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 election question on the 2026 Maine Senate race, where GOP incumbent Susan Collins faces a competitive Democratic field (Janet Mills, Graham Platner). The market price is the primary anchor, and the key research is candidate status (Collins retirement vs. running), primary dynamics, polling, and national midterm environment. Cross-venue checks on Kalshi and related Polymarket markets (Senate control, Maine primary winner) give independent signals.

# Current state Susan Collins (R) is confirmed running for a sixth term and was unopposed in the GOP primary. Graham Platner won the June 9 Democratic primary (72%) after Janet Mills withdrew, but Platner suspended his campaign July 10 amid a rape allegation and other scandals; Troy Jackson (former state Senate president, 3rd place in the gubernatorial primary) was chosen as replacement nominee at a July 25 convention. The general election is Nov 3, 2026, using ranked-choice voting; Maine hasn't elected a Democratic senator since 1988, but is the only state Trump has never won held by a GOP senator. # Timeline of key events - 2025-08: Platner announces Senate candidacy (confirmed, Wikipedia). - 2026-03: Emerson poll shows Platner/Mills both leading Collins, especially among women (reported). - 2026-05-27: UNH Survey Center poll — Platner leads Collins (reported). - 2026-06-04/09: UMass Lowell poll shows Platner 48%–43% over Collins; Platner wins Democratic primary with 72%, Mills having already withdrawn (confirmed). - 2026-07-10: Platner suspends campaign following rape allegation from ex-girlfriend and prior controversies (tattoo, offensive posts); Schumer, Gillibrand, Maine Dem Party call for withdrawal; DSCC/Senate Majority PAC pull support (confirmed, multiple outlets). - 2026-07-25: Maine Democratic Party nominating convention selects Troy Jackson as replacement nominee (confirmed, Wikipedia/CBS). - 2026-08-13 to 08-25: Platner makes first public appearances (labor rally); ongoing media commentary on scandal fallout and Jackson's viability (reported). # Event Will the Democratic nominee win the 2026 Maine U.S. Senate election (vs. Republican incumbent Susan Collins)? # Outcomes to forecast - Yes (Democrat wins) - No (Republican/Collins wins) # Kalshi market anchor No direct Kalshi price returned in this research; Polymarket is the closest live anchor: **69.5% YES** (Dem win) as of latest snapshot, down 1pt over 7 days, up 2pts over 30 days, range 51.5%–70.5% over 90 days, $732K volume. Note: this price range spans the pre- and post-Platner-scandal period, so the current level likely already reflects Jackson's nomination as the replacement candidate. # Sub-question answers 1. **Polymarket price/trend** — 69.5% currently; 90-day range 51.5%-70.5%, +2pts/30d, -1pt/7d [polymarket_direct]. The dip to 51.5% likely corresponds to the peak of Platner scandal uncertainty; recovery to ~70% suggests market views Jackson as a credible if lesser nominee, or believes fundamentals (Collins' vulnerability) dominate. 2. **Collins running?** — Yes, confirmed; she is the only Republican to qualify for the GOP nomination, seeking a sixth term [Wikipedia]. 3. **Primary leader** — Platner won the primary (72%) after Mills withdrew, but Platner then withdrew from the general due to scandal; Troy Jackson (3rd in gubernatorial primary, labor-backed) is now the actual Democratic nominee [Wikipedia, CBS, Fox]. Pre-scandal head-to-heads favored Platner (48-43) and Mills over Collins [UMass Lowell, Emerson]. No general-election polling of Jackson vs. Collins found in research. 4. **General election polls / 2020 error** — Available polls (UNH, UMass Lowell, Emerson) all predate Platner's withdrawal and Jackson's nomination; none test Jackson vs. Collins. Research does not address 2020 Maine polling error explicitly, though it's widely known Collins outperformed final polls by ~8-10 points in 2020 — not confirmed in this research set (gap). 5. **National midterm environment** — Not directly addressed in research; not found (gap). 6. **Base rate for cross-pressured incumbents** — Historical Senate incumbent loss rate ~9.5% (1990–2022, n=220) [code_execution/historical aggregation]; incumbents in "cross-pressured" states (opposite presidential lean) face ~2.5-3.5x elevated loss risk (~24-33% central ~28.6%), which would suggest a Dem win probability of ~67-76% if the race is just "generic cross-pressured incumbent" — but this analysis doesn't account for the Jackson downgrade. 7. **Related markets** — Kalshi has no direct Maine Senate market; related 2028 Senate-control markets (Dem Sweep 55%, Dem Senate control 56%) are not directly comparable/informative for this race [kalshi_related]. No Polymarket related markets found for Maine Senate primary/Collins specifically [polymarket_related]. # Key facts (high-confidence, factual) 1. [Wikipedia] Collins unopposed in GOP primary; seeking 6th term. 2. [Wikipedia/CBS/Fox] Platner won Dem primary, withdrew July 10 after rape allegation; Troy Jackson named replacement nominee July 25. 3. [Wikipedia] Maine is the only state held by a GOP senator that Trump has never won; Dems haven't won a Maine Senate seat since 1988. 4. [UMass Lowell, Emerson] Pre-scandal general-election polling favored Platner/Mills over Collins; Collins had 57% unfavorable/38% favorable rating. 5. [Cook Political, via claude_news] Cook's Senate editor: "virtually impossible" for Dems to retake Senate majority without winning Maine; post-scandal framing now leans Collins-favored against a weaker Jackson. # Cross-market signals - Kalshi related: No direct Maine Senate market found; adjacent 2028 Senate-control markets show Dem-leaning odds (55-56%) but are not race-specific comparables. - Polymarket: 69.5% YES for Dem win — this is the primary cross-market signal and appears to already price in the Jackson substitution given its 90-day trajectory (low of 51.5% during scandal, recovery to ~70%). - Sportsbook implied: None found. # Analyst opinions and speculation - Cook Political Report (via claude_news): views Jackson as lacking Platner's "candidate skills," calls a Jackson win "a real upset" — implies Collins now favored, in tension with the 69.5% Polymarket price. - Sabato's Crystal Ball piece titled "Maine Senate Do-Over" suggests rating changes reflecting new uncertainty post-scandal. - code_execution model (using assumed/possibly stale 0.63 price, pre-dating full confirmation of Jackson nomination): scenario-weighted estimate ~54.6% (range 46-63%), below market; cross-pressured-incumbent base-rate model implies 67-76% — the two approaches bracket 55-70%, roughly consistent with current 69.5% Polymarket price but with real disagreement among methods. # Directional lean per outcome - **Yes (Dem win)**: Supported by structural blue-lean of Maine, Trump never carried it, Collins' weak favorability, elevated cross-pressured-incumbent base rate (~67-76% per one model), and current Polymarket price (69.5%) recovering post-scandal. - **No (Collins wins)**: Supported by Cook Political's post-scandal Collins-favored framing, Jackson's perceived weaker candidate skills/profile vs. Platner, lack of general-election polling validating Jackson's strength, Collins' proven history of overperforming polls (2020 precedent, though not directly sourced here), and her long incumbency/moderate brand. # Gaps / unknowns - No Jackson vs. Collins general-election polling found. - 2020 Maine polling-error precedent not directly sourced in this research (widely known but unconfirmed here). - National generic ballot / presidential approval data not retrieved. - No live Kalshi YES price for this specific ticker was returned — Polymarket used as best available anchor. - code_execution model used an outdated/hypothetical 0.63 price and pre-Jackson scenario assumptions, reducing its reliability for current calibration. # Calibration anchors - Polymarket YES price (anchor): 69.5%, recovered from a 51.5% scandal-period low, +2pts/30d. - Historical Senate incumbent loss base rate: ~9.5% generic; ~24-33% for cross-pressured incumbents (opposite presidential-lean state) [code_execution]. - Precedent: Collins has won reelection five times, including 2020 despite unfavorable polling, in a state Democrats otherwise dominate statewide.

1. Both forecasters shade down from Polymarket (69.5%) to ~63-64% based on the same qualitative Cook Political framing, but neither explains why they didn't shade further given Cook's explicit language calling a Jackson win "a real upset" — this is a strong directional signal that seems only partially incorporated relative to its stated strength. 2. Neither forecast grapples with the tension flagged in the brief itself: Cook's framing "is in tension with the 69.5% Polymarket price" — if Cook (a sharp professional handicapper) thinks Collins is favored, that implies the market itself may be stale or thin ($732K volume, no confirmed recent trade reflecting Jackson's actual general-election viability), yet both forecasters treat the market as fully "pricing in" the substitution. 3. Both forecasters cite the 2020 Collins overperformance as evidence for shading toward Collins, but the brief explicitly flags this as "not confirmed in this research set (gap)" and only "widely known" — treating an unsourced/unconfirmed claim with the same weight as confirmed facts is a calibration risk neither forecaster acknowledges. 4. Neither forecaster addresses the national midterm environment sub-question, which the brief marks as "not found (gap)" — in a real 2026 midterm, generic ballot and presidential approval could meaningfully move a cross-pressured race, and this omission is unaddressed by both rather than flagged as a source of uncertainty. 5. The two forecasts are nearly identical (0.63 vs 0.64) despite using different confidence levels (0.38 vs 0.54) and slightly different reasoning chains — this convergence suggests anchoring on each other's likely output or on the same Polymarket figure rather than independent triangulation of the base-rate model (67-76%) vs. the scenario-weighted model (~55%), which actually bracket a much wider range than either forecast reflects. 6. Neither forecaster discusses ranked-choice voting dynamics explicitly, despite the brief noting Maine uses RCV — a third-party or independent candidate could siphon votes in ways that interact unpredictably with a weakened Jackson candidacy, a scenario neither considered.

62% (46% confidence)

Yes: 62% No: 38%

Polymarket's 69.5% is the only live anchor and its recovery from a 51.5% scandal-period low suggests Jackson's substitution is at least partly priced, but the market is thin ($732K) and may lag the qualitative reassessment. Structural fundamentals still favor the Democrat: Maine is the only GOP-held Senate seat in a state Trump never carried, Collins sits at 38/57 favorability, and cross-pressured incumbents lose at roughly 25-33% versus a 9.5% generic rate — plus a second-term Trump midterm environment historically tilts toward the out-party. Offsetting, Cook explicitly calls a Jackson win 'a real upset,' no Jackson-vs-Collins polling exists, and Collins has a long record of surviving unfavorable polling (2020 most notably, though unconfirmed in this brief). The base-rate model (67-76%) and scenario model (~55%) genuinely bracket a wide range; I land at 62%, shading a bit further below the market than both forecasters to give Cook's directional signal and the candidate downgrade more weight, while not fully discounting the structural case.

Total pipeline time: 165.5s