What is AFOS Analytics?

AFOS Analytics is a global electoral intelligence platform that cross-references real-money prediction markets (Polymarket), polls from 17+ institutes, live news, and strategic analysis in real time.

Is AFOS Analytics free?

Yes. Access to the platform is completely free, with no registration required. The project is open source.

What are prediction markets?

Prediction markets are platforms where people bet real money on future events. Unlike opinion polls, they reflect where people put their money — historically more accurate than traditional polls.

Which elections does AFOS monitor?

AFOS monitors elections in 14+ countries, including Brazil 2026, USA, France, Germany, UK, Canada, Australia, South Korea, Colombia, Chile, and more.

How is the data updated?

Prediction market data is updated every 30 minutes via cron job. News is updated every 30 minutes. Analysis is updated manually with source cross-referencing.

AFOS Daily — May 22, 2026

Name: AFOS Analytics Election Data
Creator: AFOS Analytics
License: https://opensource.org/licenses/MIT

Three movements define the day. The Datafolha published this Friday is the institute's first poll after the Vorcaro case and shows Lula 40% × Flávio 31% in the 1º turno (gap +9pp Lula, expansion of 6pp vs Datafolha from May 16) and 47% × 43% in the 2nd round (gap +4pp Lula vs previous technical tie 45×45). On Polymarket — USD 88.7M accumulated in Brazil 2026 cycle since opening, Flávio recovered to 28.05% (USD 5.92M accumulated), up ↑3.40pp in 24h against the low of 24.65% from the close on May 21 — the market read the poll as less catastrophic than the "Flávio on the hot slate" scenario. And the STF institutional axis hardened: André Mendonça voted to maintain the imprisonment of Vorcaro's father and cousin, Gilmar Mendes requested time to review, and the PF rejected the banker's plea deal proposal, with the attorney stepping down from the defense.

1. Prediction market

The Polymarket markets in Brazil 2026 cycle total approximately USD 88.7M accumulated since opening (snapshot May 22 17:53 BRT), with the presidential market concentrating more than 95% of total volume — where the signal is most robust.

In the presidential market, Lula remained stable at 45.50% (USD 5.72M accumulated) for the third consecutive day, unchanged since the morning of May 21. Flávio Bolsonaro recovered to 28.05% (USD 5.92M accumulated) — up ↑3.40pp in 24h against the low of 24.65% from the close on May 21. Still, it trails ↓3.25pp in five days against the 31.30% from May 17. The Lula × Flávio gap closed at +17.45pp Lula (↓3.40pp in 24h, mechanical narrowing because Flávio recovered). The counterintuitive movement — Flávio rises in the market the same day he falls in Datafolha — suggests that the market had been pricing a worse scenario than what the poll delivered (the +4pp gap in the 2º turno is less wide than part of the bets presumed).

The third way ceded nearly uniformly. Renan Santos at 11.75% (USD 5.64M accumulated) fell ↓0.80pp in 24h, but maintains second position in presidential Poly above Zema (5.05%, USD 2.75M), Haddad (4.55%, USD 4.87M) and Michelle Bolsonaro (3.35%, USD 6.51M — second largest volume position among third-way candidates). Caiado ceded to 1.05% (USD 3.04M). In the 2nd place market, Flávio rose to 58.00% at the top (↑2.50pp in 24h), while Renan advanced to 16.25% (USD 1.0M) — only candidate with volume above USD 1M in 2L. Lula at 11.50%, Haddad at 4.30%, Zema at 2.90%, Michelle at 2.65%. In the 3rd place market, Renan remained at the top with 33.00% (↓4.00pp in 24h, ceding part of the space Flávio recovered in the presidential), Zema at 23.00%, Caiado at 15.00%, and Flávio at 11.10% (↑1.80pp in 24h) — the market continues pricing material risk of Flávio finishing in third.

The contract for impeachment of STF minister before 2027 partially reversed the previous day's collapse to 6.45% (USD 80k accumulated), up ↑2.10pp in ~22h against the 4.35% recorded after the close on May 21. The movement validates the hypothesis noted in the prior update that the post-close collapse had been low liquidity distortion (a small order in USD 75k sub-market moved price disproportionately). Against the May 21 close at 17h30 (7.55%), the contract still cedes ↓1.10pp — modest net trajectory within the window.

In the Senate, the PL ceded to 69.00% (USD 240k accumulated) — relevant decline of ↓6.00pp in 24h, maintaining the continued downward trajectory since May 21. União Brasil at 4.95%, MDB at 4.90%, PSD at 4.15% (↓1.60pp), Republicanos at 4.05%, PSB at 3.60%, PT at 3.05% (↑0.20pp in 24h), Podemos at 2.50%, Novo at 1.10%. In the 2026 inflation bands, the most relevant movement was massive migration to the upper-middle band: band 5.50-5.99% surged to 23.05% (↑11pp in 24h) from 12.05% at the May 21 close, while the 4.50-4.99% band ceded to 21.45% (↓5.90pp). The 5.00-5.49% band remains modal at 25.85% (↑1.45pp). Upper tail (≥6.50%) remains at 7.55% (stable). The redistribution suggests the market shifted the center of inflationary expectation distribution from ~4.7% to ~5.3% in 24h.

2. What the institutes recorded

The Datafolha published this Friday is the first Tier 1 national poll entirely in field after the Vorcaro case and strongly confirms the deterioration captured by AtlasIntel (May 19) and Vox Brasil (May 20). In the first round, Lula has 40% × Flávio 31% — gap of +9pp Lula, expansion of 6pp against the +3pp registered by Datafolha from May 16 (field May 12-14, before the scandal). In the second round, Lula 47% × Flávio 43%, gap of +4pp Lula against previous Empate técnico 45×45. In the scenario where Michelle Bolsonaro replaces Flávio on the slate, Lula has 48% × Michelle 43% (gap +5pp Lula) — Michelle loses more than Flávio in the same scenario, signal that the substitution narrative does not hold in field. Rejection of Flávio Bolsonaro rose to 46% and numerically exceeded Lula's (45%) for the first time in the series.

A second national poll was published today: the Apex/Futura shows a drop of 4.7 points for Flávio and Lula's lead in the 2nd round. Together, Datafolha, Vox and Apex/Futura form a convergence of three institutes with distinct methodologies — convergence that reduces the chance the captured deterioration is an isolated sampling artifact. The divergence among institutes in the second round shifted to concentrate on how far ahead Lula is, not on who leads: Datafolha +4pp, Vox Brasil +8.7pp, AtlasIntel +7.1pp.

📅 Poll calendar — next 7 days

Polls registered with the TSE with publication scheduled between May 23 and May 28. Inclusion in the table does not mean publication is confirmed — institutes may delay or cancel release. Filter applied: sample ≥ 1,000. Each protocol linked to TSE public consultation.

Date	Institute	Sample	Scope	TSE Protocol	Conf.
May 26	Nova Ibrape	1,000	state	BR-006152026	0.6
May 26	IPEN Norte	1,200	state	BR-068352026	0.6
May 26	Real Time Big Data	1,600	state	BR-002822026	0.8
May 27	JOTA Jornalismo 🔥	6,000	national	BR-029972026	0.7
May 27	Veritá (3,330) 🔥	3,330	state	BR-042762026	0.7
May 27	Veritá (3,025) 🔥	3,025	state	BR-081862026	0.7
May 27	Veritá (2,020)	2,020	state	BR-032132026	0.7
May 27	INDEXA	2,000	state	BR-021542026	0.7
May 27	Real Time Big Data	1,600	state	BR-035622026	0.8
May 27	Veritá (1,525)	1,525	state	BR-098432026	0.6
May 27	Veritá (6× ≈1,220)	1,220	state	BR-010742026	0.6

Source: TSE public record via disclosure portal and AFOS Analytics API. 🔥 highlights samples ≥ 3,000. The calendar highlight is JOTA Jornalismo with n=6,000, registered for May 27 — most robust sample-size of the cycle registered so far. Status "registered ≠ published" — confirmation of actual disclosure requires verification from two primary sources before citing numbers.

3. What the press covered

In the Vorcaro case field, two institutional movements redefined the day's axis. André Mendonça voted to maintain the imprisonment of Daniel Vorcaro's father and cousin — both have remained imprisoned since May 14, in the 6th phase of Operation Compliance Zero. Gilmar Mendes requested a stay and suspended the proceedings, a standard procedural maneuver that does not overturn Mendonça's decision but postpones the collegial vote. In parallel, the PF rejected Daniel Vorcaro's plea bargaining proposal and his attorney left the defense — Folha separately reported the attorney's departure following the rejection, and O Globo detailed that the disagreement between Mendonça and Vorcaro, combined with the PF's rejection and the cell transfer, was what led to the attorney's departure. Mendonça authorized Vorcaro to return to a special cell at the PF, and O Globo reported that Vorcaro's father had an episode in a prison in Minas Gerais.

On the PL side, O Globo noted that "PL and Centrão downplay Flávio's decline in Datafolha, but fear greater losses if new facts emerge", and the PL leadership sees urgency in "holding platforms and securing the election of caucuses" amid the crisis. Jair Bolsonaro himself is acting to try to save Flávio and contain tension with Michelle, while Malu Gaspar reported that the Master scandal "stalls the reconciliation" between Michelle and Flávio. A relevant observation for the narrative of "Flávio demands CPI" recorded on May 21: Gazeta Mercantil reported that Flávio Bolsonaro did not sign 3 of the 5 CPI requests about Banco Master — a factual contradiction that exposes the senator's own version. In a parallel column, Bernardo Mello Franco wrote that "in the theater of Congress, even Flávio Bolsonaro defends the Master CPI", exploring the same discrepancy. Alcolumbre will block a new possibility for creating a Master CPI in Congress.

On the government front, Lula's campaign assesses that the Vorcaro effect is real but believes right-wing voters will return to Flávio — explicit caution from Planalto itself. Lula strengthens the leadership team of the pre-campaign "with an eye on Flávio's crisis". On the legislative front, the Lula government freezes R$ 22.1 billion in the budget to reduce the INSS queue on the eve of the election, and Lula says he will veto a bill that opens a loophole for mass messaging. Congress overrode Lula's vetoes and decided to release funds to municipalities before the election. In the narratives field, Marcos Nobre told Folha that "the timing of the Vorcaro scandal was good for Flávio Bolsonaro — it will give him time to recover" — a technical reading that helps explain Flávio's recovery on Polymarket despite deterioration in polling. Vera Magalhães wrote that "the Master case wounds but does not kill Flávio, and Lula gains short-lived momentum". In Brasília, Folha's Painel reported that a governor wants to transform Piauí into Lula's main electoral stronghold.

4. Divergencies of the day

Market × survey: Polymarket continues significantly more aggressive than any published national survey. The gap of +17.45pp Lula × Flávio in the market contrasts with a gap of +4pp in Datafolha from May 22 (runoff), +7.1pp in AtlasIntel (May 19), +8.7pp in Vox (May 20). The market × Datafolha runoff divergence is on the order of 13.5pp — significantly larger than the 14pp from the previous week, but with Datafolha now confirming the direction of the movement. The market appears to price not only what surveys have captured but also unmeasured institutional developments: PF rejected Vorcaro's plea deal, defense in fragility, Mendonça consolidates position against Master.

Market × survey (intra-day): Flávio recovered ↑3.40pp on Polymarket on the same day it fell 4pp in the 1º turno of Datafolha. The counter-intuitive movement suggests that the market had been pricing in a worse scenario than what the survey delivered — or, in technical terms, some of the bets presumed that the gap in the runoff would be larger than +4pp. The reading aligns with what Marcos Nobre wrote in Folha de S.Paulo about "timing will give Flávio time to recover". There is no individual event that justifies Flávio's isolated recovery in 24 hours; it is a reading of mean reversion via conservative data.

Market × institutional field: The STF impeachment contract partially reversed to 6.45% — a rise of ↑2.10pp in ~22 hours against the collapse of 4.35% recorded after the close on May 21. The movement validates the hypothesis registered in the previous update that the collapse was a distortion of low liquidity (sub-market of USD 80k accumulated, where a small order moves price disproportionately). In parallel, PL fell to 69% in the Senate (↓6pp in 24 hours) — market continues repricing PL hegemony in the Senate alongside Flávio's individual wear. Massive migration from the 4.50-4.99% band to the 5.50-5.99% inflation band (↓5.9pp / ↑11pp) signals realignment of the center of the inflation expectations distribution for 2026, with no individual macro event that justifies the redistribution.

In summary

The Datafolha post-Vorcaro closed the convergence among three national institutes: gap Lula × Flávio between +4pp and +9pp in the runoff (Datafolha 47×43; AtlasIntel 48.9×41.8; Vox 46.8×38.1) — methodological convergence that reduces the chance that the deterioration is an isolated artifact. For the first time in the series, rejection of Flávio (46%) numerically exceeded that of Lula (45%).
The Polymarket reacted counter-intuitively to Datafolha: Flávio recovered ↑3.40pp in 24 hours because the gap of +4pp in the runoff was smaller than some of the bets presumed. The movement confirms the pattern observed throughout the week — the market prices the worst scenario before the survey, and calibrates to "less bad than expected" when the survey delivers a less catastrophic result.
The STF axis hardened in the Master case: Mendonça voted for the maintenance of Vorcaro's father and cousin's detention, Gilmar suspended judgment pending review but did not overturn the decision, and PF rejected the banker's plea deal proposal. The Polymarket STF impeachment contract partially reversed the May 21 collapse to 6.45% — a rise that validates the hypothesis of liquidity distortion registered in the previous update. The narrative "Flávio demands CPI" registered on May 21 requires qualification: Gazeta Mercantil reported that he did not sign 3 of the 5 CPI requests on Banco Master.

Sources consulted

Articles with direct link to the news:

Secondary articles (Google News redirect URL — click resolves to article):

Sources cited in this text: Folha de S.Paulo; O Globo; Datafolha; Apex/Futura; Vox Brasil; AtlasIntel; JOTA; Gazeta Mercantil; Polymarket; TSE disclosure portal.

Method: this synthesis is automatically generated from AFOS Analytics platform auditable data, under versioned code rules in git. All claims can be verified on the platform or in the linked sources.

Integration: to see live data and detailed candidate analysis, visit the complete Dashboard. To understand the method in depth, read The Method.