Research

The SSRN paper formalizes the combined NLP + financial signal framework behind EventHorizonIQ. 173,000+ earnings transcripts analyzed. Calibration set includes SVB and six other failed banks. Each detection documented with timestamped language patterns and the financial confirmation signals that followed. The framework generalizes beyond banking to fraud-adjacent subjects, CEO coached-denial structures, and multi-quarter conviction arcs.

e.l.f. Beauty acquired Hailey Bieber's Rhode for a "$1 billion" headline. The SEC filing tells a sharper story: $896.5M in total consideration ($589.1M cash + $300.3M stock / 2,582,371 e.l.f. shares + a $7.1M-fair-valued earnout), against ~$212M of trailing net sales at signing - a ~4.7x revenue multiple, in roughly three years, on about ten products. The number the press release didn't carry: under ASC 805, e.l.f.'s FY2026 10-Qs disclose Rhode's standalone post-close results - $128.2M net sales and $54.8M net income in the December 2025 quarter alone (~43% NET margin), $180.6M / $69.9M since the August 5 2025 close. A ~40%+ net margin is the financial signature of an inverted cost structure: owned distribution substituting for a paid customer-acquisition function. The mechanism, per Tribe Dynamics: ~$248M of 2024 earned media value, +367%, #1 skincare - with only ~15% of it from posts that even named the founder. The community is the engine, not the ad budget. The structural turn: a brand can only realize that value by selling; an artist can hold it. Once an audience is owned rather than rented, the legacy middle layer - in beauty the marketing apparatus, in music the label - stops being a counterparty with leverage and becomes a vendor. The same architecture that produced a 4.7x exit in three years is the thing that decides what a record label is worth. The filing put a number on owned distribution; the music business has been carrying the same option, mostly unpriced, for a decade.

Part 2 of the Iceman series. Drake released what amounts to an engineered exit filing from Universal Music in the same window Bill Ackman put a ~$64B unsolicited bid on the company. The music press is covering the first as a rap story; the markets desks are covering the second as a French-governance story. They are one story. This piece maps the control structure of a $64B asset - Drake to UMG to Vivendi to Vincent Bolloré, with Ackman's bid, the Vivendi/Bolloré/Tencent voting bloc, and the CIAM/AMF activist-regulatory stack as the forces on the kingmaker. The core: the headline '~43% voting bloc' is far narrower than the math implies - UMG's own September 2021 listing prospectus (Section 12.11) shows a dividend-policy plus board-seats consultative pact whose stated function is the Dutch mandatory-takeover-offer exemption, a construct for avoiding a forced bid, not seizing control, and it is on a clock. Layered on top: a live French control-law case (Cour de Cassation overturned and remanded, November 2025) that could force the kingmaker to sell. The trade: everyone is bidding to own this building; nobody is bidding to run it differently - and the insurgent bid keeps the incumbent operating model.

Drake dropped three albums simultaneously on May 15 - ICEMAN (18 tracks), MAID OF HONOUR (14), HABIBTI (11). The internet is counting Kendrick shots. Read all 43 tracks the way you read a 10-K and a different document appears. ICEMAN is not a comeback album - it is an exit filing. Across one of the three albums Drake lays out a sequenced case against Universal Music Group: the grievance ("battle the stations 'til my ass is back in rotation"; "hundred million streams vanished"; a literal recoupment-accounting bar), the reframe ("I'm fightin' the man, not suin' the rapper, you boys is not listenin'" - he tells you the Kendrick story is a costume), the target ("October's Own is the label"; "check signin' is my kink, pushin' out ink" - owner, not talent), the mechanism ("private arbitration leads to resolution"), and his own prediction of the ending ("they'll act like I lost my appeal but they'll pay me for changin' they mind... they'll frame it as people retired"). And the bar that points to the top of the building: "I'm the golden goose, shakin' things up at Lucian's house." The structural proof it is engineered, not catharsis: every business bar is in ICEMAN only. All 14 MAID OF HONOUR tracks and all 11 HABIBTI tracks carry zero label content - verified track by track, including the one suspect title that resolved clean. You do not quarantine a feeling. You quarantine a plan. The triple-drop functions as a contract-fulfillment vehicle. What this is NOT: a claim Drake is contractually free. His remaining album count is confidential; delivering albums does not return the back catalog, erase recoupment, or end the live litigation - which is precisely why the lyrics point to arbitration and a quiet settlement. Drake is the single most valuable active asset in UMG's catalog. An artist that size laying out a public, sequenced exit case - in the same window as an active investor-governance fight over the same company - is an asset-risk and governance event, not a hip-hop headline.

Live Nation reported Q1 2026 on May 5: revenue up 12%, AOI up 9%, 85% of 2026 large venue shows already booked, 107M tickets sold YTD up 11%. Michael Rapino on the call: "We've seen no demand pullback anywhere. Broad, strong demand across the board on all genres, all theater sizes." Joe Berchtold added that 30% of tickets are under $50 and two-thirds under $100, even on stadium tours like Shakira, BLACKPINK, The Weeknd, Coldplay, and Stray Kids. StubHub, the secondary ticketing marketplace that IPO'd at $23.50 on September 17, 2025, has lost 73% of its market value since peak. Its first 10-K disclosed material weaknesses in internal controls so significant that PwC's critical audit matter is explicitly titled "Impact of Controls Related to Financial Reporting" - the auditor performed manual testing because the company's controls weren't operating. A securities class action was filed November 24, 2025 in SDNY. The 10-K also disclosed an Andro Capital / Colloquy Capital related-party arrangement, where the founder's own affiliate holds $7.9M in security interest against seller proceeds on the StubHub platform, up from $0.1M a year earlier. Eric Baker's Pay versus Performance Compensation Actually Paid for 2025: negative $575M. The founder is bleeding equity in real time, locked in by 88% voting control via dual-class shares with a Madrone Partners and Bessemer Venture Partners side letter requiring their approval of any successor. Brandon Ross at LightShed Partners had the cleaner read in private exchange last week: the blue dots are brokers, not buyers. The broker-mediated secondary marketplace economics broke; brokers can't move inventory at the prices they paid. Live Nation's response: hire Saumil Mehta as Ticketmaster President in November 2025. Mehta spent 10 years at Square as Chief Product Officer. His Investor Day numbers: blocked one million high-risk accounts with only 3% passing subsequent identity verification, bots blocked monthly grew from 1B in 2022 to 20B in September and October 2025, 9,000 tickets pulled back from broker accounts in one recent on-sale and returned to fans. The Saumil hire is not a routine executive appointment. It is the structural disintermediation mechanism. Drake releases Iceman this Friday May 15 on streaming. No physical. No tour announcement. The streaming-only release is the rational arbitrage of exactly that collapse.

Yesterday I made the EHIQ predictions dashboard public for the first time at eventhorizoniq.com/predictions - 39 calls paired with the prediction market price at the time we made each one. Some calls are upstream of consensus by single digits. A handful are upstream by twenty-plus percentage points. This piece walks through the five biggest gaps. (1) Buttigieg beats Newsom head-to-head: EHIQ 40% vs Polymarket-implied 12% at the time of the call. Field-wide pricing rewards operational advantages - fundraising, brand recognition, donor network; head-to-head primary mechanics reward different things - early-state retail track record, debate-stage performance, general-electability profile. (2) Tucker beats Vance, conditional on declaration: EHIQ 40% vs Polymarket 13%. The structural path constraint dominates the unconditional probability - most of the time, Tucker does not declare. Conditional on declaration with serious organization, his rhetorical command of the post-Trump base reshapes the head-to-head. (3) Drake top Spotify 2026: EHIQ revised 38% to 20% same-day vs Kalshi 2%, after Kalshi crashed on H1 streaming data showing Bad Bunny's structural Latin streaming dominance. Same-day public revision with what-I-got-wrong attribution. Iceman drops Friday May 15. (4) Musk v Altman jury verdict (the first short on the list): EHIQ 30% vs Kalshi 40%. The November 2017 Brockman diary entry that Judge Gonzalez Rogers cited in her January 15 ruling sending the case to trial is material but not a smoking gun. Defense procedural posture has been confident throughout. (5) House Democrats 2026: EHIQ 65% vs Polymarket 77.5%. Directional bet intact, magnitude wrong. The 2025-2026 mid-decade redistricting cycle has materially extended Republican map advantages - Texas +5 seats, Florida +4, North Carolina +1, with up to 14 additional seats projected across TX/FL/MO/NC/OH/TN. Mix of two LONG (Buttigieg, Tucker conditional), one calibration (Drake), two SHORT (Musk, House Dems) - the dashboard does not have a single-direction bias. Eight resolved calls so far, all directional hits, average 40.5 percentage points upstream of consensus markets at the time of the call (Brexit 2016, Biden Hur-era collapse 2024, Carney USMCA 2025, Bulgaria-Radev 2026, plus four 2026 housing and credit earnings).

The first sponsor-backed software loan above $2 billion has been publicly equitized at zero recovery for the equity holder. Bloomberg's Preeti Singh reported this afternoon that a Blackstone-led group of private credit lenders is preparing to inject at least $100 million of fresh capital into Medallia and convert a large portion of its $2.8 billion loan into equity, wiping out the roughly $5 billion of equity Thoma Bravo and its co-investors poured in since taking Medallia private in 2021. Apollo and KKR are named as co-lenders. Buried in the article: Blackstone valued the loan at 60 cents on the dollar as of the end of March and placed the debt on non-accrual status. That sentence is the entire story. Medallia is not unique. It is representative of the 2020-22 software buyout vintage, underwritten at 6-8x revenue multiples and 6-8x EBITDA leverage with no headroom for revenue compression and no consideration of generative AI as a category-level threat. Apollo's own parent 10-Q, filed yesterday, tells the same story at the asset-class level: alternative net investment earned rate of 5.79% in Q1 2026 versus 10.08% in Q1 2025 - a 4.3 percentage point compression in a single year. Alternative income down 33%. Spread Related Earnings down 11%. Net investment spread compressed 31 basis points. These are the marks already moving in the parent 10-Q. They are the asset-class-level visible expression of what name-level events like Medallia produce when they aggregate across the book. FS KKR Capital, Apollo Debt Solutions, OBDC, and the Blackstone credit vehicles will report Q1 2026 marks over the next thirty days. They cannot hold Medallia at 79 cents or 74 cents when Blackstone, the lead lender on the same loan, is at 60. The piece is the receipt event for the Pebbles I (March 26) and Pebbles II (April 1) thesis. The carry-trade transmission to foreign LPs - the Pebbles III research thread - remains in progress and is not yet at a publication threshold.

I run a concentrated portfolio. I get the question every week: 'Why is your concentration so high? Isn't diversification the whole point?' This piece answers it with the academic foundation that ties together three of my live theses. Harvard's Anita Elberse spent a decade proving entertainment markets are winner-take-all - the top one percent of products generates over fifty percent of the revenue, and substitutes do not exist for blockbusters. Arizona State's Hendrik Bessembinder proved the same pattern in equity markets - just over four percent of US-listed stocks generated all of the net wealth creation between 1926 and 2016, with the rest collectively returning the risk-free rate or less. The math says concentrate. The art is figuring out which names will be the next ones. The discipline is not betting the farm on any of them. The piece applies the framework across three markets: (1) the Drake re-sign at $830M-$1.3B that becomes a cap-rate reset for the entire top 0.1 percent of hip-hop, with the blue dot problem in concert touring (Pussycat Dolls, Meghan Trainor, Zayn, Post Malone all cancelled this week) explained as the same Elberse mechanic running with a tighter top-of-funnel; (2) the concentrated equity portfolio that captured Carvana at $15 in May 2023 (now over $300), Bombardier at C$11 in June 2020 at the bottom of the turnaround arc, Opendoor at $0.73 in July 2025 at the start of the Kaz Nejatian-led rebuild, and DAVE at $166-170 in April of this year, sized with long-dated LEAP options where available to push concentration further on highest-conviction names while staying disciplined enough to survive being wrong about any one of them; (3) the multi-asset hedged crypto treasury structure being built into EMJX through its reverse merger with SRXH - heaviest in Bitcoin because that is where the compounding has been most reliable, Ethereum as the second sleeve, select altcoins with genuine multi-bagger potential, and convexity equities alongside the digital assets, directly contrasted with Strategy's MSTR / STR* preferred stack as a single-asset book by design (Strategy 800,000+ BTC, 5.6x overcollateralization on STRC, $2.25B USD reserve - real institutional contributions, but a different exposure profile). The piece closes with 13 specific reader prescriptions - from Future the Prince and the manager of a mid-tier hip-hop artist to Sir Lucian Grainge at UMG, Vincent Bolloré controlling Vivendi, the institutional allocator running a 'diversified' book that exists to justify a fee schedule, the VC backing AI startups, the sports team GM, the Live Nation executive facing the blue dot problem, and the investor evaluating crypto treasury vehicles for the post-2026 cycle.

I sized the deal Universal Music Group will negotiate to retain Drake when his contract expires. Building on a hip-hop fan account's $1B prediction from March 14 (BlueSav, @MC_60sGang) plus my own structural call from Day 253 of the Iceman Countdown (April 29), the Wall Street structural model lands at $830M-$1.30B total, midpoint ~$1.07B. Components are explicit and resolve over 12-18 months through UMG's Euronext disclosures and the Variety/Billboard/MBW press cycle. Cash advance: $700M-$1B (BlueSav's $600-750M baseline plus a 20-45% Pershing-pressure premium drawn from Apple/Beats and Twitter/Musk activist-bid retention precedents — Bill Ackman's stalled $64.4B bid is the leverage Drake's negotiation operates inside). Tier-1 masters reversion (Thank Me Later, Take Care, Nothing Was The Same): $50M-$100M, accounting for the 1/3 ownership Drake already has on his first six albums via the 2009 Aspire/Cash Money settlement (per Hollywood Reporter, Billboard, McPherson LLP coverage). Tier-2 reversion (IYRTITL, Views, More Life): $15M-$45M, structured as 50% reversion on UMG's 2/3 of the Aspire-covered titles plus 50% of More Life. OVO Sound restructured as a UMG joint venture: $15M-$75M. Distribution + sync + publishing + ancillary rights: $90M-$250M. The piece is the first quantitatively defensible Wall Street-tier deal-pipeline model for a marquee artist re-sign — a transaction class the sell-side currently does not model. Iceman is the leverage album, not the breakup album. The receipt locks here. We will know.

EHIQ scored 16 potential Canadian Conservative leadership candidates on the conviction and specificity of their public register — using the same system that flagged Brexit (June 2016), Trump's first-term register (2016), and Mark Carney's Liberal leadership win (January 2025). The Milei conditions in Canada are partially present and the partial is growing. Mark Mulroney scored +8.3 — second in the field and the highest among candidates with a national path. The model is unambiguous on Pierre Poilievre: he will never win. His score has degraded from −2.79 (post-election concession) to −5.76 by mid-May. People have heard him. People have judged him. People have rejected him. Ben Mulroney scored −7.1 — the 15-point gap with brother Mark, same family same social world, is the cleanest calibration test in the dataset. Danielle Smith scored highest overall at +9.3 but does not move the 905 belt. Jean Charest peaked +3.6 right after the election and dropped to −0.2 as he started positioning for a run — politicians soften their language to protect optionality, and the model sees it. Jason Kenney's −2.6 places him in the same cluster as Poilievre — commentator register, not candidate register. The May 2026 Spring Fiscal Update was the first concrete test of Carney's fiscal-probity claim. The numbers do not support a clean break from Trudeau-era trajectory: $396B more debt over five years ($1.47T → $1.79T), debt-service costs jumping 50% to $81B by 2030-31, deficits above $50B every year through 2031, program spending growing 4.7% and 4.5%. The Fraser Institute's headline: 'Carney's fiscal update continues Trudeau-era approach to federal finances.' But the Milei moment in Canada is gated by something the model cannot see: the Canadian housing wealth illusion. Two-thirds of households own homes; paper wealth is anesthetic; a majority of voters can still look at their home price and tell themselves a different story than the fiscal numbers tell. A line on a Department of Finance table cannot compete with the line on the Realtor.ca page. The Milei moment in Canada does not arrive on a schedule. It arrives the morning after the illusion does.

Follow-up to today's open letter on the Pershing Square / UMG bid. The argument is sharper: Universal Music Group is anti-shareholder BECAUSE it is anti-artist. Those are not two parallel concerns. They are the same concern in two registers. Three new evidence layers documented in this 3,800-word follow-up: (1) Universal's verbatim 83-page Second Circuit appellate brief filed Friday March 27, 2026 — 11 days before Pershing announced — describing Drake as 'astoundingly hypocritical' and an 'unbothered rap artist' attempting to 'salve his wounds,' and weaponizing Drake's own November 2022 petition criticizing prosecutorial use of rap lyrics against Black artists; (2) the pre-Pershing calendar: March 18 stack (UBS cut UMG PT 36% on operational deterioration, Lorde walked after 17 years citing being 'pre-sold' as a 12-year-old, Sir Lucian Grainge spoke at Nvidia GTC about AI hyper-personalization), March 27 brief, March 30 €500M buyback, April 7 Pershing bid; (3) the 0-of-10 sell-side blindspot: across 10 reports from 7 banks (Morgan Stanley, JPMorgan, UBS, Wells Fargo, Deutsche Bank, Barclays) totaling roughly 70 pages of institutional analysis, 0 mentions of Drake by name, 0 of Iceman, 0 of the lawsuit on appeal, 0 of Sir Lucian's compensation, 0 of family concentration, 0 of Ovitz/In re Walt Disney Derivative Litigation, 0 of Tornetta v. Musk. Pershing's own letter, verbatim, on Ovitz: 'He also has a 40-year relationship with Sir Lucian Grainge.' That sentence is in the bid letter as the qualification for Chairman, not the disqualification. The 78% premium is paper; the cash floor is 29%. Drake's reply brief was filed by Willkie Farr & Gallagher LLP on April 17, 2026 arguing Judge Vargas committed reversible error.

Universal Music Group reports Q1 2026 earnings on Wednesday April 29. Drake's Iceman ships May 15. Bill Ackman's Pershing Square bid for UMG has stalled at board review for 19 days. This 5,000-word open letter documents what the financial press has not: UMG has underperformed the S&P 500 by ~91 percentage points since IPO; the platform that licenses UMG's catalog (Spotify) has returned 73.8% over the same period, more than UMG's own equity has delivered; Sir Lucian Grainge has refused to name Drake by name across his 3 most recent earnings appearances, while Drake holds the Billboard Hot 100 record for most top-10 songs in chart history; the combined Grainge family share of US current recorded music streams is now ~45.21% (UMG 37.48% + son Elliot's Atlantic Music Group 7.73%); the Pershing structure is conditioned on a NEW Grainge employment contract and Michael Ovitz, the named defendant in the canonical Delaware case on board oversight of CEO compensation, as Chairman. Pershing has been a UMG shareholder since shortly after the September 2021 IPO; the relationship has never been adversarial. The bid's true math: Bolloré 28.3% (paid €2.7B cash to bless) and Pershing 5% form the floor of 'yes'; the remaining 67%, including Tencent 11.4%, GIC 4.67%, Artisan Partners 4.66%, FMR, BlackRock, Vanguard, and Norges Bank, is the constituency that decides. The letter closes with 5 specific questions for Sir Lucian to answer on Wednesday's Q1 2026 earnings call.

Bill Cohan's Finding Satoshi, released April 22, concludes Satoshi was Hal Finney and Len Sassaman, both deceased. Ten days earlier I published a stylometric analysis of 830,000 words across 12 candidates that reached a different answer: Finney and Peter Todd. Both investigations agree on the structural two-author finding. Both name Finney as one of them. The disagreement is narrow. The math decides. Satoshi's corpus is 12.5 percent British. Finney plus Sassaman blends to 1.89 percent. Finney plus Todd blends to 12.66 percent, matching within 0.16 pp. The film eliminated Adam Back on timezone grounds in CET, then kept Sassaman in the same time zone. It never tested Peter Todd at all. Its 'smoking gun' is Hal's ambiguous eyebrow raise in a 2014 Forbes piece. Zimmermann's artillery-bracketing language, QRI's plan to get Satoshi to move a coin, and Hal's post-2011 denials to Fran and Callas all imply a living co-author. A disciplined living Todd is functionally identical to a dead one for coin non-movement. The math does not bend.

Last night Tesla raised 2026 capex to $25B+, confirmed negative free cash flow for the rest of the year, and disclosed that 3.5 million HW3 vehicles cannot host Unsupervised FSD. Pre-market: down 2.8% and holding. For a call that heavy, that is remarkable resilience, not punishment. Three times in a decade, a constrained Musk raising capex into a skeptical market has preceded a major Tesla upside run. 2015-2017 Gigafactory Nevada. 2018 Model 3 ramp and the 'mental scar tissue' call. 2019 Gigafactory Shanghai. Last night's call was the 2026 version of the same sequence. Elon stayed in the disciplined mode through genuinely hard disclosures. Three 'I think' openers from a CEO whose default is certainty. No dollar figure on the capex raise until the CFO was forced to give one. Timeline softening on robotaxi cities. The market is making the same mistake it has made in every prior Tesla capex cycle. It reads the balance-sheet commitment as thesis damage, and forgets that the commitment is the setup. Base case: the setup that worked three times works a fourth. Tail risk: physical AI substrate is less certain than manufacturing capex was.

ServiceNow beat Q1 numbers, raised full-year guidance, and dropped 13.6% after hours. Three of the most respected sell-side software analysts on Wall Street, Goldman's Gabriela Borges, Morgan Stanley's Keith Weiss, and Deutsche Bank's Brad Zelnick, cut ServiceNow price targets by a combined 100 dollars over seven days. All three maintained Buy or Overweight ratings. The rating is not the language. Morgan Stanley titled their note 'Struggling to Push the Signal Through the Noise.' Goldman's first-take: 'It is still too early to see AI driven product cycles move the needle on large SaaS revenue bases.' Deutsche Bank, pre-print: 'We struggle to see how Q1 can meaningfully catalyze the shares, even with good results.' The detail nobody is connecting: on the same afternoon NOW CFO Gina Mastantuono attributed a 75bp revenue headwind to delayed Middle East deal closings, IBM Chairman Arvind Krishna told IBM's call that IBM saw its strongest Middle East growth in decades in Q1. Same quarter, opposite operational results. Organic subscription growth at NOW decelerated from 20% to 17.25% once Armis and Moveworks are stripped. Options market prices 33% probability of hitting MS bear case of 75 dollars. Short book: ASAN, DOCU, CRM, TEAM, FIVN.

The consensus on SpaceX buying Cursor is that Elon overpaid 6x. That read is wrong. $60B is a discount to what was actually acquired once you model the stack vertically. Five value levers: compute arbitrage ($400M/year capitalizes to ~$10B), silicon roadmap compounding, developer distribution (10M seats worth $20B standalone), denial value ($10B to keep it out of OpenAI and Anthropic hands), and a self-reinforcing data moat. Chamath said this morning: zoning-approved powered land plus turnkey silicon plus distribution equals checkmate. Elon owns all three. This is the playbook, not an anomaly. What gets rerated next: data center REITs, nuclear generation, power infrastructure, and Bitcoin miners with already-approved, already-powered industrial land.

Bob Lefsetz writes the most-read music industry newsletter in the world. This week he published two letters, one on Apple's CEO transition and one on Coachella, that, without intending to, describe exactly what is happening in financial media in 2026. Replace 'festivals' with 'traditional financial broadcasts.' Replace 'musical stars' with 'Wall Street analysts.' Replace 'influencers' with 'financial influencers.' Replace 'the audience is the star' with 'your subscribers are where the alpha lives.' That is the map. Three Lefsetz observations apply directly: scarcity is back, the audience is the star, and you must create at all times. Drake's eight-month ICEMAN rollout is the music industry's 2026 proof point. Financial media is next.

The polygraph is 105 years old and barely works. TonalityIQ measures cognitive architecture instead: pronoun shift, hedging collapse, self-repair rate, distancing language, evasion frequency. We scored Prince Andrew's 2019 BBC Newsnight interview against his baseline speech. Result: 66/100, ELEVATED, coached denial confirmed. Five specific signals identified. Calibrated against a reference set including Gates (EXTREME), Wexner (EXTREME), Clinton (HIGH), Black (HIGH), and Buffett (GENUINE control). Andrew sits one tier below the Epstein associates who score EXTREME. The same system that called 7 bank failures 6-18 months early.

We ran TonalityIQ against everything Ryoshi ever wrote publicly: six recovered Medium posts and the Woof Paper, totaling 10,938 words. Tested against five candidates plus a negative control. Shytoshi Kusama, the current lead developer of the Shiba Inu ecosystem, is the closest match by a 53.6% margin over the second candidate. The system found no collaboration signal. Single author. The same cognitive fingerprint. The simplest explanation: Ryoshi did not hand the project to a stranger. Ryoshi changed his name.

TonalityIQ ranked Peter Todd first out of 14 candidates. Adam Back ranked sixth. Then we watched Cullen Hoback's HBO documentary "Money Electric" for the first time and found an 18-month filmed investigation had arrived at the same name independently. Carreyrou's NYT investigation concluded Adam Back at 99% certainty using surface features. TIQ uses cognitive structure. The machine found what the filmmaker found. Neither knew about the other. Both arrived at Peter Todd.

Follow-up to our Satoshi analysis. Blending Todd and Finney's linguistic profiles matches Satoshi 3.3x better than either individual alone -- 56.5% collaboration likelihood. A correction: Todd is Canadian, not American. Satoshi's 84/16 American/British spelling ratio matches a Canadian-American writing team. Early Satoshi writes like Todd (technical). Mid Satoshi writes like Finney (community). The temporal split aligns with Finney's ALS diagnosis. The person who writes like Satoshi is not a person. It is a partnership.

We ran TonalityIQ -- the same system that detected 7 bank failures -- on the writings of 12 Satoshi candidates across 870,000+ words. Two names separated from the field: Peter Todd and Hal Finney, just 7% apart. Everyone else was 30-300% further. The data cannot choose between them. But $78 billion in dormant coins can. One candidate is alive. One is dead. The private keys are locked inside a brain preserved at -196 degrees in Scottsdale, Arizona.

The capstone. Holmes said "I don't know" 662 times in her SEC deposition. Never once said "reliable" about a blood-testing company. SBF said "risk management" while exempting Alameda from all controls. 7 failed banks never said "uninsured" while holding 60-93% uninsured deposits. Surviving banks had 2x the constraint density of failures. 7 for 7. Zero false positives. 173,000 transcripts. 4,800 companies. The pattern holds everywhere.

Follow-up research on DAVE. Pay in Four launches Q2 2026 targeting 1.5M eligible members. Revenue estimates: $25M conservative, $58M base, $117M bull case. Zero analysts model it. Sezzle precedent: $4 to $187 (47x) on same BNPL-to-credit transition. DAVE runs the playbook without Sezzle's problems -- no merchant dependency, growing customers, off-balance-sheet credit via Coastal, no insider selling. At 10.6x forward earnings, Pay in Four optionality is free.

80 earnings calls across 20 years and 3 CEOs (Schmidt, Page, Pichai). Pichai’s conviction dropped from 10.22 (post-ChatGPT peak) to 4.64. Zero constraint language in Q4 2025. Larry Page I:We ratio 0.76 vs Pichai 0.26 — the founders chose safety over conviction. Zuckerberg running at 2x Pichai’s conviction level with more constraint acknowledgment. Google needs a CEO who says "I" instead of "we." Dead money risk until language shifts.

Full TIQ analysis across 21 transcripts and three CEO regimes. Kaz Nejatian’s posture evolution from Revolutionary to Demonstrated Builder is the fastest in our 173,000-transcript database. Acquisition velocity quadrupled. Mortgage product built in 10 weeks. October cohort most profitable in company history. The Carvana playbook: 85% of EBITDA from finance, not from buying and selling. Q2 will look ugly on the headline. The signal is in the margin.

45 transcripts across 12 years. AI language score 11.5 — top 1% of 173,000 transcripts. AI Pods: $20.6M ARR from zero, 60+ deployments, 45-60% margins. Disney, Google, Santander, FIFA as clients. Bloomberg consensus: $1.3B cumulative FCF through 2029 on a $2B market cap. Enough to retire all debt and buy back 40% of shares. Comp table: ACN 8.1x, EPAM 5.3x, VEEV 11.3x, NOW 13.3x — GLOB at 4.3x.

$554 million in revenue. 41% EBITDA margins. 8-day credit duration. 300 employees. TIQ conviction score: 91 with FUNDAMENTAL backing. The same Distress-to-Credibility pattern that preceded Carvana’s 32x run. After the Coastal transition, Dave becomes a tollbooth — collecting fees without carrying credit risk. Trading at 5.9x 2027 EBITDA vs Visa at 17x.

I called Carvana at $15. It went to $487. The same Distress-to-Credibility pattern exists in 10 companies right now. 169,000 transcripts. 4,664 companies. 18 scored a perfect 100. Two names being added as long positions in EMJ Capital and EMJX. Announcing soon.

83 transcripts across 24 years. Steve Jobs appeared once. Tim Cook built the tightest CEO baseline in the Mag 7. The conviction-to-constraint ratio just hit 3.72, the third-highest in 83 transcripts, at the exact moment Cook outsourced AI to Google. Every prior spike to these levels was followed by a pullback. Morgan Stanley, JPMorgan, Barclays integrated. Three-CFO map. Late April catalyst.

55 transcripts across 14 years. The maturation arc: from firing his CFO at 24 to producing the most disciplined transcript in the dataset at 38. TIQ detected the $97 bottom weeks before Cramer apologized. Sandberg-to-Li transition. Conviction Cycling pattern. Goldman Sachs, Morgan Stanley integrated. Position on April 30.

64 transcripts across 15 years. Three complete cycles of euphoria, crisis, and forgetting. The conviction-to-constraint ratio just hit 0.37, the most disciplined reading since 2018. Morgan Stanley, Goldman Sachs, and Wedbush coverage integrated. Five-CFO map. Forward return table. April 22 catalyst.

The sequel. Oratomic paper: 10,000 neutral atom qubits. Three orders of magnitude reduction in 5 years. Armstrong reversed from dismissal to "urgent" in 7 days. Why I own BTQ and not IONQ. The offense is priced. The defense is not. Three EHIQ quantum sensors tracking the gap in real time.

84 earnings calls across Ballmer and Nadella. Five distinct conviction phases from RESTRUCTURING_EARLY to CASH_MACHINE to AI_TAILWIND. The narrative-to-fundamental ratio has inverted for the first time since 2014. Goldman, UBS, and Morgan Stanley all maintained Buy after the January call. April 29 is the test.

716 earnings calls. The AI Paradox: companies in the top quintile of AI language density underperformed by 5.4pp over 90 days (p=0.043). Typology: Roadkill, Survivors, Quietly Recovering. Next Chegg scoring.

The full institutional analysis behind “Pebbles.” — 12 sections covering risk scores, Monte Carlo results, contagion mapping, FSK 7-quarter TIQ arc, ARCC comparison, monitoring framework, and systemic risk assessment.

Quarterly updates available with Intelligence tier ($999)