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 the September 2026 Berlin state election. The key drivers are current Berlin state polling (AfD historically weak in Berlin, ~15%, vs CDU ~25% and Linke/Greens strong), national AfD momentum, and how much polls can move in ~9 months. I'll anchor on the Polymarket price, pull cross-market signals, and gather polling/news evidence.

# Current state Berlin's Abgeordnetenhaus election is confirmed for 20 September 2026 (set by Berlin Senate decision, 3 June 2025). As of the most recent polling (through 22 Aug 2026), Die Linke leads most surveys (21–22%), with CDU second (19–20%) and AfD third (16–19%); only one poll (Insa/Nius, 23 Jul 2026) shows AfD tied/ahead. AfD winning the *most seats* — the resolution criterion — currently looks unlikely based on mainstream survey consensus, despite a well-documented AfD national polling surge. # Timeline of key events - 2023-02-12 (confirmed): Berlin repeat election — CDU wins with 28%, SPD 18.4%, Grüne ~18.4%, Linke 12%, AfD 9% (upswing from prior). CDU-SPD coalition formed under Mayor Kai Wegner. [Wikipedia] - 2025-06-03 (confirmed): Berlin Senate formally sets 20 Sept 2026 as the next Abgeordnetenhaus election date. [Wikipedia/berlin.de] - 2026-06-07 (reported): AfD pushing gains in district elections nationally. [DW] - 2026-07-05 (reported): German defense minister warns of possible far-right state government (national context, not Berlin-specific). [Straits Times] - 2026-07-10 (reported): Chancellor Merz repositions as "reformer" to counter far-right rise. [Politico] - 2026-07-23 (reported, outlier): Insa/Nius poll shows AfD first at 19%, tied CDU/Linke at 18% each — notable outlier from right-leaning commissioning outlet. [t-online] - 2026-07-20/08-03 (reported): Civey/Tagesspiegel poll — Linke 21%, CDU 19%, AfD 18%, Grüne 17%. [t-online] - 2026-07-31 (reported): AfD hits all-time Politbarometer national high, largest-ever lead over CDU (national, not Berlin). [Investmentwatchblog] - 2026-08-07/13 (reported): Insa/t-online poll — Linke 21%, CDU 19%, AfD 18%, Grüne 17%. [t-online] - 2026-08-21/23 (reported): AfD surging in East German state polls (Saxony-Anhalt), framed as stepping stone to federal power — national/other-state trend cited as bullish signal for AfD generally. [Yahoo, bankingnews.gr] - 2026-08-22 (confirmed, most recent): Forsa/t-online poll — Die Linke leads "by a distance," followed by CDU and AfD roughly level for 2nd/3rd. [t-online] # Event Will AfD win the most seats in the 2026 Berlin state (Abgeordnetenhaus) elections, held 20 Sept 2026? # Outcomes to forecast Yes (AfD most seats) / No (any other party/coalition most seats) # Kalshi market anchor No Kalshi-direct price was returned in raw research (kalshi_direct tool not present in this pull; kalshi_related returned only irrelevant markets — UK, US, Canada elections). **Primary usable cross-market anchor is Polymarket**: current price 10.5% YES for AfD, down from a 28% high, -6% (7d) and -11.8% (30d) trend, on $2.38M volume — a persistent downward drift. # Sub-question answers 1. **Polymarket prices** — AfD "most seats" market: 10.5% current, trending down (-11.8% 30d). CDU/Greens/SPD/Linke-specific prices were not retrieved in this pull (gap); polymarket_related found no additional matching markets. 2. **Latest Berlin polls (Jul–Aug 2026)** — Die Linke leads nearly all mainstream polls (21–22%), CDU second (19–20%), AfD third (16–19%), Grüne fourth (16–17%), SPD ~13%. Only the Insa/Nius poll (23 Jul) shows AfD in a share of first place, treated as an outlier given the commissioning outlet's political lean. [t-online, multiple polls] 3. **2023 election / 2025 federal trend** — 2023 Berlin repeat election: CDU 28%, SPD 18.4%, Grüne ~18.4%, Linke 12%, AfD 9% (up from prior cycle). No specific 2025 federal-election-in-Berlin AfD figure was retrieved in this research (gap), though national AfD polling has since surged strongly (2nd-largest Bundestag party post-2025). 4. **Historical polling movement 9 months out** — No empirical historical dataset was retrieved; the Monte Carlo model assumed sd=3–7pp independent noise as a plausible proxy for German state-poll drift, without citing specific historical precedent (gap/assumption, not sourced fact). 5. **Postponement/annulment risk** — No indications found of any risk to the 20 Sept 2026 date; election date reconfirmed by official Senate decision and Wikipedia. Low risk (Berlin already experienced an unusual repeat-election precedent in 2023, but no signals of recurrence noted). 6. **Berlin-specific shocks favoring AfD** — Coverage of AfD's political shocks (Politbarometer peak, Saxony-Anhalt surge, "path to federal power") is overwhelmingly national/other-state, not Berlin-specific; Berlin polling has not shown AfD leading except in the one outlier poll. No CDU-SPD governing collapse or major Berlin scandal reported. # Key facts (high-confidence, factual) 1. [Wikipedia/berlin.de] Election date confirmed: 20 Sept 2026. 2. [Wikipedia] 2023 Berlin election: CDU 28% (1st), AfD 9% (5th of 5 major parties). 3. [t-online, multiple, Jul–Aug 2026] Linke leads current polling (21–22%), CDU 2nd (19–20%), AfD 3rd (16–19%), one outlier poll shows AfD tied/1st. 4. [Polymarket] AfD "most seats" priced at 10.5%, declining trend, high volume ($2.38M). 5. [Wikipedia] No two-party coalition would have a majority under current polling; three-party coalitions (CDU-Grüne-SPD or Linke-Grüne-SPD) seen as realistic outcomes — implying AfD would need outright plurality, not just governing role. # Cross-market signals - Kalshi related: no directly relevant markets found (irrelevant UK/US/Canada tickers only). - Polymarket: 10.5% YES, falling from 28% high over past ~4 months — market has moved decisively away from AfD. - Sportsbook implied: none available. # Analyst opinions and speculation - German media (t-online) frame the race as a "close head-to-head" among top 3 parties, but consistently place AfD third, not first. - National AfD surge narratives (Politbarometer, Saxony-Anhalt) are cited as reasons for concern but not directly tied to Berlin's own polling lead order. - Insa/Nius outlier suggests AfD's ceiling may be underestimated in mainstream polls, but multiple other institutes (Forsa, Insa/t-online, Civey) contradict this. # Directional lean per outcome - **Yes (AfD wins most seats)**: Supported by national AfD momentum, one outlier poll (Insa/Nius) showing AfD tied/ahead, historical volatility in Berlin (2023 repeat election precedent). Opposed by: consistent multi-pollster leads for Linke and CDU over AfD by 3–6pp margin, declining Polymarket price, Monte Carlo estimate (~1–9%, base ~2.6%). - **No (other party wins)**: Strongly supported by nearly all mainstream polling (Linke/CDU lead), declining prediction-market price, structural difficulty of AfD overturning multi-point deficit in 9 months without unprecedented events. # Gaps / unknowns - No Kalshi-direct YES price retrieved (event's own primary market data missing from this pull). - Per-party Polymarket prices (CDU/Greens/SPD/Linke) not retrieved. - No confirmed 2025 federal election AfD-in-Berlin vote share. - No sourced historical polling-error/movement dataset for German state elections. # Calibration anchors - Polymarket YES price (best available cross-market anchor): 10.5%, trending down. - Monte Carlo synthetic estimate: ~2–9% depending on volatility/correlation assumptions. - Precedent: AfD has never led a Berlin state poll cycle by a wide margin; 2023 result was 9% (5th place).

1. Both forecasts converge tightly (7% and 9%) around the Polymarket anchor, but neither addresses the discrepancy with the brief's own Monte Carlo estimate (~2.6% base, 1–9% range) — Forecast 2 explicitly rejects moving toward that lower estimate without justifying why polling-error risk should dominate over the synthetic model's tighter distribution. 2. Neither forecaster grapples with the "close time" being 20 Sept 2026 while polling data cited runs only through 22 Aug — nearly a month of potential campaign movement (debates, late-breaking scandals, tactical voting shifts) is treated as low-risk without justification, especially given AfD's noted national momentum trend. 3. Both treat the Insa/Nius outlier as clearly dismissible due to "right-leaning outlet" bias, but neither considers the alternative explanation flagged in the brief itself: "AfD's ceiling may be underestimated in mainstream polls" — a house-effect/shy-voter argument that could apply symmetrically to Berlin as it does nationally, where AfD has hit record highs. 4. Neither forecast addresses seat-allocation mechanics separately from vote-share polling — Germany's mixed-member system with overhang/leveling seats could produce a seat outcome diverging from a close 3-way vote-share race (e.g., a 2-3pp popular vote gap could still translate into AfD placing first or second in seats depending on direct-mandate wins), yet both forecasts reason purely in vote-share terms as if seats = votes. 5. Both anchor almost entirely on Polymarket without discussing its known limitations for low-probability/thin-comparable political contracts (potential manipulation, non-representative trader base, $2.38M volume is not verified as deep/liquid relative to the specific "most seats" contract) — no independent stress-test of the anchor's reliability was performed. 6. The brief notes "no CDU-SPD governing collapse or major Berlin scandal reported," but neither forecaster explicitly prices in tail scenarios (e.g., a late scandal hitting Linke or CDU, differential turnout collapse among rival voters) that could disproportionately benefit AfD as the campaign's momentum party — both distributions look like point-estimate extrapolations rather than accounting for late-race volatility skew toward the trending party.

7% (73% confidence)

Yes: 7% No: 93%

Berlin polling through late August consistently places AfD third (16–19%) behind Die Linke (21–22%) and CDU (19–20%), with only the right-leaning Insa/Nius poll as an outlier; AfD must overtake two parties in under four weeks in its structurally weakest large-state terrain (9%, 5th place in 2023). The critique's seat-mechanics point is largely inapplicable: Berlin's Abgeordnetenhaus uses a proportional system with leveling seats, so seat rank closely tracks vote rank above the 5% threshold, and AfD wins few direct mandates in Berlin. House-effect/shy-voter arguments and residual late-campaign volatility (plus national AfD momentum) justify staying above the ~2.6% Monte Carlo base rather than collapsing to it, but the falling, reasonably liquid Polymarket price (10.5%, down from 28%) plus multi-institute polling agreement argues for a value slightly below the market anchor. I settle at 7% YES, modestly below Polymarket to reflect the short remaining horizon and consistent polling order.

Total pipeline time: 291.2s