Tools chosen: polymarket_direct polymarket_related kalshi_related claude_news gdelt_news wikipedia code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a Polymarket question about whether AfD wins the most seats in Berlin's September 2026 state election. The key empirical inputs are current Berlin state-level polling (AfD is historically weak in Berlin relative to eastern states, with CDU and Linke/Greens strong), the trend in AfD support since 2025, and the Polymarket price plus any cross-venue signals. Berlin's proportional system means 'most seats' tracks vote share closely.

# Current state Berlin's 2026 Abgeordnetenhaus election is set for 2026-09-20 (~6 weeks out as of the latest polling in this research, early August 2026). No party has a stable lead: CDU and Die Linke alternate first place across pollsters, with AfD surging from its 2023 result (9.1%) into the high teens but leading in only one outlier poll (Insa, commissioned by a right-leaning outlet). "Leading in a single poll" is distinct from "winning the most seats" — the resolution criterion requires an actual election-night seat plurality. # Timeline of key events - 2023-02-12 (confirmed): Berlin repeat election — CDU wins with 28.2%, AfD 9.1% (Wikipedia, statistiken-aktuell.de). - 2025-02 federal election Berlin second votes: not surfaced in raw research (gap — see below). - 2026-06-05 (reported): AfD Bundestag members push renewed Russia ties (national-level, not Berlin-specific; Yahoo). - 2026-06-18 (reported): National poll shows AfD 9-point lead over CDU nationally (Zerohedge) — national, not Berlin. - 2026-07-15 (reported): Chancellor Merz reaches out to AfD voters ahead of "pivotal regional election" (Reuters/Internazionale). - 2026-07-16 (confirmed): Infratest dimap/RBB poll — Linke 22%, CDU 20%, Grüne 17%, AfD 16%, SPD not specified. - 2026-07-23 (reported): CDU nominates Stefan Evers as new lead candidate, described as "right-wing hardliner" (WSWS). - 2026-07-23 (confirmed): Insa/Nius poll — AfD's best-ever Berlin result, 19% (ahead of CDU/Linke tied at 18%), Grüne 16%, SPD 13%. - 2026-07-27 (reported): Berlin Pride attack sparks fears of far-right exploitation; officials call for tighter extremism laws. - 2026-08-02 (rumored/fringe): Sott.net op-ed alleging "election rigging" against AfD — unreliable source, treat as speculation. - 2026-08-05 (reported): >1,000 lawyers call for AfD party ban nationally (Breitbart) — not Berlin-specific, low near-term resolution likelihood. - 2026-08-05 (confirmed): Civey/Tagesspiegel poll — Linke 21%, CDU 19%. # Event Will AfD win the most seats in the 2026 Berlin Abgeordnetenhaus election (Sept 20, 2026)? # Outcomes to forecast Yes (AfD most seats) / No (another party, most likely CDU or Linke, wins most seats) # Kalshi market anchor No direct Kalshi price returned for this ticker in raw research; kalshi_related search found no matching AfD/Germany/Berlin market. **Use Polymarket as the best available cross-market anchor: 17.25% YES**, 7-day +2.3%, 30-day -5.3%, range 11.55–28% over 91 days, $2.35M volume — moderate confidence, some recent softening from a June/July high. # Sub-question answers 1. **Latest polls**: Multiple July–Aug 2026 polls: Infratest dimap (Linke 22, CDU 20, Grüne 17, AfD 16); Insa (AfD 19, CDU/Linke 18, Grüne 16, SPD 13); Civey (Linke 21, CDU 19); early-July aggregate (CDU 20.5, AfD 18, Grüne 17.1, Linke 16.1, SPD 14.2). [claude_news] 2. **Gap**: AfD trails the leader by 0–5pp depending on pollster; only Insa (right-leaning-commissioned outlier) shows AfD ahead. 12-month trend: AfD up from 9.1% (2023) to high teens, but never leading in the two most established institutes (Infratest dimap, Civey). [claude_news] 3. **2023/2025 baselines**: 2023 Berlin election: CDU 28%, AfD 9.1% [Wikipedia]. Feb 2025 federal election Berlin second votes: not found in raw research — gap. 4. **National AfD momentum**: National polls show AfD with a large (~9pt) lead over CDU nationally (June 2026), plus ban-debate agitation (1,000 lawyers petition, Aug 2026) and a Berlin Pride attack fueling far-right narratives — these create tailwinds but Berlin-specific polling still lags national AfD strength. [gdelt_news] 5. **Election date risk**: Sept 20, 2026 date is formally set by Berlin Senate and confirmed by Wikipedia/berlin.de; no postponement signals found. Risk of delay past Dec 31, 2026 appears negligible. 6. **Polymarket pricing**: AfD YES = 17.25% (current); no explicit CDU/Linke/Greens/SPD sub-market prices were returned, only the AfD contract. # Key facts (high-confidence, factual) 1. [Wikipedia] 2023 election: CDU 28.2%, SPD 18.4%, Greens ~18.4%, Linke 12%, AfD 9.1%. 2. [Wikipedia/berlin.de] Election date confirmed as 2026-09-20. 3. [claude_news] AfD has never led in Infratest dimap or Civey Berlin polls through Aug 2026; led once in Insa (Jul 23, 2026). 4. [Polymarket] Current AfD YES price 17.25%, down from a 30-day high of 28%. # Cross-market signals - Kalshi related: no direct or comparable Kalshi market found. - Polymarket: 17.25% YES, declining from 28% peak over past month — market has cooled on AfD chances despite the one favorable Insa poll. - Sportsbook implied: none available. # Analyst opinions and speculation - Monte Carlo model [code_execution] using illustrative CDU~22/AfD~19 gap estimates AfD win probability ~30% (range 21–52% depending on assumed gap), noting gap size is the dominant driver (each 1pp shift ≈ 5-6pp probability change). - WSWS frames CDU's new candidate (Evers) as a "right-wing hardliner," implying CDU may compete for AfD-adjacent voters, potentially suppressing AfD ceiling. - National ban-petition and Pride-attack narratives are speculative accelerants/decelerants, not yet reflected in Berlin-specific poll shifts. # Directional lean per outcome - **Yes (AfD)**: Supporting — AfD momentum from 9% to high-teens, one poll showing outright lead, strong national AfD polling. Opposing — AfD trails in 2 of 3 most-cited pollsters, Berlin AfD structurally weaker than national average (west/east and urban dynamics), Polymarket pricing near 17%, model central estimate ~30% but with wide uncertainty band still favoring "No." - **No (other party)**: Supporting — CDU/Linke consistently alternate lead across most recent polls, 2023 precedent of CDU dominance, Polymarket pricing implies ~83% for "not AfD." Opposing — narrowing gap and one AfD-led poll show race is tightening. # Gaps / unknowns - No Feb 2025 federal-election Berlin second-vote figures retrieved (would sharpen swing analysis). - No Kalshi-specific price for this ticker was returned; anchor relies on Polymarket. - No polling after Aug 5, 2026 captured; six weeks remain before the election — meaningful movement possible. - Coalition/seat-conversion nuances (5% threshold dropouts for FDP/BSW) could affect seat allocation disproportionately to vote share. # Calibration anchors - Polymarket AfD YES: 17.25% (primary cross-market anchor, since no Kalshi price found). - Historical precedent: AfD has never led statewide polling averages in Berlin (west German-leaning capital), unlike its dominance in eastern states (Thuringia, Saxony). - Quantitative model central estimate: ~30% (range ~20-40%) based on current CDU/Linke–AfD gap and historical polling error volatility.

1. Both forecasts anchor almost entirely on Polymarket without flagging its low reliability for this niche market ($2.35M volume, no Kalshi comparator, 91-day range 11.55–28%) — neither discusses how thin/volatile prediction markets can misprice single-country regional elections, which warrants wider uncertainty bands than a tight 15-22% cluster. 2. Neither forecaster meaningfully engages with the Monte Carlo model's wide range (21-52%) or explains why they dismiss the upper tail so heavily; Forecast 1 asserts the model "likely overweights" the gap but doesn't quantify why 30% central estimate should collapse to 15% rather than, say, 20-25%. 3. Both forecasters underweight the six-week window for polling movement explicitly flagged in the brief ("No polling after Aug 5, 2026 captured; six weeks remain before the election — meaningful movement possible") — given AfD's rapid rise from 9% to high-teens in ~3 years and accelerating national momentum, tail scenarios where AfD continues climbing (e.g., from Pride-attack backlash or ban-petition backlash effects) deserve more explicit modeling rather than a passing mention. 4. Neither addresses the seat-allocation nuance flagged in the brief: "5% threshold dropouts for FDP/BSW could affect seat allocation disproportionately to vote share" — a multi-party fragmentation scenario could let AfD win a seat plurality with a lower vote share than CDU/Linke, which is a structurally important pathway to Yes that's mentioned in the brief but not incorporated into either probability. 5. Forecast 1's reasoning that "AfD needs to overtake both CDU and Linke simultaneously" somewhat conflicts with Forecast 2's point that "a plurality could be won with under 25% in a fragmented multi-party race" — this is a genuine analytical disagreement between the two forecasters on how hard the plurality threshold actually is, and neither reconciles it against the actual current standings (Linke 21-22, CDU 19-20, AfD 16-19), where the gap to overtake is only 2-6pp, not requiring AfD to leapfrog by a wide margin. 6. Both largely ignore the confirmed asymmetry in polling institute quality (Insa's outlier vs. Infratest dimap/Civey as "most-cited/established") without discussing potential herding or systematic pro-establishment bias in the established pollsters — if AfD has a history of under-polling (as seen in other European contexts), the "never led in two most-cited pollsters" framing may still understate true support, an angle neither forecaster interrogates despite the brief noting "historical polling error volatility" as a named model input.

19% (medium confidence)

Yes: 19% No: 81%

The Polymarket anchor (17.25%, cooling from a 28% peak) and Berlin-specific polling both lean No: AfD sits at 16-19% and trails Linke (21-22) and CDU (19-20) in the two most established institutes, leading only in the Insa/Nius outlier. Berlin is structurally AfD's weakest terrain (west-leaning, urban, left-tilted), and AfD has historically polled at or slightly above its actual Berlin result (2023: ~10% polled, 9.1% actual), which cuts against the 'systematic under-polling' argument. The devil's advocate is right that a thin market and six weeks of campaign time justify wider uncertainty and some weight on the Monte Carlo upper tail — a 2-6pp gap in a fragmented five-party field is genuinely surmountable, and Berlin's proportional seat allocation means a vote plurality translates fairly directly into a seat plurality (so no big extra hurdle, but also no free path via threshold dropouts). I therefore set Yes modestly above the market anchor and below the model's ~30% central estimate, at 19%, reflecting real but sub-even-odds tail risk of continued AfD momentum.

Total pipeline time: 288.1s