AFOS Analytics
Real-time political pricing
Prediction markets × polls × news
no smoothed averages
BrazilThe first week of the second round widened the gap in the winner contract from 68.20pp at Monday's close to 78.50pp at Friday's, with Flávio Bolsonaro at 89.00% and Lula at 10.50%. That is +10.30pp in five trading sessions, the third largest weekly move in absolute value among the 26 weeks with at least four paired trading sessions since 14 April, and the second largest among those that opened in favour of Flávio Bolsonaro. The high of 78.50pp, touched at 19:00 UTC on Friday, is the highest point in the entire paired series, 5.90pp above the previous record, of 72.60pp on the night of the first-round count.
The week did not move in a straight line. On Monday the gap opened at 61.40pp at 07:30 UTC, with Lula at 19.50%, and closed the day at 68.20pp, below Sunday's 72.60pp. Tuesday and Wednesday stayed between 68.80pp and 70.00pp. The widening came on Thursday and Friday, +3.40pp and +6.30pp, the same days on which the first four national polls conducted after the first-round vote were released. AFOS measures both with each one's timestamp and does not attribute the price move to the polls.
The four measured different things. Datafolha (49% to 45%), PoderData/Aya (49% to 44%) and AtlasIntel (51.1% to 45.7%) put Flávio Bolsonaro ahead by 4 to 5.4 points in total votes. Vox Brasil put Lula ahead, by 44.2% to 42.7%, within the margin. The voters of Augusto Cury, who declared his support for Flávio Bolsonaro on 8 October, went to him by 56% to 27% in PoderData/Aya and to Lula by 50.4% to 38.9% in AtlasIntel.
The contract that took in the most new money was the one that lost the most price: USD 3.70M on Lula, against USD 2.09M on Flávio Bolsonaro, in a book that took in USD 5.78M over the week and reached USD 179.00M accumulated. On Saturday, 10 October, after the window, Ipespe measured 49% to 44% for Flávio Bolsonaro, and the week ahead brings ten national polls registered for 13 to 15 October and the first second-round debate, on 15 October.
1.Executive Summary
The week's theme is the first reading of the second round by the two instruments. The market widened the gap over two days, Thursday and Friday, and closed the week with Flávio Bolsonaro above 89%. The polls, published on the same two days, did not give a single distance: three of them measured between 4 and 5.4 points for him and the fourth measured 1.5 point for Lula. The contract today pays more than eight times as much for Flávio Bolsonaro as for Lula, and the pollsters measure a race of a few points, with one of them within the margin for the other side. Price and polls answer different questions, the probability of winning on 25 October and the week's voting intention, and this issue measures the two side by side.
2.Why AFOS does not smooth
The four national polls released on 8 and 9 October measured the second round in total votes, with samples of 2,100 to 5,026 interviews and fieldwork between 3 and 8 October. A simple average of each one's lead returns 3.2 points for Flávio Bolsonaro, a number that reads as a clear, moderate lead.
Why it matters: the average of 3.2 points describes a race no pollster found and erases the only one that measured Lula ahead. The interval shows what the week offered the reader: three pollsters close to one another and one outside the group, and the same slice of voters, Augusto Cury's, going in opposite directions in two of them. If the next round of polling closes the interval, the divergence between the pollsters was first-week noise; if it holds, the divergence is the data.
Issue №20 assigned about 55% to the base case, in which the first second-round national polls would repeat the split of the eve of the vote, with the distance within 4 points either way, and the contract would stay between 75% and 90% for Flávio Bolsonaro with a smaller range than in the week of the vote. The contract behaved as expected: Flávio Bolsonaro's five closes stayed between 83.70% and 89.00%, with an intraday low of 80.90%, and his range over the week was 8.10pp. The polls, only in part: the pollsters split, but two of the four, PoderData/Aya and AtlasIntel, measured more than 4 points for him. About 30% went to the contrarian case, in which the first four national polls would put Lula ahead outside the margin. It did not happen: one of the four put Lula ahead, and within the margin. About 15% went to the tail, an institutional act that would change the race. It did not happen during the week; the acts on record, Alexandre de Moraes's ruling on Eduardo Bolsonaro's sentence and the TSE's individual rulings on Lula campaign videos, did not reach the candidacy of either of the two.
Erratum to issue №20. That issue's watch list said that Augusto Cury, Renan Santos and Ronaldo Caiado added up to 7.31% of valid votes and the other names 0.50%. Counted from the votes in the official tally, they are 7.32% and 0.49%. The total for the eliminated candidates, 7.81%, does not change.
3.Weighted scenarios for the week
The week of 12 to 16 October has ten national polls registered with the TSE for release between 13 and 15 October, among them Quaest, AtlasIntel, Datafolha and Palver, and the first second-round debate, on 15 October, on SBT, RedeTV! and CNN Brasil. Record cancelled the one scheduled for 11 October. On Saturday, 10 October, Ipespe measured 49% to 44% for Flávio Bolsonaro, the fourth pollster to put him ahead by 4 points or more.
The national polls of 13 to 15 October repeat the shape of the first week: most put Flávio Bolsonaro ahead by 4 to 6 points, and one or two pollsters measure the race within the margin. The contract stays between 85% and 92% for him, and the gap oscillates around the 70pp to 80pp band without repeating Friday's 6.30pp jump.
Quaest, AtlasIntel and Datafolha, the three with fieldwork after the endorsements of 8 October, measure the distance falling to within the margin, as Vox Brasil did. In that case the contract, which pays more than eight times as much for Flávio Bolsonaro as for Lula, stands against the reading of three reference pollsters in the same week, and the gap gives back part of what it opened.
The 15 October debate, the first in which the two face each other in this election, or an institutional act concerning one of the two candidates, changes the race before the next round of polling. The Supreme Court impeachment contract, at 3.40%, is the closest gauge of institutional risk on the dashboard, and spent the week between 2.00% and 8.50%.
4.Indicator Grid
| Contract | Current | Δ week | Vol USD acc. | Implied reading |
|---|---|---|---|---|
| Presidential gap, Flávio minus Lula | +78.50pp | +10.30pp | USD 179.00M | Path: +68.20, +70.00, +68.80, +72.20, +78.50. Low of 61.40pp at 07:30 UTC on Monday and high of 78.50pp at 19:00 UTC on Friday, the largest in the paired series. Third largest weekly Δ in absolute value out of 26 weeks. |
| Flávio Bolsonaro, presidential | 89.00% | +5.30pp | USD 19.09M | Low of 80.90% at 07:30 UTC on Monday and high at Friday's close. Took in USD 2.09M over the week, and the reading normalized by the pair's sum gives +5.07pp. |
| Lula, presidential | 10.50% | -5.00pp | USD 21.75M | High of 19.50% at 07:30 UTC on Monday and low at Friday's close. Took in USD 3.70M, more than Flávio Bolsonaro, and still has the largest accumulated volume in the book. |
| Supreme Court impeachment before 2027 | 3.40% | +0.90pp | USD 0.10M | High of 8.50% at 03:30 UTC on Monday and low of 2.00% at 08:00 UTC on Tuesday, with USD 2,147 of new money over the week. A thin book, where a few trades move the price. |
| 2026 inflation, 5.00% to 5.49% | 47.20% | -2.90pp | USD 25 thousand | Still the leading band. Ranged between 41.20% and 56.20% in the week's readings, with USD 1,054 of new money. |
| 2026 inflation, 4.50% to 4.99% | 35.00% | +5.20pp | USD 31 thousand | Low of 10.80% at 01:30 UTC on Monday and high of 36.80% at 18:00 UTC on Friday. The band just below the leader gained weight over the week. |
| 2026 inflation, 5.50% to 5.99% | 12.20% | -4.50pp | USD 20 thousand | High of 31.70% at 12:30 UTC and low of 11.00% at 20:30 UTC, both on Friday: 20.70pp in a single day, with USD 1,753 of new money over the whole week. |
5.Liquidity and market structure
The presidential book took in USD 5.78M of new money between Monday's and Friday's closes, and all of it went to the two contracts still open: USD 3.70M on Lula and USD 2.09M on Flávio Bolsonaro. The contract that took in the most is the one that lost the most price, from 15.50% to 10.50%. Volume counts trades in both directions, buying and selling, and does not tell anyone's direction: the week in which Lula took in more money than his opponent was the week in which his price fell 5 points. The other seventeen contracts in the book have been closed since the first round, resolved at zero, and their volume is history.
Volume is accumulated since each contract opened. The week's new money is the difference between the volumes recorded at the closes of 5 and 9 October, and the prices in the table are those of Friday's close, at 21:30 UTC. AFOS does not cite book depth in editorial text, because low liquidity in a continuously arbitraged market does not mean the price is wrong.
6.Calendar of price-relevant prints
| Date | Sample | Why it matters | |
|---|---|---|---|
| 13/Oct | PoderData | n=3,000 | On 8 October it measured 49% to 44% for Flávio Bolsonaro in total votes. Fieldwork from 10 to 12 October. |
| 14/Oct | Quaest | n=2,154 | The pollster's first second-round national poll. On the eve of the first round it measured 44% to 42% for Flávio Bolsonaro. |
| 15/Oct | AtlasIntel | n=5,000 | On 9 October it measured 51.1% to 45.7%, the largest distance among the week's national polls, and Lula ahead among Augusto Cury voters. |
| 15/Oct | Datafolha | n=2,520 | On 8 October it measured 49% to 45%, at the edge of the margin of 2 points. Fieldwork from 13 to 15 October, ending on the day of the first debate. |
| 15/Oct | Palver | n=5,000 | The pollster's first national poll after the first round. On the eve of the vote it measured 49% to 44% for Flávio Bolsonaro in the second round. |
Registration with the electoral court is not confirmed release. These are five of the ten national polls with samples of 1,000 or more and release scheduled between 13 and 15 October; the others are two from Gerp, Nexus, 100 Cidades and Indexa. DataTrends has release scheduled for 11 October.
7.Watch list, week triggers
- Ten national polls in three days, 13 to 15 October It is the first time in the second round that Quaest and Palver publish, and AtlasIntel and Datafolha return with fieldwork after the endorsements of 8 October. The first week's interval ran from 1.5 point for Lula to 5.4 points for Flávio Bolsonaro.
- The first debate, on 15 October SBT, RedeTV! and CNN Brasil host the first encounter between the two candidates in this election. Band and Globo have debates on 18 and 23 October.
- Augusto Cury's voters Three measurements: 56% to 27% for Flávio Bolsonaro in PoderData/Aya, 50% to 29% in Ipespe and 50.4% to 38.9% for Lula in AtlasIntel. AtlasIntel returns on 15 October with a sample of 5,000.
- The pair at more than eight times With the other contracts closed, the sum of Flávio Bolsonaro and Lula stayed between 99.00% and 99.80% at the week's closes, and the move on one side is almost all of the move on the other: Friday's normalized reading gave 5.07pp against 5.30pp and 5.00pp raw.
- The Supreme Court The impeachment contract spent the week between 2.00% and 8.50% with USD 2,147 of new money. A move of price in a thin book, not of flow.
8.Methodology
AFOS does not produce a weighted average and does not arbitrate between sources. It publishes the three signals side by side, contract price, registered poll and news flow, and reports the distance between them when it exists. A market price goes live only after two independent readings eight minutes apart that agree within 0.20pp, contract by contract, and the reference series are measured in the daily backup of the database and not in the public interface. The week runs from Monday to Friday: the published value is Friday's close, the path is the five daily closes, the Δ is Friday minus Monday, and the gap is measured at paired instants, both sides with the same timestamp.
The polls cited are those registered with the Superior Electoral Court, with protocol, sample, margin and field dates declared, and in the second round AFOS compares the price with each pollster's total vote, not with valid votes. The code and the method are open under the Apache 2.0 licence. Every factual claim in this issue links to its primary source.
9.Additional reading · macro coverage
Complementary macro context. AFOS is a primary source on divergence between market and polls; the links below cover the economic and institutional backdrop of the week.
- TSE, official results · Official first-round tally, by office and by state., resultados.tse.jus.br/oficial/app/index.html
- Valor Econômico · Daily coverage of the 2026 election and of the currency and interest-rate reaction to the second round.(paywall), valor.globo.com/politica/eleicoes-2026/
- Reuters Brazil · English-language flow on politics and currency, useful for the non-resident investor., www.reuters.com/world/americas/
- Poder360 · Registry of electoral polls with protocol and history by pollster., www.poder360.com.br/pesquisas/
Get it by email
AFOS Tradeoff is the weekly read on Brazilian political risk, for people who decide on data rather than headlines.
No spam. Unsubscribe anytime.
After signing up you choose whether to receive it in English, Portuguese or Spanish.