May 20, 2026

---

1. Prediction market

In the presidential market at Polymarket — market with USD 86.80M accumulated since opening (snapshot May 20, 18:00 BRT) —, Lula remained at 45.50% (USD 5.65M), stable vs May 19 evening, while Flávio Bolsonaro continued yielding to the 25-26% range (USD 5.75M) — third consecutive day of erosion post-Vorcaro audio. The Lula × Flávio gap closed around +19-20pp Lula — variation of approximately ↑2.5pp in 24h.

The third way reorganized itself. Renan Santos surged to the 14% range (USD 5.50M), consolidating second position in the presidential Polymarket above Zema (≈5.65%) and Haddad (≈3.50%, low base but rising). In the 3rd place market, Renan held the top around 35%, Zema yielded to ≈29%, and Caiado rose to ≈15-16% — first recovery after colossal decline on May 18-19. Relevant initial movement: Flávio began appearing in the 3rd place market around 6-7% (vs 4.80% on May 19) — the market pricing the hypothesis of Flávio finishing third if he exits the main race.

In the 2nd place market, Flávio began yielding the PL base to the 59-60% range (vs 62.50% stable on May 19) — first material erosion in the 2L contract since the audio. Michelle Bolsonaro registered first measurable traction rising to ≈2.50-3.00% in 2L (vs 1.55% on May 19) — narrative of substitution in the PL ticket gaining hedge capital for the first time.

The contract for impeachment of STF justice before 2027 registered ≈6.60% (USD 71k), modest upward correction sustained vs the May 18 minimum of 5.30% — solid institutionality returns to pricing. In the Senate, PL at ≈77-78% (stable), Republicanos near 3-4%. In the 2026 inflation bands, high tail (≥6.50%) around 9%, stabilizing after May 19 collapse.

2. What the institutes recorded

Vox Brasil released a national survey with Lula 46.8% and Flávio Bolsonaro 38.1% in the 2º turno (gap +8.7pp). Lula rejection was 52.8% and Flávio 49.2%. Parallel coverage of the same survey in Exame, Folha de Alphaville and Revista Fórum confirmed the same numbers — second institutional confirmation of the Vorcaro effect after AtlasIntel.

Vox Brasil is the second national survey to register Flávio deterioration post-Vorcaro audio, after AtlasIntel on May 19 (1st round Lula 47% × Flávio 34.3%; 2º turno 48.9% × 41.8%). Taken together, the two institutes show consistent gap around +7-9pp Lula in the 2º turno — significant methodological convergence, given that AtlasIntel and Vox use distinct sample designs. The Datafolha from May 16 (TSE record BR-00290/2026) remains the most conservative reference point (1st round 38×35 gap +3pp; 2º turno 45×45 tie); the 5.7pp discrepancy between Vox 1st round and Datafolha 1st round became the largest divergence between institutes in the cycle.

The next expected national release is Datafolha on May 22, n=2.004, including scenario with Michelle Bolsonaro as substitute on the PL ticket — direct test of the reorganization discussed openly by political actors today.

📅 Survey calendar — next 7 days

Surveys registered at TSE with expected publication between May 21 and May 27. Table inclusion does not mean confirmed publication — institutes may delay or cancel release. Filter applied: sample ≥ 1.000. Each protocol linked to TSE public consultation.

Source: TSE public record via AFOS API. 🔥 highlights samples ≥ 3.000 or national anchor surveys. Status "registered ≠ published" — confirmation of effective release requires verification of two primary sources before citation of numbers.

3. What the press covered

The day's agenda was dominated by the public reorganization of the right around Flávio's vulnerability. The PSDB initiated internal debate about launching Aécio Neves to the Presidency, the first explicit signal of reorganization by the non-Bolsonaro right. Ronaldo Caiado declared that "anyone contaminated by Vorcaro cannot be president" without naming Flávio directly — a clear message directed at the senator. Romeu Zema delivered a speech on "credibility to lead the country" also without direct reference (Folha de S.Paulo, panel). Symmetrical movement between the two governors: positioning themselves as alternatives without burning bridges.

More expressive, Michelle Bolsonaro ignored Flávio in her speech and criticized an "alliance with evil" in reference to Ciro Gomes (Folha de S.Paulo, power). Ricardo Salles stated that "a slate with Michelle in place of Flávio would gain much support" (BBC News Brasil interview) — the first public statement by a Bolsonaro ally suggesting internal replacement. Carlos Bolsonaro, simultaneously, suggested that Zema would be "surfing" Flávio's crisis — a signal that internal tension in the Bolsonarist camp is public.

On the operational side, the erosion materialized in two concrete actions. Flávio switched marketing advisors following the Dark Horse case — the advertising executive Eduardo Fischer is set to assume campaign marketing (Folha de S.Paulo). And PL allies admit reviewing the candidacy: O Globo reports that allies of Flávio Bolsonaro admit reconsidering support for the presidential candidacy if new facts emerge about the Master case, and the Valdo Cruz column on G1 reports that allies assess that Flávio needs to escape the "bad vibe" at risk of becoming a non-viable candidate. Peripheral reports amplify: there is record that Valdemar Costa Neto faces pressure and the PL may remove Flávio from 2026 Elections (cautious attribution — lower-prestige source). The directional signal converges with what is being stated publicly by the governors themselves and by Salles.

On the institutional level, records from JOTA (May 20 publication) point to judicialization as the central tactic of the Lula × Flávio campaign — TSE as the axis of dispute before even official registration. BBC News Brasil published a retrospective one week after Vorcaro highlighting the accumulated contradictions in the Bolsonaro camp, and investigations into Banco Master continued advancing — seven CPMI requests in Congress remain on the table.

4. Day's divergences

In summary

1. Vox Brasil is the second institutional confirmation of the Vorcaro effect — after AtlasIntel May 19, Vox publishes on May 20 2T Lula 46.8% × Flávio 38.1% gap +8.7pp, with rejection of Flávio at 49.2%. Two large national institutes (n=5,000 and n=1,200) converging on the deterioration signal — level of institutional evidence that makes it difficult to sustain a narrative of "passing noise".

2. Non-Bolsonarist right publishes positioning openly for the first time — PSDB debates Aécio Neves; Caiado speaks of "contamination by Vorcaro"; Zema of "credibility"; Michelle of "alliance with evil"; Salles defends replacement by Michelle. Five coordinated public moves in different directions but with a common vector: distance themselves from Flávio publicly.

3. Flávio operation acknowledges the crisis without abandoning the candidacy — change of marketing strategist after Dark Horse + reports (O Globo, G1 Valdo Cruz) on PL allies reconsidering the candidacy + pressure on Valdemar Costa Neto. The public campaign denies it; the operational signals converge with what internal adversaries are saying.—

Sources consulted. Articles with direct links to news (anchor outlets): O Globo (Caiado contaminated, Flávio allies admit reconsidering), G1 (Valdo Cruz allies Flávio bad luck), Folha de S.Paulo (Zema credibility, Michelle ignores Flávio, Eduardo Fischer campaign strategist — via Google News redirect and panel). Secondary articles via Google News redirect: Poder360 (Vox Brasil 46.8% × 38.1%), CNN Brasil and G1 (PSDB Aécio Neves), Estado de Minas (Carlos Bolsonaro on Zema), Bnews and Área VIP (Valdemar pressure cautious attribution). Outlets cited by plain-text without direct URL: BBC News Brasil (Salles "Michelle ticket would gain support"; Dark Horse retrospective one week), JOTA (judicial campaign tactics), Exame and Revista Fórum (Vox Brasil parallel coverage).

Method. Cross-reference Polymarket (prices and accumulated volumes live via gamma API — Brazil presidential, 2nd place, 3rd place, STF impeach, Senate, 2026 inflation), electoral surveys registered with the TSE (Vox Brasil May 20 published nationally; AtlasIntel May 19 n=5,000; Datafolha May 16 n=2,004 BR-00290/2026) and daily journalistic coverage. Electoral surveys registered with the TSE are verifiable on the public disclosure portal. Each factual claim appears with inline link to primary source or plain-text attribution to outlet with exact date. Synthesis generated by AFOS Analytics and reviewed before publication.