AFOS Analytics
Real-time political pricing · Prediction markets × polls × news · no smoothed averages
Week of cross-directional validation: Polymarket priced in deterioration of Flávio Bolsonaro post-Vorcaro audio between May 19-22, and Datafolha of May 22 confirmed in polling — 1º turno gap expanded +6pp vs May 16 (40×31, gap +9pp). May 23 consolidated via heavy editorial reverberation: 48% support Flávio stepping down from candidacy (G1), Lula approval rises +4pp excellent/good Datafolha (32% vs 28%), Folha publishes O rachadão dos Bolsonaros. Market stays ahead: Polymarket gap Lula × Flávio at +17.75pp (Lula 45.50%, vol USD 5.73M accumulated · Flávio 27.75%, vol USD 5.94M). STF impeach makes cycle low 4.35% (May 21 19h) and corrects to 6.65% after Mendonça and Fux vote to maintain Vorcaro's imprisonment + Congress stalls PF installation of the CPI. PL plurality Senate recovers to 70.00% (↑1.00pp 24h). Inflation bands ≥6.50% normalize at 7.70% after May 22 transitory anomaly. Next week (May 26-30): Lula+Motta meeting on ending the 6×1 and Flávio's trip to the USA as editorial triggers.
1.Executive Summary
The week of May 19–23 resolved a divergence open since May 14–16: Polymarket priced the deterioration of Flávio Bolsonaro before Datafolha registered it — the market × research gap reached +13pp × +3pp on May 16 (a 10pp divergence). The Datafolha of May 22 (n=2.034) expanded the research gap to +9pp 1º turno (40%×31%) and +4pp 2º turno (47%×43%), with the additional reading that Flávio's rejection at 46% exceeds Lula's at 45% for the first time in the series. The AtlasIntel of May 19 (n=5.000) had already indicated the direction with a +12.7pp gap 1º turno. Three tier-1 prints in the same week, all validating market pricing. The technical reading is that the divergence cycle closed partially — but the market continues pricing a larger magnitude (+17.75pp 1º turno) than any poll has registered. The next Datafolha print will resolve whether the market over-priced or if the validation cycle continues.
2.Why AFOS does not smooth
Most of the electoral signal aggregation industry reduces divergence through weighted averaging — "consensus tracker" models treat differences between market, polling, and coverage as noise to be canceled out. For this audience, that is exactly the error to avoid — even when signals converge in the same direction.
Why it matters: four signals aligned in the same direction are not redundant. Datafolha publishes +9pp in a survey registered with sample n=2.034 and certified methodology. Polymarket prices +17.75pp in a market denominated in USD with accumulated vol ~USD 80M in the presidential race. The magnitude difference (~8.75pp) carries information: the market prices additional deterioration not yet captured by survey, or survey underestimates an effect the market has already seen. Averaging destroys the question — which of the two will be correct in the next measurement.
Recent weekly history: market-survey gap was at +13pp × +3pp on May 16 (divergence of both magnitude and weak direction). After May 19-22, the gap became +17.75pp × +9pp (divergence only of magnitude, convergent direction). AFOS Tradeoff reports the four numbers separately and maintains the magnitude difference as an observed variable — not as noise to be corrected. The May 19-22 validation cycle was precisely what makes the price-relevant magnitude divergence: the market was directionally correct and the survey confirmed it. The next national survey will determine whether magnitude also converges.
3.Weighted scenarios for the week
Three plausible paths for the next screenshot window (May 26–June 9):
Next national poll (Quaest, AtlasIntel or Datafolha in Jun) registers first round gap between +10pp and +13pp, partially compressing the distance to the market without fully closing it. Polymarket reacts with mild gap compression to the +14 to +16pp range. STF impeach sustains above 6%. Net-neutral for BRL; mildly supportive for local names linked to fiscal continuity and privatizations.
Next print shows stabilization of the gap at ~+8pp (without additional expansion), suggesting that Lula's +4pp Datafolha approval was a transitory peak of the Dark Horse effect. Polymarket compresses gap below +12pp in the window. STF impeach revisits 5%. BRL bid, sectoral rotation in Ibov exiting incumbency-linked names. Implicit: market over-priced Flávio's deterioration.
New poll introduces Michelle Bolsonaro as a substitute on the PL ticket and registers viability significantly above the 43% runoff of the Datafolha scenario (>46%). Structural repricing of the right-wing trade — Michelle contract currently at 2.90% with USD 6.56M accumulated vol (second largest of the third way) would be the first to move. Tail has gradually declined this week (Datafolha May 22 showed Michelle 43% vs Flávio 43% — tie, NOT overtaking).
4.Indicator Grid
| Contract | Current | Δ week | Vol USD acc. | Implied reading |
|---|---|---|---|---|
| Lula [1º turno](/en/glossary#primeiro-turno) (Polymarket) | 45.50% | flat 48h+ | 5.73M | Stable after [Datafolha](/en/glossary#datafolha) validates thesis |
| Flávio [1º turno](/en/glossary#primeiro-turno) | 27.75% | ↓3.55pp 5d | 5.94M | Yields to post-Vorcaro audio cycle low |
| Gap Lula–Flávio | +17.75pp | ↑0.30pp 24h | — | Above any poll this week |
| [STF](/en/glossary#stf) impeach <2027 | 6.65% | ↑2.30pp 48h | ~79k | Post-minimum correction 4.35% (May 21, 19:00 BRT) |
| [PL](/en/glossary#pl) plurality Senate | 70.00% | ↑1.00pp 24h | ~243k | Recovers after 67% test |
| Renan Santos third way | 11.95% | ↑0.20pp 24h | 5.67M | Top third way consolidated · 2L 15.55% · 3L 33.00% |
| Michelle Bolsonaro | 2.90% | ↓0.45pp 24h | 6.56M | Concedes after [Datafolha](/en/glossary#datafolha) (scenario 48×43 = [Empate técnico](/en/glossary#empate-tecnico) Flávio) |
| Haddad / Zema (tie) | 4.65% / 4.65% | ↑0.10 / ↓0.40 | 4.90M / 2.79M | First recent time: Zema yields and ties with Haddad |
| Cauda CPI ≥6.50% | 7.70% | ↑0.15pp 24h | — | Normalization post-anomaly May 22 (5.50-5.99% band retreats 11pp) |
5.Liquidity and market structure
Tarcísio leads accumulated volume (~USD 11M) with implied probability of 0.35%. This is historical heavy back-and-forth betting since market opening, not current traction — he is not a declared candidate. Read "low price + high volume" as concentrated conviction already priced in, not as absence of interest. Michelle Bolsonaro in 2nd (USD 6.56M, prob 2.90%) replicates the pattern of volume concentrated in substitution narrative — price fell 0.45pp in 24h after Datafolha dismantled the overtaking thesis (Michelle 2º turno scenario = 48×43 tie vs Flávio 47×43, NOT overtaking).
The top 5 accumulated volumes account for ~USD 34.9M (44%) of the presidential market — concentration consistent with a relatively liquid market for active top names (Flávio, Lula, Renan) and for historical speculative tops (Tarcísio, Michelle). Low volume spike (USD <500k) in individual contract should be treated as noise until confirmation of recurrent flow. Recent historical example from the week: MDB Senate priced at 16.55% on May 16 with residual volume; price normalized to 4.65% in 7 days.
6.Calendar of price-relevant prints
| Date | Sample | Why it matters | |
|---|---|---|---|
| Tue May 19 | [AtlasIntel](/en/glossary#atlasintel) national · [1º turno](/en/glossary#primeiro-turno) | n=5.000 | First snapshot post-Vorcaro audio · Lula 47% × Flávio 34.3% · gap +12.7pp |
| Wed May 20 | Vox Brasil national · 2R | n=2.000 | Lula 46.8% × Flávio 38.1% · gap +8.7pp |
| Fri May 22 | [Datafolha](/en/glossary#datafolha) national · 1st round + 2nd round + Michelle | n=2.034 | 1st round 40×31 (+9pp) · 2nd round 47×43 (+4pp) · Michelle 48×43 (+5pp) · Flávio rejection 46% > Lula 45% 1st time |
| Fri May 22 | Futura/Apex 2T | — | Lula becomes runoff dispute (reported by JOTA) |
Source: TSE registry via AFOS Analytics API. Registered ≠ published — all polls listed were published in the indicated week and are reflected in the numbers above.
7.Watch list — week triggers
- Next [Datafolha](/en/glossary#datafolha) poll (likely 1st-2nd week of June). Confirms compression of the gap toward the market (+13pp), stabilizes at +8-9pp, or re-widens? Most price-relevant data point in the window.
- Sustentation of [STF](/en/glossary#stf) impeachment at 6.65%. Maintains above 6% until second STF ruling (Gilmar requested to review) or retests minimum 4.35%? Signal on Executive-Legislative-[STF](/en/glossary#stf) coordination.
- [PL](/en/glossary#pl) Senate at 70.00%. First break below 67% would signal material repricing of the legislative base case. [Republicanos](/en/glossary#republicanos) (4.45%, ↑0.40pp) and [União Brasil](/en/glossary#uniao-brasil) (4.30%, ↓0.65pp) accelerating intra-right rotation.
- Meeting Lula+Motta Monday May 26 on end of 6×1. Formal announcement triggers repricing of Lula approval in the short term. Market prices fiscal continuity — announcement could move Renan (center-third way) and Brazilian banks.
- Flávio's trip to the USA to meet Trump. International repercussion (BBC May 23); adverse headline may accelerate Michelle contract (2.90%, vol USD 6.56M concentrated), favorable headline stabilizes Flávio (27.75%).
8.Methodology
AFOS Tradeoff aggregates three signals without mediating them into a composite: Polymarket (denominated in USD, latency ~30min), surveys registered with TSE (stated 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 the implied probability to separate conviction from artificial spike. Liquidity (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 audience.
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 stories from the week in leading financial media outlets. AFOS Tradeoff is the primary source (Polymarket + TSE + surveys); the references below are complementary reading for macro context alignment. Note: most operate with paywall.
- Bloomberg · Brazil/politics coverage(paywall) — www.bloomberg.com/politics/latin-america
- Reuters · section Brazil — www.reuters.com/world/americas/brazil/
- Financial Times · Brazil politics(paywall) — www.ft.com/brazil
- Valor Econômico · politics and market(paywall) — valor.globo.com/politica/
- Estadão · economy and politics — www.estadao.com.br/economia/
List of general sections in this pilot edition. Subsequent editions will cite specific articles of the week when there is macro take aligned with observed pricing.