null voting, and differential participation can all change the electorate. A 1.87-point first-round lead is therefore a structural advantage, not a floor beneath Flávio’s runoff vote share.

Late runoff polling provides the second major evidence base. The available readings point in Flávio’s direction overall, but their spread says the contest remains close. The late comparable history in the research is below.

Poll	Fieldwork / release	Published result	Comparable Flávio margin
AtlasIntel/Bloomberg	Oct. 13–18; published Oct. 19	49.8% Flávio, 47.6% Lula, total intent	+2.2 pp valid
BTG/Nexus	Fieldwork ending Oct. 18; release Oct. 19	46.5% Flávio, 45.9% Lula, total intent	about +0.6 pp total
PoderData/Aya	Oct. 16–18; published Oct. 19	52% Flávio, 48% Lula, valid votes	+4.0 pp valid
Veritá	Fieldwork through Oct. 16	51% Flávio, 49% Lula, valid votes	+2.0 pp valid
Futura	Fieldwork ending around Oct. 17	50% Flávio, 50% Lula, valid votes	0.0 pp
Datafolha	Oct. 13–14; published Oct. 15	47% Flávio, 45% Lula, total intent	+2.0 pp valid
Ipespe/Abrapel	Oct. 13–15; published Oct. 17	51.2% Flávio, 48.8% Lula, valid votes	+2.4 pp valid
Quaest	Oct. 10–13; published Oct. 14	46% Flávio, 44% Lula, total intent	roughly +2.2 pp among named candidates

The fresh signals are positive but compressed. Atlas’s October 13–18 survey showed Flávio leading 49.8% to 47.6% in total intent, equivalent to 51.1% to 48.9% after excluding its reported blank, null, and undecided share (AtlasIntel reporting). PoderData showed a wider 52%–48% valid-vote advantage, while the Nexus reading was close to even. The supplied polling summary also identifies the newest Atlas, PoderData, and Nexus releases as favorable to Flávio but narrow (Politon Data).

Datafolha offers an in-person benchmark. Its October 13–14 survey found 47% for Flávio and 45% for Lula in total intentions, or 51%–49% among valid votes, and indicated that Flávio’s earlier 52%–48% edge had narrowed (Datafolha). Ipespe/Abrapel similarly reported a 51.2%–48.8% valid-vote result (CNN Brasil). Another late survey, Vox, reportedly put Lula ahead. That disagreement is why the evidence supports a moderate favorite rather than a commanding one.

The polling record also contains a useful prior: pre-first-round surveys appear to have understated Flávio relative to the eventual count. A comparison cited in the research found that Datafolha’s final first-round reading undercounted Flávio by 5.03 points and Atlas by 2.93 points (ViésLab). The data shows a meaningful polling miss. I read it as support for treating Flávio’s small late lead as more than a coin flip, but not as grounds to add past error mechanically to every runoff poll. The electorate and question have changed, and firms can revise their methods after a miss.

Voter transfers give Flávio a further structural edge. Romeu Zema, Ronaldo Caiado, and Augusto Cury endorsed him, while Renan Santos did not endorse either finalist (Itatiaia). That matters because eliminated candidates’ voters are the principal pool available to alter the first-round balance. Yet endorsements are imperfect signals of where voters actually go. Atlas crosstabs reported that 69.8% of Caiado voters moved to Flávio and 24.5% to Lula, but Lula led among Cury voters, 52.1% to 42.4% (AtlasIntel reporting).

Turnout is Lula’s clearest path. First-round abstention was high: the research cites 33.4 million abstainers, compared with Flávio’s raw-vote lead of about 2.2 million (Datafolha). AP described more than 33 million nonvoters, roughly 21% of the electorate and the highest first-round abstention rate since 1998 (AP). These people will not all return, so their existence does not itself establish a Lula advantage. But Atlas found that first-round abstainers who said they would vote in the runoff favored Lula 55.7% to 40.2% (AtlasIntel reporting). That is enough to make mobilization a decisive variable.

The remaining campaign calendar is short. The candidates held a two-hour televised debate on October 18, described as hostile and light on policy rather than as an obvious race-changing event (AP). The short runway limits the scope for a long trend, but it does not eliminate a late shock, a turnout shift, or a polling miss.

What's non-obvious

The obvious reading is that a 1–2 point polling lead is nearly a toss-up. The evidence is stronger than that for Flávio because the polls sit on top of an official 1.87-point first-round lead, favorable endorsements, and a recent record of polls understating him. Those facts do not make the election safe. They do mean the small survey edge should not be read in isolation.

The equally tempting opposite error is to treat the first round as a runoff preview with no meaningful change possible. That misses the size and composition of abstention, plus the fact that eliminated-candidate voters are not transferred automatically by elite endorsements. Lula does not need a wholesale realignment to win. A modest turnout advantage among previously absent voters, combined with better transfers from selected eliminated candidates, can reverse a narrow first-round deficit.

Uncertainties

The forecast remains 64.2% Yes and 35.8% No because the evidence favors Flávio but leaves substantial room for Lula. The principal uncertainty is correlated polling error. Surveys use different modes and turnout assumptions, but they can still share the same error in weighting hard-to-reach voters, estimating participation, and measuring late decisions. Nominal sample size cannot resolve those common weaknesses.

The trend is also uncertain. Atlas and Datafolha both showed narrowing from earlier Flávio advantages, and one late poll reportedly gave Lula the lead. It is not known from the evidence here whether that reflects a real late Lula gain, normal poll-to-poll variation, or differences in survey method. A genuine final-week movement toward Lula would make the saved forecast too favorable to Flávio; a continued undercount of Flávio similar to the first round would make it too cautious.

Three information gaps would most reduce uncertainty:

• More final-wave surveys with transparent fieldwork dates, methods, turnout models, and full undecided treatment.
• Verified evidence on whether first-round abstainers actually return for the runoff, rather than only their stated preference if they vote.
• Better cross-tabs on transfers from each eliminated candidate’s electorate, including blank and null voting.

There is no information in the supplied materials establishing that an extraordinary legal, health, or procedural event will disrupt the ordinary runoff. The evidence instead describes an imminent second-round vote and a normal official process. The live risk is electoral, not procedural: a very close election in which a modest Flávio advantage can hold, or be overtaken, on turnout and late choice.