1. Prediction market

The gap widens by 14.25 points. At the confirmed reading of October 24 at 20:06 BRT, the winner contract pays 83.75% for Flávio Bolsonaro (USD 27.06M) and 16.50% for Lula (USD 30.28M). Against the confirmed reading of October 23 at 19:10 BRT, Flávio Bolsonaro rises 7.25pp and Lula falls 7.00pp; since the pair's sum went from 100.00 to 100.25, the normalized move is 7.04pp for each side. The gap goes from 53.00pp to 67.25pp. At the confirmed reading of October 24 at 17:21 BRT, before the evening polls, Flávio Bolsonaro already stood at 83.70%; between 17:21 and 20:06, with the six national polls released, his price moved 0.05pp. This reading does not attribute the move to the debate or to the polls. The total volume accumulated in the presidential book is USD 195.79M, and the largest volume in the book is still Lula's.

Supreme Court and inflation. The STF justice impeachment by 2027 contract stands at 2.75% (USD 0.10M) at the confirmed reading of October 24 at 17:21 BRT, 0.60pp below the October 23 reading. In the 2026 inflation contract, at the 19:58 BRT reading against the October 23 reading at 19:16 BRT, the 5.00% to 5.49% band goes from 72.00% to 62.90% (USD 30 thousand), the 4.50% to 4.99% band from 11.60% to 20.95% (USD 34 thousand) and the 5.50% to 5.99% band from 13.45% to 10.85% (USD 22 thousand).

2. What the polling firms registered

Six national polls on the eve. Datafolha, with 4,050 in-person interviews from October 23 to 24, gives 48% to 47% in total votes and 51% to 49% in valid votes. Quaest, with 3,702 interviews in the same period, gives 46% to 45% and 51% to 49%. AtlasIntel, with 4,990 online interviews from October 18 to 23, gives 49.9% to 48.4% and 50.8% to 49.2%. Palver, with 5,000 interviews from October 22 to 24, gives 50% to 49% in total votes. BTG/Nexus, with 2,000 interviews, gives 47% to 46% and 50.5% to 49.5%. Gerp, with 2,400 telephone interviews, gives 49% to 46% and 52% to 48%.

Wave against wave. Against each pollster's previous wave, Lula rises in total votes in all six, by 0.1 to 1 point. The distance falls in AtlasIntel, from 2.2 to 1.5 points, in Palver, from 2 to 1, and in Gerp, from 5 to 3; it stays at 1 point in Datafolha and Quaest; and it rises from 0.6 to 1 point in BTG/Nexus.

What else they measured. In Datafolha, 46% reject Lula and 45% Flávio Bolsonaro; in Quaest, Flávio Bolsonaro's rejection goes from 52% to 49% and Lula's stays at 47%. Among voters with a high propensity to vote, 76% of the electorate, Datafolha measures 52% to 48% for Flávio Bolsonaro in valid votes, and among those with a low or medium propensity, 52% to 48% for Lula, according to InfoMoney.

The count of national polls. Of the fifty-two conducted after the first round, forty-seven put Flávio Bolsonaro ahead, four put Lula ahead in total votes and one measured a tie in valid votes.

Polling calendar, next 7 days

Polls registered with the TSE scheduled for publication between 25/Oct and 25/Oct. Appearing in the table does not mean publication is confirmed, because pollsters can delay or cancel a release. Filters applied: national scope and sample of 1,000 or more. Each registration links to the TSE public record.

Date	Pollster	Sample	Scope	TSE registration	Conf.
25/Oct	PoderData	4,000	national	BR-02828/2026	0.7

Source: public TSE record via the AFOS API. Samples of 3,000 or more in bold. The table lists 1 national poll in the window; 2 records from the same period were left out by scope or by sample. The status is registered, not published.

3. What the press covered

The debate. The last debate, on Globo, was on the night of October 23, with proposals and an exchange of accusations, according to Agência Brasil and the BBC. In Datafolha, 54% said they watched it, in full or in part; 36% thought Lula did better, another 36% pointed to Flávio Bolsonaro, and 98% said the debate did not change their vote, according to Folha de S.Paulo and O Globo. The fact-checking agency Lupa flagged 20 statements it considered inadequate and 11 entirely true, according to Folha de S.Paulo.

Willer Tomaz. On October 23, Justice Alexandre de Moraes lifted the secrecy of the offshore Alcazar Holdings, linked to the lawyer Willer Tomaz, and ordered up to R$ 6 billion of his assets frozen, according to Gazeta do Povo and O Globo. Flávio Bolsonaro's defense says the relationship between the two is limited to a friendship between the families.

Secrecy waivers. On October 24, the Electoral Prosecutor's Office argued for dismissing the case on opening Lula's and Flávio Bolsonaro's bank, tax and phone records, on the view that it is not for the Electoral Court to decide on their disclosure, according to Folha de S.Paulo and G1.

TSE. In a decision reported on October 24, the court barred Nikolas Ferreira and Pablo Marçal from publishing false content until the election, under a fine of R$ 100 thousand, according to Folha de S.Paulo, and ordered Meta and X to submit reports every six hours, according to Valor. TSE president Kassio Nunes Marques denied the transport of Indigenous voters with a simple prior notice, and on October 24 Prosecutor-General Paulo Gonet asked the STF to suspend that decision, according to O Globo and Estadão.

4. Divergences of the day

Market × polls: the winner contract pays 83.75% for Flávio Bolsonaro, and the six national polls on the eve put him ahead by 1 to 3 points in total votes. The price measures the probability of winning on October 25, and the poll measures voting intention during each one's fieldwork.

Poll × poll: both released on the same October 24, Colectta puts Lula ahead by 46.5% to 46% in total votes, and Gerp puts Flávio Bolsonaro ahead by 49% to 46%.

Debate × vote: in Datafolha, Lula and Flávio Bolsonaro tie at 36% on who did better, and 98% say the debate did not change their vote.

In summary

The six national polls on the eve put Flávio Bolsonaro ahead by 1 to 3 points in total votes, and in all of them Lula rises against the pollster's previous wave.
The winner contract takes Flávio Bolsonaro from 76.50% to 83.75%, and the gap goes from 53.00pp to 67.25pp.
In Datafolha, 98% say Globo's debate did not change their vote.

Fontes consultadas

matérias com link direto para a notícia (veículos âncora):

matérias secundárias (URL Google News redirect, clique resolve à matéria):

Fontes técnicas: Polymarket (quotes via the AFOS proxy, confirmed reading of 24/Oct 20:06 BRT and reading of total volume and the inflation book at 19:58 BRT), TSE registry (official electoral polls).

October 24, 2026