AFOS Analytics
Real-time political pricing
Prediction markets × polls × news
no smoothed averages
Week in which divergence hit a record for the entire cycle. Between June 29 and July 3 the Lula × Flávio gap moved from +32.95pp to +39.55pp (historic peak on July 3), with Lula breaking 60% and then 61% for the first time (56.50% → 61.50%) and Flávio declining (23.55% → 21.95%). The tradeable fact is the source and resistance of the movement: the week had a single vote print, the AtlasIntel national from July 1 (n=5.000, field June 25-30), which confirmed the DIRECTION (Lula widening: 1º turno 46.3 × 36.6; 2º turno 48.8 × 42.3) but at a vote level well below the market gap, +6.5pp in the 2º turno against +39.55pp of implicit probability. The interval opened with two days of convergence (June 29-30, gap at +31.55pp) and turned into a three-day surge driven by momentum and news flow (Michelle × Flávio crisis, tariff increase). By the weekend the give-back came: the gap declined to +38.20pp (July 4-5), a sign that the July 3 peak may have been an overshoot. Below the top, two structural signals: Renan fell in the winner (12.55% → 9.85%, back below 10%) but WIDENED favoritism to 3rd place in the 1º turno (53.5% → 62.5%), and the PL jumped in the Senate by seats (75.5% → 87.5%, ↑12.0pp). The presidential contract accumulates ~USD 110M, so the record gap was opened with real money. The window from July 6-12 brings no registered national print: the next vote catalyst is the convention season, starting July 20.
1.Executive Summary
The week of June 29-July 3 pushed the market × polling divergence to the widest point of the entire cycle. The presidential market opened a gap of +32.95pp reaching a record +39.55pp on July 3, with Lula breaking through 60% and 61% for the first time, in a three-day move priced on momentum (Michelle crisis × Flávio, tariff hike). The only vote test in the window, the AtlasIntel of July 1 (n=5.000), confirmed the direction but not the level: 1º turno Lula 46.3 × Flávio 36.6 and 2º turno 48.8 × 42.3, a vote gap of +6.5pp against +39.55pp of implied probability. Over the weekend the market returned part of the record (gap at +38.20pp, snapshot July 5 16:54 BRT), a sign of possible overshoot. Below the top, Renan fell in the winner (12.55% → 9.85%) but widened third place (53.5% → 62.5%), and PL jumped in the Senate by seats (75.5% → 87.5%). Snapshot aligned to the AFOS Dashboard; markets operate 7 days, and Sunday's data is the most recent on the covered window.
2.Why AFOS does not smooth
The aggregation industry reduces 'the market versus polls' to a single, smoothed reading. This week, that average erases the fact that defines the window: the market opened the gap to a historic record (+39.55pp) while the only vote print in the interval measured a 2º turno with a margin of only +6.5pp.
Why it matters: the AtlasIntel from July 1 (n=5.000, fieldwork June 25-30) is the only voting snapshot from the window and confirmed the direction the market was pricing in, Lula expanding (1º turno 46.3 × 36.6; 2º turno 48.8 × 42.3). But the 2º turno vote margin, +6.5pp, contrasts with the probability gap that the market pushed to +39.55pp in the same interval. An aggregated reading would eliminate the most useful information: market and poll agree on direction and diverge on magnitude in record fashion, and the market itself began returning from the peak over the weekend without new data. AFOS Tradeoff reports both side by side and treats the divergence as an observed variable. Methodological note: market gap is probability of victory, not vote margin, so the comparison is one of direction and conviction, not pp to pp.
Different from Edition №6, in which the record gap was opened WITHOUT any vote poll in the window, this week the record came WITH a print (AtlasIntel July 1) that confirmed the direction but not the level, and was followed by a give-back over the weekend. The methodology reading is the same: the AFOS Tradeoff registers the level divergence between probability of victory and vote margin and maintains the ballot results as final arbiter. The next vote test only arrives with the convention season (starting July 20), so the window July 6-12 tends to be market drift without a polling catalyst.
The same framework we apply to Brazil—measuring the distance between the prediction market and polls and checking the signal against the actual election results—has already been tested against nine elections across four continents: South Korea (2025, Asia's first case), Peru and Colombia (2026), Chile, Germany and Canada (2025) and United Kingdom, Mexico and United States (2024). In some cases the signal was one of high divergence; in others, of near-zero convergence (Colombia, Germany); and in the United States in 2024 two markets disagreed with each other (the Electoral College was right, the popular vote was wrong), recorded with radical honesty. What validates the method is the actual result, not the direction of the signal, which is why we don't speak of 'validated divergences', but rather of cases checked against what the polls delivered. It is auditable and open track record, not claims of predictive superiority. The cases are in the AFOS global hub.
3.Weighted scenarios for the week
Three paths for the window 06-12/Jul, which does NOT bring any national registered vote snapshot (the next catalyst is the convention season, starting 20/Jul):
Without a national poll in the window, the market consolidates the gap in the +36 to +39pp range, digesting the weekend give-back without a new voting catalyst. Lula around 60%, Renan stable near 10% in the winner and ~62% in 3rd place, maintaining the ~6pp divergence versus the ~4% of national polls. Market and research remain aligned in direction; the probability level remains well above the voting margin. Net-neutral for BRL.
The give-back from July 4-5 extends: with no polling data to support the record gap, the market converges toward the voting signal and the gap compresses to +33 to +35pp (closer to the 2º turno +6.5pp from AtlasIntel and the Empate técnico from in-person polls, BTG/Nexus 47 × 44). Implicit: the peak of +39.55pp was momentum overshoot, and the absence of a polling catalyst accelerates normalization.
Evento fora da dinâmica de momentum desloca o quadro: um desdobramento agudo do caso Master/Vorcaro no STF ou um movimento de convenção antecipado (chapas, federações) que altere o campo da oposição. O contrato de impeachment a 2.75% sinaliza que o mercado não precifica ruptura institucional iminente; reprecificação aqui exigiria gatilho novo.
Event outside momentum dynamics shifts the frame: an acute development in the Master/Vorcaro case at the STF or an early convention movement (slates, federations) that alters the opposition field. The impeachment contract at 2.75% signals that the market does not price in imminent institutional rupture; repricing here would require a new trigger.
4.Indicator Grid
| Contract | Current | Δ week | Vol USD acc. | Implied reading |
|---|---|---|---|---|
| Lula · winner | 60.50% | ↑4.00pp semana | 7.2M | Broke through 60% and 61% during the week (peak 07/03); gave back part of gains over the weekend |
| Flávio · winner | 22.30% | ↓1.25pp semana | 7.2M | Retreated under adverse flow (Michelle × Flávio crisis, tariff hike); continues as the certain opponent |
| Gap Lula × Flávio | +38.20pp | ↑5.25pp semana | - | Historic peak +39.55pp on July 3; pullback to +38.20pp over the weekend |
| Renan · 3rd place 1º turno | 62.50% | ↑9.00pp semana | 125k | Favoritism expanded; ahead of Caiado (12.5%) and Zema (6.5%) |
| Renan · winner | 9.85% | ↓2.70pp semana | 7.8M | Back below 10%; highest volume among active candidates; diverges ~6pp from the 1º turno of polls (~4%) |
| Flávio · 2nd place 1º turno | 72.50% | ↓6.00pp semana | 142k | Leader of 2nd place, but ceded part of its share to Renan (11.4%) over the weekend |
| STF impeach <2027 | 2.75% | ↓0.15pp semana | 82k | At floor; Master case cooled in the window; thin volume, fragile signal |
| PL plurality Senate | 87.50% | ↑12.00pp semana | 250k | Jump in the week; broad favoritism without close contender (MDB 11.25%) |
| Inflation 2026 modal band 5.0-5.49% | 37.80% | ↑1.80pp semana | - | Modal band firmed near 38%; consensus 4.50-5.49% totals 66.20% |
5.Liquidity and market structure
The five largest volumes in the entire book continue in probability contracts ≤1.65%: Tarcísio (USD 13.31M), Carlos Massa (10.29M), Eduardo Bolsonaro (10.02M), Michelle (8.65M) and Eduardo Leite (7.78M), all legacy positions never unwound. It is legacy volume, not current traction; none are active contenders in the range. Read 'low price + high volume' as concentrated conviction already priced in and resolved, not as live interest. The relevant data for the week sits just below the top-5: the contenders with material probability, Renan (7.8M), Flávio (7.2M) and Lula (7.2M), remain clustered around USD 7-8M, and Renan still leads accumulated volume among the active candidates, sustaining 9.85% in the winner (diverges ~6pp from 1º turno polling) while their third-place contract WIDENED from 53.5% to 62.5% during the week.
The 5 largest accumulated volumes account for ~USD 50M (~45%) of the presidential market (total ~USD 110M). The cross-reading of price × volume for the week is at the top: the opening of the gap to a record (+39.55pp on July 3) was made on real flow (Lula USD 7.2M, Flávio USD 7.2M), not on shallow book, and with a single vote print in the window (AtlasIntel July 1). Low volume spike (USD <500k) in individual contract should be treated as noise until confirmation of recurrent flow.
6.Calendar of price-relevant prints
| Date | Sample | Why it matters | |
|---|---|---|---|
| Mon Jul 06 | Veritá (state) | n=2.000 · BR-04161/2026 | Opens the window; state-level scope, no reading on the national presidential landscape |
| Tue Jul 07 | Social Research Opinion (state level) | n=2.000 · BR-01893/2026 | State-level; regional context, not price-relevant to the presidential gap |
| Wed Jul 08 | Gerp (state) | n=2.000 · BR-03067/2026 | State-level; no direct impact on national pricing |
| Wed Jul 08 | Ideia/Canal Meio (state level) | n=1.500 · BR-05628/2026 | State level; contextual reinforcement |
| 20/Jul+ | Convention season | - | Next real catalyst: definition of tickets and federations, with potential to reprice the opposition field |
Source: TSE registry via AFOS API. The window July 6–12 brings no NATIONAL vote poll registered: all are state-level (the 12 samples ≥ 1,500 are listed in the AFOS Daily synthesis). Without a poll catalyst until the convention season (starting July 20), the week tends toward market drift. Registered ≠ published: inclusion does not guarantee disclosure or numbers. Filter applied: sample ≥ 1,000.
7.Watch list, week triggers
- Give-back vs new regime. The gap returned part of the July 3 peak (+39.55pp) to +38.20pp over the weekend. Monitor whether it stabilizes above +36pp (consolidated high gap regime) or compresses to +33-35pp (overshoot reversing) in the absence of vote print.
- Research vacuum (July 6-12). No national print recorded in the window. Without voting data to arbitrate, any gap movement is pure momentum; the first real test only arrives with the conventions (July 20).
- Renan's Divergence. The winner contract fell below 10% (9.85%) while the third-place contract EXPANDED to 62.5%. Monitor whether any state or national poll registers Renan above 4% in the 1º turno, or if Caiado (12.5%) approaches third place.
- PL surge in the Senate (75.5% → 87.5%). The Senate plurality contract jumped 12pp in the week, with no close contender (MDB 11.25%). Monitor whether the movement reflects convention repositioning or thin book spike (USD 250k).
- Caso Master / Vorcaro at STF. The case cooled off in the news this week, but remains open (Toffoli's confidentiality order, Mendonça's rapporteurship). Any act that moves the impeachment contract away from 2.75% is an institutional signal to monitor.
8.Methodology
AFOS Tradeoff aggregates three signals without mediating them into a composite: Polymarket (denominated in USD, latency ~30min), polls registered with the TSE (declared intent, variable frequency) and 400+ press sources (event flow). When the three diverge, the divergence is the signal, not the consensus. USD volume is reported alongside implied probability to separate conviction from artificial spike. Liquidity (order book depth) is not cited inline because low liquidity on Polymarket does not mean wrong price (active arbitrage in minutes), and exposing the technical number generates misreading in lay audiences.
Weekly Δ series derive from snapshots persisted daily on Neon (June 29–July 5), on a basis consistent with the dashboard pipeline. Source code, raw data and daily editorial synthesis (PT/EN/ES) under Apache 2.0 license: afos-analytics.com · github.com/AFOS-Analytics.
9.Additional reading · macro coverage
Relevant articles and sections from the week in reference outlets. AFOS Tradeoff is the primary source (Polymarket + TSE + polls); the references below are complementary reading for macro context alignment. Note: some content operates behind a paywall.
- O Globo · Datafolha: Haddad has the highest rejection rate among SP voters, with 47% (July 5)(paywall), oglobo.globo.com/politica/noticia/2026/07/05/datafolha-haddad-tem-maior-rejeicao-entre-eleitores-de-sp-com-47percent.ghtml
- Folha de S.Paulo · PT fears Tarcísio's victory in 1º turno in SP, but assesses that there is still room for Haddad to grow (July 5)(paywall), www1.folha.uol.com.br/colunas/painel/2026/07/pt-teme-vitoria-de-tarcisio-em-1o-turno-em-sp-mas-avalia-que-ainda-ha-espaco-para-haddad-crescer.shtml
- Reuters · section Americas/Brazil, political flow and country risk, www.reuters.com/world/americas/
- Bloomberg Línea · coverage Brazil/emerging markets and BRL(paywall), www.bloomberglinea.com.br/
- Valor Econômico · politics and markets, electoral cycle coverage(paywall), valor.globo.com/politica/
- InfoMoney · politics and market (open access), www.infomoney.com.br/politica/
Macro context references. The primary signal in this edition (Lula × Flávio gap at historic cycle peak, +39.55pp on July 3, opened with a single vote print that confirmed direction but not magnitude, and the subsequent give-back) is direct observation of Polymarket pricing crossed with TSE records, not derived from the articles above.