AFOS Analytics
Real-time political pricing
Prediction markets × polls × news
no smoothed averages
BrazilThis issue covers seven days rather than five: from Monday, 28 September, to the close of the first-round count on the night of Sunday, 4 October. Closing on Friday would leave out the day on which the presidential contract moved more than in any full week of the series.
The gap between Flávio Bolsonaro and Lula widened from 16.50pp at Monday's close to 72.60pp at 23:30 UTC on Sunday, with Flávio Bolsonaro at 86.10% and Lula at 13.50%. That is 56.10pp in seven days. Sunday's high exceeds by 48.80pp any instant of the paired series recorded since 14 April, whose previous record was 23.80pp, on 21 September. From Monday to Friday the move was small, from 16.50pp to 13.40pp, with a low of 8.30pp at 07:00 UTC on Friday, hours after the cancellation of the TV Globo debate. Almost all of the shift was left for the weekend: 16.00pp on Saturday, the day nine national polls were released, and 43.20pp on Sunday, the day of the vote and the count.
The Superior Electoral Court (TSE) closed the official tally on 5 October with Flávio Bolsonaro 47.03% and Lula 45.16% of valid votes, and the two contest the second round on 25 October. On the eve of the vote, five of the nine national polls put Lula ahead in the first round. The final distance, 1.87 points for Flávio Bolsonaro, fell within the interval of the nine, which ran from 5.1 points for Lula to 4 points for Flávio Bolsonaro.
The side books reached resolution on the same Sunday. The first-round second-place book closed at 99.30% for Lula and the third-place book at 99.40% for Augusto Cury, the same order as the ballot box. The book on the party with the most Senate seats closed at 97.70% for the PL, which won 19 of the 54 seats in contention. The presidential book took in USD 13.41M of new money over the window and reached USD 169.90M accumulated.
1.Executive Summary
The week's theme is the passage from measurement to result. For five days the contract moved within the band of the previous weeks, while more than twenty national polls split over who led each round. On Saturday the last nine national polls came out with the same split, five to four for Lula in the first round, and the gap widened 16.00pp. On Sunday the official count answered the question the polls could not settle: Flávio Bolsonaro finished ahead by 1.87 points, and the contract went to 86.10%. This issue measures what each instrument showed before the result, and what changed when the ballot box became the reference.
2.Why AFOS does not smooth
The nine national polls released on 3 October, the eve of the vote, measured the first round with samples of 2,000 to 5,000 interviews. A simple average of each one's lead returns 0.3 point for Lula, a number that reads as a tie with a slight lean toward him. The ballot box gave 1.87 points to Flávio Bolsonaro.
Why it matters: the average of 0.3 point for Lula was a number no pollster measured and that pointed to the wrong candidate for the lead. The interval, from 5.1 points for Lula to 4 points for Flávio Bolsonaro, contained the result. What the eve of the vote offered the reader was not a point, it was the band and the split between pollsters, and the band is what was checked at the ballot box.
Issue №19 assigned about 55% to the base case, in which the last wave would repeat the split of the previous weeks, a simple majority giving the second round to Flávio Bolsonaro and a majority giving the first to Lula, with the gap within the band of 10pp to 24pp. Through Friday that is what happened: the five closes stayed between 13.40pp and 22.30pp, and the intraday low of 8.30pp fell 1.70pp below the edge of the band. About 30% went to the contrarian case, in which AtlasIntel, Quaest and Datafolha would put Lula ahead in the second round at the same time. It did not happen in full: on the eve of the vote Datafolha gave Lula 47% to 46% and AtlasIntel 47.6% to 47.4% in total votes, and Quaest gave Flávio Bolsonaro 44% to 42%. About 15% went to the tail, one of the pollsters measuring a majority of valid votes in the first round outside the margin. None measured it, and the ballot box did not deliver it either: Flávio Bolsonaro finished with 47.03%, below 50%.
3.Weighted scenarios for the week
From 5 October the winner contract becomes, in practice, a pair between Flávio Bolsonaro and Lula, and the second-place, third-place and Senate books leave the picture with resolution. The calendar registers four second-round national polls for 8 and 9 October: Datafolha and PoderData on Thursday, AtlasIntel and Vox Brasil on Friday. The eliminated candidates add up to 7.81% of valid votes, with Augusto Cury, Renan Santos and Ronaldo Caiado between 2.18% and 2.89% each.
The first second-round national polls repeat the shape of the eve of the vote, with the pollsters split and the distance within 4 points either way. The contract stays in a high band for Flávio Bolsonaro, between 75% and 90%, because the reference is no longer the polls but the first-round result. The intraday range falls relative to the week of the vote, with four prints over two days instead of nine on a single Saturday.
The first four second-round national polls put Lula ahead outside the margin, driven by the transfer of the 7.81% of valid votes that went to the eliminated candidates and by the 21.08% abstention. In that case the contract, which today pays more than four times as much for Flávio Bolsonaro as for Lula, ends up against the majority of the pollsters in the very first week, and the gap gives back part of what it opened on Sunday.
An institutional act changes the race before 25 October, such as a court decision on a candidacy, a question about the official tally or a new development in the Master case reaching one of the two. The Supreme Court impeachment contract rose 0.80pp on Sunday, to 3.40%, and is the closest gauge of this kind of risk on the dashboard.
4.Indicator Grid
| Contract | Current | Δ week | Vol USD acc. | Implied reading |
|---|---|---|---|---|
| Presidential gap, Flávio minus Lula | +72.60pp | +56.10pp | USD 169.90M | Path: +16.50, +18.00, +22.30, +22.00, +13.40, +29.40, +72.60. Low of 8.30pp at 07:00 UTC on Friday and high of 72.60pp at 23:30 UTC on Sunday, the largest in the paired series. |
| Flávio Bolsonaro, presidential | 86.10% | +28.10pp | USD 15.94M | Low of 53.80% at 07:00 UTC on Friday and high at Sunday's close. Took in USD 3.78M over the window, and the reading normalized by the pair's sum gives 28.15pp. |
| Lula, presidential | 13.50% | -28.00pp | USD 16.31M | High of 45.50% at 07:00 UTC on Friday and low at Sunday's close. Took in USD 3.78M, the same inflow as Flávio Bolsonaro, and his contract remains above his opponent's in accumulated volume. |
| Lula, 2nd place in the first round | 99.30% | +74.80pp | USD 2.05M | From 24.50% to 99.30%, almost all of the shift on Sunday. Took in USD 980 thousand over the window, almost half of the contract's accumulated volume. |
| Flávio Bolsonaro, 2nd place in the first round | 0.80% | -75.70pp | USD 2.51M | The mirror of Lula's contract: the book stopped paying for Flávio Bolsonaro in second when the count put him first. |
| Augusto Cury, 3rd place in the first round | 99.40% | +55.00pp | USD 0.53M | Low of 21.00% at 04:30 UTC on Sunday, hours before the vote. The ballot box gave him 2.89% of valid votes, ahead of Renan Santos, with 2.24%, and Ronaldo Caiado, with 2.18%. |
| Renan Santos, 3rd place in the first round | 1.10% | -42.40pp | USD 0.97M | High of 62.50% at 14:00 UTC on Friday. Took in USD 520 thousand over the window, the largest inflow in the third-place book, and finished fourth at the ballot box. |
| PL, most Senate seats | 97.70% | +4.20pp | USD 0.32M | Low of 90.50% at 01:00 UTC on Monday. The MDB, second at the ballot box with 7 seats, gave up ground from 3.80% to 1.40%. |
| Supreme Court impeachment before 2027 | 3.40% | +0.40pp | USD 0.09M | Range of 0.80pp, with a low of 2.60% on Thursday and the high at Sunday's close, on USD 37 of new money over the whole week. |
5.Liquidity and market structure
The presidential book took in USD 13.41M of new money over the window. Lula and Flávio Bolsonaro took USD 3.78M each. Right behind come three contracts that paid less than 1% throughout the week: Renan Santos, with USD 2.49M, Geraldo Alckmin, with USD 1.77M at 0.20%, and Tarcisio de Freitas, with USD 0.79M at 0.10%. The three add up to USD 5.04M, more than a third of the new money, on names that did not finish among the top two and, in the case of Alckmin and Tarcisio, were not on the presidential ballot. Accumulated volume is trading history since each contract opened and is not current conviction, and anyone who reads the volume ranking as a probability ranking is reading the book backwards.
Volume is accumulated since each contract opened. The week's new money is the difference between the volumes recorded at the close of 28 September and at that of 4 October, at 23:30 UTC, and the prices in the table are those of this last close. 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 | |
|---|---|---|---|
| 08/Oct | Datafolha | n=2,520 | The pollster's first national poll after the first round. On the eve of the vote Datafolha measured Lula 47% to 46% in the second round, with n=4,006. |
| 08/Oct | PoderData | n=3,000 | The pollster measured a tie at 46% in the second round on the eve of the vote, the only one of the nine national polls of 3 October with no lead for either of the two. |
| 09/Oct | AtlasIntel | n=5,000 | The largest sample on the calendar. On the eve of the vote it measured 47.6% to 47.4% for Lula in second-round total votes, the narrowest distance among the pollsters. |
| 09/Oct | Vox Brasil | n=2,100 | On 2 October it measured Flávio Bolsonaro 48.2% to 45.2% in the second round, the second largest distance in his favour that week. |
Registration with the electoral court is not confirmed release. These are the four second-round national polls with samples of 1,000 or more and release scheduled through 11 October, all registered before the first round.
7.Watch list, week triggers
- The first second-round national polls, on 8 and 9 October Datafolha, PoderData, AtlasIntel and Vox Brasil publish over two days. Three of them measured the second round on the eve of the vote, with distances of 1 point for Lula, a tie and 0.2 point for Lula, against the 1.87 points for Flávio Bolsonaro in the first round at the ballot box.
- Where the 7.81% of the eliminated candidates goes Augusto Cury, Renan Santos and Ronaldo Caiado added up to 7.31% of valid votes, and the other names 0.50%. Declared second-round endorsements, with date and source, are the first new data point that none of the pollsters had on the eve of the vote.
- The resolution of the side books The second-place, third-place and most-Senate-seats books resolve on the official result. They closed at 99.30% for Lula in second, 99.40% for Augusto Cury in third and 97.70% for the PL, the same order as the official tally.
- The winner contract as a pair With the other names at the 0.10% floor, the sum of Flávio Bolsonaro and Lula went to 99.60%, and the move on one side becomes almost all of the move on the other. The normalized and raw readings tend to coincide, which was not the case while third place was still in contention.
- The Supreme Court after the vote The impeachment contract rose 0.80pp on Sunday, to 3.40%, after spending the week between 2.60% and 3.00%. It is a book of USD 0.09M accumulated and took in USD 37 of new money over the week, so the move is one of price, 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. This issue covers seven days, from Monday to the close of the count on Sunday, by editorial decision: the convention for the Brazil issues is Monday to Friday, and the week of the vote concentrates on the weekend the shift that defines the series. The window's Δ is Sunday's close minus Monday's, 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 the result is that of the TSE's official tally, read from the court's public files. On the night of the count AFOS published hourly editions of the dashboard and summary, archived with the tally and the price of each hour and open on Hugging Face and Harvard Dataverse. 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 window.
- 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 result.(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.