Tools chosen: polymarket_direct polymarket_related kalshi_related claude_news claude_news gdelt_news wikipedia code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a Polymarket election market on the 2026 Maine Senate race, where Republican incumbent Susan Collins (if she runs) faces a contested Democratic primary between Gov. Janet Mills and Graham Platner. The market price is the primary anchor; key drivers are Collins' retirement decision, the Democratic primary outcome, general-election polling, and the national midterm environment. I'll pull the Polymarket price, cross-check Kalshi/Polymarket related markets, and gather news/polling evidence.

# Current state The Democratic nominee is now **Troy Jackson** (selected via July 25, 2026 convention) after original primary winner Graham Platner withdrew July 10 amid sexual assault allegations; he faces five-term incumbent GOP Sen. Susan Collins in the Nov. 3, 2026 general election, conducted under ranked-choice voting. Nonpartisan raters (Cook, Sabato) call it a **Toss-up**; polling is roughly tied to slight Jackson lead, while prediction markets lean Democratic (~65-70%). # Timeline of key events - 2025-08: Graham Platner (D) announces Senate candidacy (confirmed, Wikipedia). - 2025-10: Gov. Janet Mills (D) launches Senate campaign (reported). - 2026-02-10: Susan Collins (R) confirms she will seek a sixth term (confirmed, Bangor Daily News). - 2026-04: Mills suspends her campaign but stays on primary ballot (reported, multiple sources). - 2026-06-09: Primary held — Collins is sole GOP nominee; Platner defeats Mills for Dem nomination (confirmed, Wikipedia). - 2026-07-10: Platner withdraws from race following rape allegation and prior controversies (Nazi-style tattoo, offensive posts) (confirmed, Wikipedia). - 2026-07-25: Troy Jackson (former state Senate president) selected as replacement Dem nominee at convention (confirmed, Wikipedia). - 2026-08-06/10: Fox News/Beacon poll: Jackson 48%, Collins 46% (reported). - 2026-08-14: Axios/Cook/Sabato: race rated Toss-up; Maine seen as Dems' best pickup chance (reported). - 2026-08-21 (approx): Kalshi-linked aggregator (predictionedge) shows Jackson ~64-66% vs Collins ~34% (reported). # Event Will the Democratic nominee (Troy Jackson) win the 2026 Maine U.S. Senate election against Susan Collins (R)? # Outcomes to forecast Yes (Democrat wins) / No (Democrat does not win) # Kalshi market anchor No direct kalshi_direct feed returned for this exact ticker. The **matching ticker on Polymarket** (identical ID, likely mirrored/same market) shows **YES = 70.5%**, up +1% (7d) and +3% (30d), range 51.5%-79% over 120 days, volume $722,542 — this is the best available direct anchor. A separate Kalshi-based aggregator (predictionedge) cites Jackson ~64-66% as of Aug 21, roughly consistent but somewhat lower than the 70.5% anchor. # Sub-question answers 1. **Collins reelection/approval** — Confirmed running for 6th term (announced 2026-02-10, BDN). No specific current approval number found; unfavorable ratings noted at 53% (June poll). 2. **Democratic primary** — Platner won June 9 primary over Mills (who suspended in April but stayed on ballot); Platner later withdrew July 10 after rape allegations; Troy Jackson replaced him via July 25 convention (Wikipedia, NBC). 3. **Head-to-head polling** — Highly volatile: March Emerson had both Dems leading Collins; June UMass Lowell had Platner 48-43; late June poll had Platner +2; July 1 Fox poll had Collins narrowly ahead; Aug 6-10 Fox/Beacon poll has Jackson 48-Collins 46; RCP average Jackson +2.3; Pollsmax average nearly tied (Collins +0.1). 4. **Collins polling-bias base rate** — In 2020, Collins trailed all public polls yet won by 8.6 points (12.6-pt miss vs. polling average); this is the key historical wildcard cited repeatedly by analysts (Wikipedia, Axios, code_execution model). 5. **National environment/2024 baseline** — Harris won Maine by ~7 points in 2024; environment described as "terrible for Republicans" nationally; Maine hasn't backed GOP presidential nominee since 1988 (Axios). This structurally favors Democrats but Collins has repeatedly overperformed the state's partisan lean. 6. **Third-party/RCV** — Race uses ranked-choice voting (confirmed, Wikipedia/2020 precedent). No significant independent/third-party candidates identified in current research; RCV applied but 2020 resolved without additional rounds since Collins won outright majority — no indication 2026 differs materially. 7. **Betting markets** — Polymarket (matching ticker): YES 70.5%. Kalshi-linked aggregator: Jackson ~64-66%. Model-based forecast (Pollsmax) gives Collins 61.9% win probability (Tilt Republican) — sharply diverging from market pricing. AI panel (oddsshopper) favors Collins 57-43. # Key facts (high-confidence, factual) 1. [Wikipedia] Collins seeking 6th term; sole GOP primary candidate; Jackson is Dem nominee since July 25 convention. 2. [Wikipedia] Platner withdrew July 10 after rape allegation; multiple prior controversies (tattoo, social posts). 3. [Axios/Cook/Sabato] Race rated Toss-up as of mid-Aug 2026; one of ~4 Senate toss-ups nationally. 4. [Wikipedia] Harris won Maine by ~7 pts (2024); Democrats haven't won a Maine Senate seat since 1988. 5. [Fox/Beacon poll] Jackson led 48-46 (Aug 6-10, 2026). 6. [SLF/GOP PAC] $42M committed by Senate Leadership Fund to defend Collins — signals serious GOP investment despite environment. # Cross-market signals - Kalshi related: Democratic Senate control markets show 55-57% for 2028 sweep/control scenarios (not Maine-specific). - Polymarket (same ticker): YES 70.5%, upward trend. - Other Kalshi-aggregator (predictionedge): Jackson ~64-66%, Collins ~34%. - Model-based (Pollsmax): Collins 61.9% win probability — contradicts market pricing. - AI panel (oddsshopper): Collins 57% vs Jackson 43% — also contradicts market. # Analyst opinions and speculation - NRSC memo warns against underestimating (former candidate) Democratic strength; Collins "has won tough races before." - Sabato/Cook: genuine toss-up; Democrats need Maine to have any realistic shot at Senate majority. - Multiple analysts flag Collins' 2020 polling-outperformance (+12.6 pt miss) as reason to discount current Jackson-favoring polls/markets. # Directional lean per outcome - **Yes (Dem win)**: Favorable state partisan lean (Harris +7), bad national GOP environment, current polling average slight Jackson edge, prediction markets (Polymarket/Kalshi-aggregator) both above 60%. - **No (Collins wins)**: Collins' strong personal brand/crossover appeal, historical polling-underestimation pattern (2020), $42M GOP investment, model-based forecasts (Pollsmax, AI panel) favor Collins, Democratic nominee (Jackson) is a late-entry replacement candidate with less vetting/name recognition than original nominee. # Gaps / unknowns - No true kalshi_direct quote returned for this exact ticker; anchor relies on Polymarket data with matching ID. - Limited fresh polling after mid-August; race dynamics after Jackson's convention nomination still stabilizing. - No detailed fundraising/name-ID data for Jackson vs. Collins. - Third-party/independent candidate filings not clearly detailed beyond RCV mechanism confirmation. # Calibration anchors - Polymarket/matching-ticker YES price: **70.5%** (primary anchor, upward trending). - Kalshi-aggregator alternate: ~64-66% Jackson. - Model-based counter-signal: Pollsmax ~38% Dem win probability (61.9% Collins). - Historical precedent: 2020 Collins won by 8.6 pts despite trailing every poll — supports meaningful downward adjustment from raw polling-implied Democratic lean.

1. Both forecasters anchor heavily on Polymarket's 70.5% and discount toward the low-60s, but neither adequately explains why they trust the market-implied Democratic lean over the model-based forecasts (Pollsmax 61.9% Collins, AI panel 57% Collins) — they simply "net" the two without a principled weighting, producing a somewhat arbitrary landing point (60% vs 63%) rather than a reasoned synthesis. 2. Neither forecast grapples with the magnitude of the Collins 2020 base-rate adjustment: a 12.6-point polling miss is enormous, and if applied literally to an Aug 6-10 Jackson+2 poll, it would flip the race decisively to Collins — yet both treat this as a modest few-point discount rather than seriously stress-testing a scenario where Collins wins comfortably (e.g., 55%+ probability of No). 3. Both largely ignore the nominee-instability risk quantitatively — Jackson was selected less than a month before these forecasts via convention (not primary), with essentially no general-election polling track record before Aug 6-10 and no fresh polling after mid-August; this "thin post-convention data" gap should widen uncertainty bands rather than support fairly confident 60-63% Yes calls. 4. Neither forecaster addresses the divergence between Polymarket (70.5%, trending up) and Kalshi-aggregator (64-66%) as evidence of market disagreement/liquidity differences — both just average toward the lower end without discussing which market is more reliable or why Kalshi (the actual event platform) might be more trustworthy than a "matching ticker" Polymarket proxy. 5. The $42M SLF investment in defending Collins is treated as a minor factor by both, but this level of spending on a single Senate race signals the GOP itself views this as genuinely competitive/losable, which arguably should be read as confirming toss-up status rather than simply another data point favoring "No." 6. Both forecasts converge suspiciously close together (60% vs 63%) despite citing nearly identical evidence, suggesting anchoring bias on the Polymarket price rather than independent reasoning — neither seriously entertains the AI panel/Pollsmax view that Collins is actually the favorite, which would justify a Yes probability meaningfully below 50%.

60% (53% confidence)

Yes: 60% No: 40%

The best direct anchors (Polymarket matching ticker 70.5%, Kalshi-linked aggregator 64-66%) favor a Democratic pickup, supported by structural fundamentals: Harris +7 in Maine in 2024, a poor national GOP environment, and Jackson leading 48-46 in the most recent (Aug 6-10) poll. I discount meaningfully below the market anchor because of three well-documented counter-signals the critique correctly emphasizes: Collins' extraordinary 12.6-point polling outperformance in 2020 (a documented, repeated pattern of crossover appeal in exactly this matchup type), model-based forecasts that actually favor Collins (Pollsmax 61.9%, AI panel 57%), and the thin, stale data on Jackson, a late convention replacement with limited general-election vetting and name ID. However, the critique's push toward Collins-as-favorite overreaches: the 2020 miss cannot be applied literally as a point estimate (it reflects one cycle with a presidential-year GOP surge and Collins' incumbency-plus-committee-chairmanship pitch), Cook/Sabato both rate this a genuine toss-up, and the $42M SLF defensive commitment signals Republicans themselves see the seat as losable. Netting a ~65-70% market against ~38-43% model signals with a modest additional haircut for nominee instability lands me just below both forecasters' midpoint at 60% Yes, with wide effective uncertainty.

Total pipeline time: 229.1s