Forecast
Projection from round-1 votes: the headline emulates how the 4 October votes transfer to the 25 October runoff, state by state. Polls and the prior appear for comparison only.
Win probability · projection from round-1 votes
Left · Lula
47%
Right · Flávio Bolsonaro
53%
National share of valid votes
Left: median 49.60% · 80% interval: 43.16–56.04%
Right: median 50.40% · 80% interval: 43.96–56.84%
Axis from 35% to 65%; center tick = 50%.
Updated on 10/19/2026 · freezes on 24 Oct 2026 at 23:59 BRT·model 2026.1
Statistical forecast based on official TSE results and on polls registered with the TSE; it is not an opinion poll (TSE Resolution 23.600/2019).
How the number is computed
Transfer model (round-1 votes)
49.67%·50.33%- Baseline: proportional split of round 1 between the two finalists, by state (48.90% for the left nationally).
- Adjustment for each bloc’s eliminated candidates (β left -0.09 · β right 0.22): how much each side’s eliminated share moves the result beyond the proportional split.
- Abstention drift by state: expected runoff turnout weights each state in the national total.
- Uncertainty: the historical between-round swing, 4.77 pts national standard deviation, plus regionally correlated state error, over 20,000 simulations.
Calibrated on the 2002, 2006, 2010, 2014, 2018, 2022 runoffs (168 state × year observations). · sd of year effects α_y (between-round national swing) floored at 2.5 pp
Comparison (not part of the number)
These two references are shown alongside the projection but do not feed it.
Polls
- Raw average (left · right)
- 48.89% · 51.11%
- House-effect-corrected average
- 47.18% · 52.82%
- Polls
- 8
Selected institutes only: Datafolha, AtlasIntel, BTG/Nexus, Quaest, MDA/CNT.
Polls registered with the TSE; figures transcribed from the institutes’ releases (per-row source in polls.json).
Incumbency prior
51.27%·48.73%
5 previous runoffs, σ 8.31 pts.
Scenarios
Variations of the base model; definitions in the methodology
Base
Left
47%
Right
53%
- Left
- 49.60% (43.16–56.04)
- Right
- 50.40% (43.96–56.84)
High abstention in the Northeast
Left
45%
Right
55%
- Left
- 49.34% (42.89–55.77)
- Right
- 50.66% (44.23–57.11)
Strong transfer from eliminated candidates
Left
37%
Right
63%
- Left
- 48.28% (41.83–54.71)
- Right
- 51.72% (45.29–58.17)
By state
Color by the gap between expected medians (right − left) in each state.
Margin in percentage points of valid votes (right − left).
| State | Left | Right | 80% (left) | P(left) | Weight | Round 1 |
|---|---|---|---|---|---|---|
| RR Roraima | 24.77% | 75.23% | 17.47–32.03% | 0% | 0.26% | left 22.86% · elim. left 2.76% · elim. right 3.32% |
| SC Santa Catarina | 28.09% | 71.91% | 20.88–35.38% | 0% | 3.77% | left 25.04% · elim. left 3.16% · elim. right 5.14% |
| RO Rondônia | 28.22% | 71.78% | 21.04–35.53% | 0% | 0.78% | left 25.89% · elim. left 2.74% · elim. right 3.91% |
| AC Acre | 31.21% | 68.79% | 24.01–38.46% | 0% | 0.36% | left 28.73% · elim. left 3.00% · elim. right 3.71% |
| MT Mato Grosso | 31.40% | 68.60% | 24.21–38.65% | 0% | 1.66% | left 29.18% · elim. left 2.12% · elim. right 3.55% |
| PR Paraná | 35.13% | 64.87% | 27.90–42.46% | 0% | 5.50% | left 31.20% · elim. left 3.28% · elim. right 5.61% |
| MS Mato Grosso do Sul | 37.73% | 62.27% | 30.47–44.92% | 2% | 1.22% | left 34.68% · elim. left 2.77% · elim. right 3.94% |
| GO Goiás | 39.49% | 60.51% | 32.19–46.69% | 3% | 3.15% | left 31.06% · elim. left 1.50% · elim. right 13.84% |
| RS Rio Grande do Sul | 39.96% | 60.04% | 32.72–47.22% | 4% | 5.38% | left 35.73% · elim. left 3.21% · elim. right 5.42% |
| ES Espírito Santo | 41.43% | 58.57% | 34.13–48.83% | 7% | 1.88% | left 37.76% · elim. left 3.04% · elim. right 4.42% |
| SP São Paulo | 43.30% | 56.70% | 36.00–50.66% | 12% | 20.97% | left 38.20% · elim. left 3.86% · elim. right 6.01% |
| RJ Rio de Janeiro | 43.34% | 56.66% | 36.06–50.72% | 12% | 7.68% | left 39.41% · elim. left 2.88% · elim. right 4.69% |
| DF Distrito Federal | 44.03% | 55.97% | 36.75–51.27% | 14% | 1.46% | left 38.11% · elim. left 2.81% · elim. right 7.77% |
| TO Tocantins | 46.78% | 53.22% | 39.46–54.05% | 28% | 0.76% | left 43.42% · elim. left 2.48% · elim. right 3.65% |
| MG Minas Gerais | 48.00% | 52.00% | 40.74–55.35% | 36% | 10.12% | left 43.33% · elim. left 3.58% · elim. right 4.85% |
| AP Amapá | 50.55% | 49.45% | 43.33–57.82% | 54% | 0.36% | left 45.71% · elim. left 4.26% · elim. right 4.36% |
| AM Amazonas | 52.00% | 48.00% | 44.75–59.28% | 64% | 1.73% | left 48.23% · elim. left 3.70% · elim. right 3.07% |
| ZZ Exterior | 53.05% | 46.95% | 45.83–60.36% | 71% | 0.27% | left 47.57% · elim. left 3.56% · elim. right 5.38% |
| PA Pará | 53.18% | 46.82% | 45.97–60.51% | 71% | 3.93% | left 49.91% · elim. left 2.69% · elim. right 2.90% |
| AL Alagoas | 57.78% | 42.22% | 50.50–65.00% | 92% | 1.54% | left 54.73% · elim. left 2.45% · elim. right 2.37% |
| RN Rio Grande do Norte | 63.48% | 36.52% | 56.24–70.78% | 99% | 1.75% | left 59.75% · elim. left 2.88% · elim. right 2.60% |
| PB Paraíba | 65.17% | 34.83% | 57.92–72.43% | 100% | 2.18% | left 61.31% · elim. left 3.05% · elim. right 2.57% |
| CE Ceará | 67.25% | 32.75% | 60.03–74.47% | 100% | 4.76% | left 63.29% · elim. left 2.61% · elim. right 2.83% |
| PE Pernambuco | 67.50% | 32.50% | 60.30–74.79% | 100% | 4.83% | left 63.45% · elim. left 2.55% · elim. right 2.97% |
| SE Sergipe | 67.67% | 32.33% | 60.43–75.03% | 100% | 1.16% | left 62.75% · elim. left 2.97% · elim. right 3.65% |
| MA Maranhão | 67.85% | 32.15% | 60.58–75.08% | 100% | 3.33% | left 63.99% · elim. left 2.21% · elim. right 2.90% |
| BA Bahia | 70.19% | 29.81% | 63.02–77.49% | 100% | 7.39% | left 66.17% · elim. left 2.38% · elim. right 2.91% |
| PI Piauí | 74.98% | 25.02% | 67.73–82.20% | 100% | 1.81% | left 70.99% · elim. left 2.32% · elim. right 2.61% |
Weight = state share of national valid votes. Round 1: share of the left finalist and of each bloc’s eliminated candidates.
Polls registered with the TSE — for comparison
Polls are not part of the headline. The table lists those used in the comparison average.
| Institute | Field dates | n | TSE registration | Valid (left · right) | House effect | Weight |
|---|---|---|---|---|---|---|
| AtlasIntelto verifysource | 10/13/2026 – 10/18/2026 | 4,998 | BR-02418/2026 | 49·51 | +3.41 pts | 29.86% |
| BTG/Nexusto verifysource | 10/16/2026 – 10/18/2026 | 2,002 | BR-04499/2026 | 50·50 | +3.43 pts | 15.20% |
| MDA/CNTto verifysource | 10/13/2026 – 10/15/2026 | 2,002 | BR-00652/2026 | 50·51 | +2.83 pts | 13.56% |
| Datafolhato verifysource | 10/13/2026 – 10/15/2026 | 2,520 | BR-00466/2026 | 49·51 | +3.96 pts | 12.99% |
| Quaestto verifysource | 10/10/2026 – 10/13/2026 | 2,154 | BR-08729/2026 | 49·51 | +3.26 pts | 10.01% |
| BTG/Nexusto verifysource | 10/10/2026 – 10/12/2026 | 2,002 | BR-07997/2026 | 48·52 | +3.43 pts | 6.62% |
| AtlasIntelto verifysource | 10/03/2026 – 10/08/2026 | 5,026 | BR-03663/2026 | 47·53 | +3.41 pts | 7.49% |
| Datafolhato verifysource | 10/06/2026 – 10/07/2026 | 2,520 | BR-02949/2026 | 48·52 | +3.96 pts | 4.28% |
Polls registered with the TSE; figures transcribed from the institutes’ releases (per-row source in polls.json).
Other registered polls (not used) — 7
| Institute | Field dates | n | TSE registration | Valid (left · right) |
|---|---|---|---|---|
| PoderData/Ayato verifysource | 10/16/2026 – 10/18/2026 | 3,000 | BR-02586/2026 | 48·52 |
| Vox Brasilto verifysource | 10/15/2026 – 10/17/2026 | 2,100 | BR-04332/2026 | 51·49 |
| Futura/100% Cidadesto verifysource | 10/16/2026 – 10/17/2026 | 2,000 | BR-06715/2026 | 50·50 |
| Ipespe/Abrapelto verifysource | 10/13/2026 – 10/15/2026 | 1,500 | BR-00223/2026 | 49·51 |
| Futura/100% Cidadesto verifysource | 10/14/2026 – 10/15/2026 | 2,000 | BR-03764/2026 | 49·51 |
| Futura/Apexto verifysource | 10/07/2026 – 10/10/2026 | 2,000 | BR-00650/2026 | 48·52 |
| Ipespe/Abrapelto verifysource | 10/06/2026 – 10/08/2026 | 1,500 | 47·53 |
Registered with the TSE but outside the comparison average (tier-2 institutes).
Backtest
Leave-one-year-out validation using transfer + prior only — there are no historical polls in the repository — so the error is wider than the full ensemble’s.
| Year | Predicted (median) | 80% interval | Actual | Error | P(left) | In interval |
|---|---|---|---|---|---|---|
| 2002 | 66.89% | 60.29–73.39% | 61.27% | +5.62 | 100% | ✓ |
| 2006 | 53.07% | 48.27–57.71% | 60.83% | −7.76 | 79% | ✗ |
| 2010 | 57.29% | 50.28–64.22% | 56.05% | +1.24 | 91% | ✓ |
| 2014 | 54.85% | 47.63–61.95% | 51.64% | +3.21 | 80% | ✓ |
| 2018 | 36.48% | 30.74–42.08% | 44.87% | −8.39 | 0% | ✗ |
| 2022 | 53.96% | 47.11–60.73% | 50.90% | +3.06 | 77% | ✓ |
4 of 6 years inside the 80% interval.