Today's Biggest Divergences

ETH sits at ~$1,560 — only ~4% above the $1,500 strike — in an established downtrend, as the higher-beta asset in a BTC-led risk-off regime with no altseason bid.

September 2026 read, same blind basis and fact-check as the Lula pair (38 confirmed, 0 dropped; basis drafts/RESEARCH_brazil_2026_race_blind_2026-09-18.md).

September 2026 read, machine-blind, basis in drafts/RESEARCH_brazil_2026_race_blind_2026-09-18.md (fact-checked 9/18: 38 confirmed, 4 softened and applied, 0 dropped).

Ratings agencies call this race a tossup.

Democrats need a net of 4 seats on a map with only 2 Republican-held seats in genuinely competitive states (Maine, and the open North Carolina seat); the next tier (Ohio, Texas, Iowa, Florida) all went for Trump by 8 or more.

Iowa voted R+13 for president in 2024 and has reelected Republican governors by double digits; the venue is paying full price for one strong Democratic recruit (Auditor Rob Sand) before the general-election ad war, and open-seat races in states this red have closed toward the sta

Machine proposal, blind to the market: polls say a dead tie (Emerson early Sept Acton 46/45; Fox Aug R+2; aggregates RCP Acton +0.3 to R +0.8) — a tie in an R-lean state with the midterm penalty already priced.

⛔ EHIQ SAYS NO: Dems not as likely to take the House as the market thinks (−27.5pp) Polymarket prices Dem House control at 92.5% (volume $5.7M, as of 2026-09-26). EHIQ at 65% — 27.5pp below market. The midterm-incumbent-loss dynamic favors Dems, and Republicans hold a thin major

✅ SHOCKER: EHIQ sees Husted winning in Ohio (+18.5pp) NEW CALL 2026-09-27: Polymarket prices Jon Husted at 41.5% to hold Ohio's Senate seat against Sherrod Brown, and Kalshi's book agrees with the shape (Husted 43 cents on a $2.3M book). Our language engine reads the race the ot

Polymarket prices Dem Senate control at 62.5% (volume $2M+, as of 2026-09-27).

Machine proposal, blind to the market: a split-champion year — Spain won the WC but FIFA's awards went to Rodri (Golden Ball), Simón, Cubarsí, so voters leave the winning side (2010/2014/2018 pattern).

NEW CALL 2026-09-27: Polymarket priced David Eby's NDP at 50 cents at open (2026-09-27); the live leg now prints 36.5 cents — the market has crashed 13.5 points off the coin flip — while the September 24 Angus Reid poll has the Conservatives ahead 43-35.

RECONCILIATION 2026-10-03 (stamped desk:eric, "FIX BOTH and UPDATE BOTH"; TIQ/LAS-only constraint, v6 doctrine - NO SEM, NO polling): this row is CONFIRMED as the nomination read of record at 0.32 (range 30-38).

Machine proposal, blind to the market: generic ballot D+6.8 (9/11 aggregator average) converts to a 25-35 seat D House gain against a +3 threshold → P(D House) ≈ 0.82.

Cross-venue read, form table verified 2026-09-02: Spain won the World Cup (Yamal scored once in 8 matches); Kane delivered 73 club goals, the Bundesliga, the Pokal, the European Golden Shoe and 6 World Cup goals with England third.

September 2026 read. Wrapped scores a single calendar year, and the mid-year scoreboard is not close: Luminate all-platform YTD has Bad Bunny near 11.5B streams to Drake's 6.5B.

⛔ SHOCKER: EHIQ says El-Sayed far weaker than the 70.5% coronation price (−25.2pp) RESTATE 2026-09-27 (0.60 -> 0.453): we think El-Sayed is much weaker than Polymarket's 70.5% coronation price — and slightly weaker than even the toss-up polls. The market prices certainty; the po

FRESHENED 2026-10-03 (Eric-approved freshness rescore): the May register floor is RETRACTED as evidence.

The right-of-center math, not the vibe.

NEW CALL 2026-09-28 (market-aligned seat): Polymarket prices James Talarico at 64.5% to win Texas's open Senate seat — the market already disagrees with the ratings consensus (Lean R per Cook/Sabato), and the register reads the same direction from language alone: Talarico -0.04 m

Polymarket extremely low at 0.85%.

D leader by Polymarket pricing.

✅ SHOCKER: EHIQ says Turek is leading in Iowa (+19.5pp) NEW CALL 2026-09-27: Polymarket prices Josh Turek at 41.5% to flip Iowa's open Senate seat — the models lean Ashley Hinson (DDHQ 56%). The register reads the flip: across 8 spontaneous speeches each, Hinson reads defensive

Conditional on 2026 GA Senate retention (EHIQ 62% to retain per senate-ga-ossoff-2026 entry).

Machine proposal, blind to the market, fact-checked before stamping (19 claims: 11 confirmed, 6 softened, 0 dropped): the 2018 open-seat comp (Kemp won +1.4 under a D+8.6 wave with Trump at ~41%) adjusted for a worse 2026 environment puts R near 0.43; +~6pp for the nominee draw —

PA Governor (~13M state population), strong general-election profile in the most pivotal swing state.

Machine proposal, blind to the market: Sanders' 2026 screen is explicit (Medicare for All, big money, Gaza) plus the candidate must seek it.

EHIQ register-derived underpriced: Buttigieg has the structurally STRONGEST D primary register in the corpus — materially ahead of Newsom in side-by-side scoring.

RECONCILIATION 2026-10-03 (stamped desk:eric, "FIX BOTH and UPDATE BOTH"; TIQ/LAS-only constraint, v6 doctrine - NO SEM, NO polling): ehiq_prob 0.27 -> 0.16.

EHIQ register read shows mid-tier register — neither dominant nor weak.

STRUCTURAL / THESIS CALL.

TIGHTENED MAY 10 from 10% to 8% after primary linguistic analysis of Cohen's May 4-6 media tour.

Polymarket prices Chow at 77% (thin volume $28K — shallow consensus).

Polymarket prices Rubio at 15.15% — 3rd-highest behind Vance + Newsom.

↗️ EHIQ LEAN: Collins stronger than expected (+3.5pp) September 2026 rebuild for the Jackson matchup — the prior text tested a withdrawn candidate (Platner withdrew July 10; Troy Jackson nominated at the July 25 convention). The September polls say a true toss-up: UML's survey c

TonalityIQ language analysis flagged Newsom as the weakest of the top-tier 2028 Democrats months before the markets moved.

The market underprices Bad Bunny because it is metric-confused.

EHIQ register read DOMINATES the field — strongest primary-stage register of all 7 candidates scored, both parties.

EHIQ register scoring of Biden's Hur interview language identified collapse markers 4 months before mainstream polls reflected the trajectory.

A Fed HIKING into a 20-30% risk-asset drawdown, record crypto-ETF outflows, a chip-led equity selloff, AND a +57K July payroll miss is historically rare and self-limiting — tightening financial conditions plus a cracking labor market usually forces a pause, not a hike.

Retrospective validation: EHIQ register scoring on Cameron / Boris / Farage public language identified LEAVE outcome 59 days before the referendum, when Betfair / Predictit had REMAIN at 82%.

EHIQ prices a 2026 World Cup final halftime appearance well above the Kalshi market, which has faded the name on rumors of a lineup snub that were subsequently reported as false.

EHIQ prices a 2026 World Cup final halftime appearance well above the Kalshi market, which has faded the name on rumors of a lineup snub that were subsequently reported as false.

Polymarket now ~62%, matching EHIQ — 12pp gap CLOSED pre-resolution (May 12).

EHIQ has Nathan Turek as the favorite (~66%) to win the Iowa Democratic Senate primary, against a market (Kalshi) that has him around 35%.

TIQ/LAS debate leg (4 verified French debates, two-ways controlled) reads the four-way opposition split HOLDING: PSPP zero blame / zero core-claim doubt / desperation only as return fire; Milliard the field clearest deterioration; Frechette partial recovery only.

Pezeshkian was elected in mid-2024 and is still in early stages of his presidency with no credible removal mechanism in play.

EHIQ reads El-Sayed a modest underdog (~42%), not the market's ~65% favorite.

Polymarket prices Cepeda at ~42% (Vol $28M).

Polymarket prices Cepeda at ~42%.

QUESTION-SEMANTIC RESOLVED 2026-05-22: this market resolves on first-round / primary first-place finish (consistent with Polymarket sub-market structure where candidate probs sum to ~1 across the field, not the 0.5 D-vs-R split a general-only market would produce).

QUESTION-SEMANTIC RESOLVED 2026-05-22: reframed from "win the office" to "first-round / primary first-place finish" — matches Polymarket sub-market structure (probs sum to ~1 across field).

EHIQ 18% vs market 35%, -17pp below.

Polymarket prices Espriella at ~42% (Vol $28M).

QUESTION-SEMANTIC RESOLVED 2026-05-22: LARGEST MOVE IN THIS UPDATE — prior EHIQ 9% reflected "win the office" (deep-blue CA Republican constraint).

Kalshi over-priced Drake to ~18-29% on the May triple-album pulse — a decaying, US/Anglo-weighted volume spike — while Bad Bunny's sustained global tour and reigning Global Top Album make him the structural favorite for the full-year global metric.

Pebbles argues private credit's problem was never the returns, it was that nobody could see inside.

Polymarket prices Espriella at ~40% ($28M volume).

EHIQ holds Musk at 30% to win, 10pp below Kalshi current of ~40%.

Polymarket prices Valencia at ~20%.

We sit above the market's 72%.

QUESTION-SEMANTIC RESOLVED 2026-05-22: reframed from "win the election" to primary first-place finish.

EHIQ called Radev correct on the morning of April 19, before exit polls and before Polymarket fully priced it.

Polymarket prices Lula at 40.5% (Vol $67.8M, real consensus).

Polymarket prices GameStop acquiring eBay at 20%.

Day 250 (April 26) Embassy Toronto video called Kalshi 16% on a Kendrick mention as overpriced.

Polymarket prices Flávio Bolsonaro at 41.6% (Vol $67.8M, real consensus).

Polymarket prices Chong at 92.5% near-lock ($36M volume — deep market, real consensus).

Polymarket prices Oh at 7.5% — EHIQ at 15% (+7.5pp).

We expect the firing to stand and Humphrey's Executor to be narrowed for ordinary independent agencies.

Polymarket prices López at 0.1% (likely she hasn't formally declared candidacy yet — market is dormant).

QUESTION-SEMANTIC RESOLVED 2026-05-22: reframed from "win the election" to primary first-place finish.

Polymarket prices Valencia at ~16-24% (Vol $28M).

Iceman is already 2 weeks at #1 (first multi-week #1 for Drake since FATD).

Wildcard, revised DOWN May 11 after corpus deepening.

QUESTION-SEMANTIC RESOLVED 2026-05-22: this market resolves on CA jungle-primary first-place finish (Polymarket sub-market probs sum to ~1 across the 5-candidate field — Becerra 0.49 + Steyer 0.335 + Hilton 0.097 + Porter 0.014 + Mahan 0.022 — which is first-place semantics, not

QUESTION-SEMANTIC RESOLVED 2026-05-22: reframed from "win the office" to first-round / primary first-place finish.

Machine proposal, blind to the market: 2/9 outright R1 wins since 1989 (both FHC mid-90s); no leader polling ≤43% of total votes at ~3 weeks has ever closed to >50% of valid votes (2002 Lula ~46% valid failed; 2022 Lula ~48% failed).

QUESTION-SEMANTIC RESOLVED 2026-05-22: reframed from "win the election" to primary first-place finish.

CONSENSUS-CONFIRMING (not a fade).

Day 258 video + X thread posted 2026-05-04, 11 days before Iceman's May 15 release: public point-forecast of 525,000 first-week US equivalent album units (revised UP from earlier 250K call after methodology refinement).

TIQ language model on the BoC Governing Council statement (latest ingested = 2026-03-18) reads HOLD at 2026-06-04 with raw confidence 0.766.

CONDITIONAL CALL: this resolves only if Tucker Carlson declares as a 2028 R primary candidate with serious organization (donors, early-state infrastructure).

AI analysis on April 18 (Substack) called Mark Mulroney as the better and more likely successor to Pierre Poilievre as CPC leader.

Odet trades for less than its 71%-owned Bolloré SE stake alone.

BRZE is in the EMJ Capital short book on the AI-SaaS-incumbent decline thesis (Mar 31 piece).

Decay-model build from actual W1 (463K) + W2 (225K) prints, applying Iceman's better-than-typical retention curve and the Chartmetric MAU uplift (89.5M → 99.0M monthly Spotify listeners in 27 days, global rank #7 → #3, listener-to-follower ratio 0.81 → 0.885 toward the streaming-

The conservative majority has been telegraphing readiness to overturn Humphrey's for years (Seila Law 2020 distinguished it; Collins v Yellen 2021 narrowed it further).

Principal-agent thesis: $1.18B of BCRED fees flowing to Blackstone Inc.

Sheeran precedent (May 22 2026): Ed Sheeran exited Warner after 15 years via amicable distribution-only structure — catalog stays at Warner, go-forward via Gingerbread Man + ADA.

6-month post-IPO performance is highly path-dependent and the 10pp-outperformance-vs-QQQ threshold is a non-trivial hurdle.

AI CEO comparison work (tonality_iq/analysis/AI_CEO_COMPARISON) tracks Altman's vision-mode Invincibility Triad and the structural unsustainability of the current MSFT-OpenAI partnership terms.

Lock-up expiry typically creates 5-10% downward pressure on close-day (Snap -9%, Coinbase ~-7%, Lyft -9%) but breaking BELOW IPO price 180 days out requires either a fundamentally disappointing 6-month run OR a structurally massive secondary supply hit.

Bookrunner press leak target is $1.75T IPO valuation.

Pebbles I-III walked the structural argument: BCRED's 0.6% non-accrual rate is implausibly low given the 5.3%->7.2% PIK trajectory over 4 quarters and 25% software-book exposure (7pp above peer mean) facing AI disruption.

Extension of the May 26 v18 MIXED->STRESS print.

Day 274 video + X thread (2026-05-20).

Conservative majority has consistently sided with executive immigration authority (Trump v Hawaii 2018, DHS v Thuraissigiam 2020, Biden v Texas 2022).

The EHIQ v18 gate moved to risk-ON today, 2026-05-20, with a model score of 0.619 — above the 0.5 ON threshold, into the B→C interpolation region of the EMJX Gen2.1 v1.3 allocation matrix.

Pop > 20% is the modal mega-cap-tech-IPO outcome of the last decade (Snap +44%, Airbnb +112%, DoorDash +86%, Rivian +29%, Coinbase intraday +47% / close ~+13% vs ref, Reddit +48%) but not the dominant outcome (Facebook ~+0%, Uber -7%, Lyft +9%).

EHIQ holds 75% the Second Circuit AFFIRMS Judge Jeannette Vargas's 38-page October 9 2025 opinion dismissing Drake's defamation claim against UMG under Rule 12(b)(6).

We expect the Court to split the removal-power question rather than decide it as one.

FIVN contact-center cloud is the most acutely AI-disrupted business in the 5-short cohort.

JPMorgan analyst Daniel Kerven (Nov 25 2025 Vivendi note) decomposed the Paris Court of Appeal ruling into three scenarios: 35% court upholds the prior "had control" finding (mandatory tender at fair value ~€3.9/share, +56% VIV upside); 25% Bolloré wins and does nothing; 40% Boll

On July 31 AM Best revised the outlook to negative from positive and affirmed the Financial Strength Rating of A- for Delaware Life, Clear Spring Life and Annuity and Gainbridge Life.

First Pebble Substack thesis (Medallia 60c) predicted distribution coverage breakdown + structural stress in BDC space.

Drake 'Iceman' Week 4 EAU forecast: greater than 120K.

Warner and Bain stood up a joint vehicle of up to $1.2B for catalog acquisitions, and reporting attaches the Chili Peppers catalog to it as the first deal out of the gate.

Catalyst-sourced. Kill trigger: A termination 8-K, a withdrawn proxy, or a second-request/injunction that pushes the outside date past the resolution date.

Stamped on IPO day with SPCX indicated ~$200 pre-open (gray markets $180-208, ~$2.3-2.6T implied).

Pre-print thesis called all directional markers correctly: 28-DPD record, $300M new buyback, raised FY26 guide, CashAI v5.5 structural credit improvement.

STRUCTURAL / THESIS CALL (not TIQ- or engine-derived; the framework is unbacktested and this dated call is the test).

FSK is in the EMJ Capital short book (puts) on the Pebbles thesis.

EHIQ register scoring of Carney's public language pre-PM and immediately post-PM identified specific concession points.

Drake Spotify Monthly Listeners 100M crossing forecast.

This is the call we'd point you at first, because the coverage has the wrong company.

On August 18 the company filed notice that TWG Global, Walter's holding company, agreed to buy up to $6.5B of affiliated assets from Delaware Life and hand back an equal amount of unaffiliated assets.

The first correction took affiliated investments from 3% of the book to 42% at year-end 2025, on AM Best's own description.

AUTO-SOURCED from the FDA catalyst calendar.

Catalyst-sourced. Kill trigger: A termination 8-K, a withdrawn proxy, or a second-request/injunction that pushes the outside date past the resolution date.

Quality core, same thesis (identity/doc auth + deepfake scoring, Fraud & Identity +28% YoY, buyback).

The league has hit every public milestone on this project: exploration announced March 2025, team-engagement process December 2025, Silver in Berlin January 2026 ('very much on schedule'), final bids due end of June 2026.

The NFL called a one-day meeting in Atlanta on 2026-08-26 to vote the Khosla group's $9.612B purchase, a league record.

The Seahawks just repriced the league.

Apr 10 Kaz Nejatian backchannel signaled cohort margin curves work in progress (3 of EHIQ's 3 suggestions).

The money runs out December 31.

VRA Section 2 has been progressively narrowed by Roberts Court (Shelby 2013, Brnovich 2021); Allen v Milligan (2023) was a surprise upholding but narrow.

The board voted to explore bids in Las Vegas and Seattle exclusively, with a final vote signalled for later in 2026.

Walter's 2 life insurers carry roughly $19.8B of investments in his own affiliates, restated onto that line after February 2026 grand jury subpoenas, and Bloomberg reports the Delaware Department of Insurance wants nearly all of it unwound by 2026-12-31.

Two valuation regimes run on the same sponsor's borrowing.

STRUCTURAL / DATED CALL, made ahead of any market — the existing SC-nominee markets are stale artifacts of the resolved June 9 Graham primary; no live special-primary contract exists yet.

Drake Spotify Monthly Listeners 100M crossing forecast.

Last week the open question was the vote.

Apple confirmed the event itself on August 26: September 9, 'Surprise and shine.' The two best-sourced Apple reporters both say the foldable leads it.

Catalyst-sourced. Kill trigger: A termination 8-K, a withdrawn proxy, or a second-request/injunction that pushes the outside date past the resolution date.

Judge Vargas dismissed in October 2025, holding the 'Not Like Us' statements nonactionable opinion.

Seventh consecutive hold is the overwhelming consensus (Reuters 35/35; CORRA OIS ~95-99%) and the data support it both ways the Bank would want: CPI ticked UP to 3.0% (July, gasoline/Hormuz), so no cover for a cut; Q2 GDP ~3.4% annualized and unemployment 6.4% falling, so no urge

AUTO-SOURCED from the FDA catalyst calendar.

Walter's core platform debt is distressed - Guggenheim Partners' first-lien TLB marked 100.0805 at 30 Jun 2026 by 3 registered funds, 98.1250 at 31 Jul, and 73.25 on Bloomberg BVAL at 25 Sep, with direct observations at 100% weight.

CERO required monthly working-capital advances at consolidation ($775,665 funded 2026-08-27, more scheduled), so repayment of up to $11.67M plus interest inside 6 weeks requires an external financing event, not operations.

AUTO-SOURCED from the FDA catalyst calendar.

EHIQ's AdCom model reads REPL as reject-leaning (~70%).

The pick. AI made content abundant -> verified provenance/identity is the scarce good (the inversewire/manufactured-consensus story is the demand proof).

Highest torque on the exact catalyst (co-authored the C2PA watermarking standard; Art 50 recommends C2PA).

To be precise about the holding, because most coverage is loose with it.

AUTO-SOURCED from the FDA catalyst calendar.

AUTO-SOURCED from the FDA catalyst calendar.

EO 14160 attempts to deny birthright citizenship to children of unauthorized immigrants and certain temporary visa holders by executive order.

Catalyst-sourced. Kill trigger: A termination 8-K, a withdrawn proxy, or a second-request/injunction that pushes the outside date past the resolution date.

Market anchor (0.842, as-of 2026-10-06) + base-rate leg (R1 leader 7/7 in runoffs since 1989) with a DECLARED -4.2pp haircut for cascade-assumption risk (post-R1 extension 28.7pp->68.7pp prices landslide momentum + minor-candidate flows + bias band partly self-priced).

Catalyst-sourced. Kill trigger: A termination 8-K, a withdrawn proxy, or a second-request/injunction that pushes the outside date past the resolution date.

Peters won 2020 by 1.5%.

September 2026 read, desk call seeded by the register instrument and checked against the venues before writing.

Bloomberg AltD Similarweb showed Q1 web traffic +14.08% YoY (R²=53%, 7/8 directional, 100% of period) vs consensus revenue +7.32% — implied beat of +6.76pp.

Roberts Court has steadily struck down campaign finance limits (Citizens United 2010, McCutcheon 2014, Cruz 2022).

v18 already prints STRESS as of late May 2026.

AUTO-SOURCED from the FDA catalyst calendar.

The public comment window closes 2026-09-04, after which a federal judge must find the consent judgment in the public interest under the Tunney Act.

OpenAI's major-version gaps have run 12 to 15 months (GPT-4 March 2023, GPT-4o May 2024, GPT-5 August 2025); September 2026 sits 13 months after GPT-5, inside the cadence.

Both sides' incentive to renew is gone.

EHIQ's AdCom model reads CAPR as approve-leaning but fragile (~60%).

Tillis won 2020 by 1.7%.

Stamped 2026-08-05 with OPEN common around $3.79.

EHIQ +26pp upstream of Polymarket-implied.