Research & Analysis
AI ensemble equity analysis, earnings deep dives, and lessons learned building a multi-model research platform.
Featured
Caterpillar: Dominant Moat, $51B Backlog, Record Revenue — But Earnings Fell 13%. What Five Lenses Found
E&T power gen surged 30% past $10B on data center demand. $51B record backlog (+71%). Yet EPS fell 13% and tariffs hit $1.8B, escalating to $2.6B in 2026. At ~20x P/E, the market prices structural growth — but only E&T is delivering it while CI runs merchandising programs and RI margins compress to 10.7%. Five lenses, 7 signals, 14 debates, 5 cross-lens tensions. Dominant moat. Demanding valuation. One central question.
Top Posts(popularity tracking coming soon)
Airbnb: Dominant Brand, Proven Moat, and a Regulatory Corridor That Keeps Narrowing
10 minHims & Hers: Five Federal Proceedings, 60-70% Compounding Dependency, and 48 Hours That Proved the Risk Is Real
12 minRobinhood: 100% Revenue Growth, 57% Transaction-Dependent, and a 17.7x P/S Priced for Permanence
10 minAll Posts
Inside the Play Portfolio: Committee Discourse, Quarter-Kelly Sizing, and Systematic Discipline
Every position goes through a 4-step committee discourse. Every size is computed by quarter-Kelly. Every rejection is transparent. Here's how the play portfolio turns thesis assessments into positions — and when the formula says no.
Okta: 3 Breaches in 4 Years for an Identity Company, Yet 106% NRR and $870M FCF — What Six Lenses Found
An identity security company breached 3 times in 4 years — yet NRR holds at 106%, $1M+ customers grew 15%, and FCF hit $870M on $2.15B net cash. Microsoft Entra offers free identity to 350M+ M365 users. Management claims $80B+ AI TAM while 2 of 345 positions are AI-dedicated and zero insiders buy stock. At ~26x non-GAAP forward P/E (vs SaaS peers at 55-80x), the market is skeptical — not euphoric. Six lenses, 9 signals, 3 cross-lens conflicts, 9 monitoring triggers.
China's 75/25 Paradox: The Demand Pivot That Would Help China and Hurt Everything Else
We ran China's pre-NPC stimulus composition through 6 macro-specific analytical lenses — stimulus composition, commodity transmission, inflation regime, global spillover, trade transmission, financial conditions — producing 12 signals and 6 conditional market pairs. The headline: supply-side dominance (75/25) sustains a deflationary export cycle keeping global conditions 150-200bp looser than policy implies. The demand-side pivot that would help China (10-25% probability) would paradoxically tighten global conditions. March 5 NPC is the binary catalyst.
DocuSign: 1.4% IAM Adoption, 23pp Margin Gap, and a Valuation That Prices Zero Upside
DocuSign's IAM pivot has 25,000 customers — 1.4% of its 1.8M base. Management frames 8% growth as 'reacceleration' while non-GAAP margins exceed GAAP by 23pp ($610M SBC). An $820M one-time tax benefit inflated headline net income. At ~2.5x EV/Revenue, the market prices the 'no' case — no IAM upside, no margin expansion, no acceleration. Five lenses found 6 signals, unanimous convergence, and one binary question: does IAM reach critical mass within 2-3 years? The answer determines everything.
Marvell: 73% Revenue From One Bet, a $5.5B Pre-Revenue Acquisition, and a 35% Correction That May Not Be Enough
Marvell's moat and its biggest risk are the same thing. The custom silicon relationships that lock in 18+ hyperscaler programs also concentrate 50%+ of revenue in two customers. Revenue grew 37% but is decelerating (base-normalized 2-year CAGR is ~21%). The $5.5B Celestial AI acquisition has $0 revenue. At ~$80 (down 35% from $127 peak), expectations still embed 28-32% CAGR through FY2028. Four lenses produced 8 signals — 3 favorable, 3 cautionary, 2 mixed — and 3 unresolved conflicts. The central paradox: strengths and weaknesses are structurally inseparable.
The Fed-ECB Gap Is Closing Through One Channel. Here's What 5 Lenses Found.
The Fed-ECB policy divergence is narrowing through a single, fragile transmission channel — the 12.7% EUR/USD appreciation — while credit channels are impaired on both sides of the Atlantic. Five lenses independently found that nearly all cross-border monetary transmission runs through FX, not credit. The ECB has cut 200bp yet produced tighter credit conditions; the Fed has cut 75bp and achieved its loosest measured conditions. 5 conditional market pairs quantify what Fed delivery would change — the EUR/USD pair shows a +30pp causal delta, the largest in the framework.
BOJ Is Normalizing. Here's What 6 Analytical Lenses Found About the Risks Nobody Sees.
We ran the BOJ policy normalization cycle through 6 macro-specific analytical lenses — rate transmission, inflation regime, financial conditions, labor dynamics, fiscal interaction, global spillover — producing 12 signals and 5 conditional market pairs. The hedging cost trap has activated for the first time in two decades: Japanese investors now earn -68bp on hedged US Treasuries vs JGBs. The speculative carry trade unwound 81,652 contracts in 5 weeks. Spring 2026 Shunto wage negotiations are the single highest-leverage catalyst identified by all 6 lenses.
Costco: 7 Lenses Found Zero Operational Weakness. At 54x P/E, That May Be the Problem.
81M members, 92.2% renewal, 35.6% ROIC, 3 years of margin expansion, SCOTUS tariff win (yesterday), fortress balance sheet. Seven lenses found no material operational weakness. But at 54x P/E (39% above 10-year average), the market prices Costco as a 'compounder' not a 'retailer' — and that classification is the most fragile assumption embedded in the stock. 12 signals, 19 debates, 4 compound scenarios including the 'Perfection Trap' at 15-25% probability. Zero published contrarian theses exist.
CrowdStrike: 97% Retention Through the Worst IT Outage in History, DOJ/SEC Investigating the Core Metric — What Nine Lenses Found
CrowdStrike survived its own Black Swan — 8.5M systems crashed, 97% retention held, record $265M net new ARR. But DOJ/SEC is investigating the ARR metric that proves recovery. Scope expanded beyond original $32M Carahsoft deal to HHS and DOE contracts. At ~20x P/S with 41/41 analyst Buy ratings, the market prices zero probability of adverse outcome. Nine lenses, 12 signals, 4 cross-lens conflicts, 12 monitoring triggers — all converging on the same question: what discount does a federal investigation into the core operating metric warrant when the market applies none?
The Supreme Court Struck Down Tariffs. Here's What 5 Analytical Lenses Found.
The SCOTUS ruling struck down IEEPA tariffs. Trump pivoted to a 15% flat tariff under Section 122 with a 150-day expiration. We ran the transition through 5 macro lenses — producing 10 signals and 5 conditional market pairs. The headline: tariffs transmit as a consumption tax (+35pp import prices, +27pp goods CPI) but fail as industrial policy (+4pp manufacturing jobs). July 24, 2026 is the single convergence event where everything reprices.
Target: $5B CapEx Bet Into Declining Revenue, 33% Rally With Zero Positive Comps — Q4 Earnings Preview
4 consecutive negative comp quarters. $2.5B revenue decline over 2 years. Market share lost in ~60% of tracked categories. Stock rallied 33% to ~14x P/E anyway. Kohl's historical analog matches on 5 of 6 signal dimensions — 73% value destruction despite adequate financials. The $5B CapEx bet may be addressing the wrong problem: Moat Mapper found the root competitive issue is pricing and convenience, not store quality. CEO transition with $26M in discretionary selling vs $281K in retention. Seven lenses, 12 signals, unanimous convergence. March 3 earnings in 12 days.
MongoDB: 30% Atlas Growth, 52% SBC-to-Revenue, and a Profitability Question Nobody Agrees On
MongoDB SBC is 52% of revenue — 2x the next closest SaaS peer. Non-GAAP operating margin is 18%. GAAP is approximately -30%. Atlas grew 30%. Beat-and-raise x3. $2.3B cash, zero debt. But 3 CFOs in 18 months, zero insider purchases, AI catalyst 12-18 months ahead of reality, and active securities litigation. Seven lenses, 11 signals, unanimous convergence — five lenses flagged SBC as the defining structural issue. The profitability is real on one layer and illusory on another.
The Fed Held in January. Here's What 6 Analytical Lenses Found.
We ran the January 28, 2026 FOMC decision through 6 macro-specific analytical lenses — rate transmission, inflation regime, financial conditions, labor dynamics, fiscal interaction, global spillover — producing 12 signals and 5 conditional market pairs. The headline: inflation convergence has stalled at 2.5-3.0%, financial conditions are significantly looser than policy implies, and equity wealth effects have displaced housing as the primary transmission channel. 5 paired conditional markets test what a surprise cut would change.
Autodesk: Confirmed Accounting Manipulation, SEC + DOJ Both Cleared, Stock Flat — What Seven Lenses Found
Autodesk deliberately manipulated non-GAAP metrics tied to exec comp. The Audit Committee confirmed it. SEC and DOJ both investigated and closed with no charges. Starboard settled with governance reforms. Stock at ~19x forward PE — bottom decile of 5-year range — while 93% subscription revenue, $7.4B RPO (+20%), and 9-10% organic growth. Seven lenses, 10 signals, zero forced convergences — the market appears anchored to risks that have already been resolved. AI disruption: genuinely uncertain.
32 Companies, 2,234 Lens Reports, and 5 Patterns We Didn't Expect to Find
In 55 days we analyzed 32 companies through 13 analytical lenses, generated 2,234 reports, created 255 forecast markets, and published 53 articles. Here are the cross-company patterns that emerged — insider-narrative divergence, inverted market narratives, governance-operations disconnect — and three things the multi-LLM process taught us about itself.
Intuit: Stock Down 43% on 'AI Will Kill TurboTax' — Revenue Up 18%, AI Growing Fastest
The market priced in AI disruption — stock down 43% from $814 to ~$390. But TurboTax Live grew 47%, 2.8M AI agent users in 4 months, revenue accelerated 13% to 18%. The bear thesis is directionally wrong. Then why did insiders sell $375M with zero purchases? Mailchimp revenue declining 4 years after $12B acquisition. SBC creates 48% non-GAAP/GAAP EPS premium. Seven lenses, 9 signals, unanimous consensus — the central unresolved tension: if the stock is mispriced, why are insiders selling?
Salesforce: AgentForce 330% ARR Growth — But It's 1.3% of Revenue. Q4 FY2026 Earnings Preview
AgentForce ARR grew 330%. It represents 1.3% of revenue. Pricing changed 3+ times in 18 months. Meanwhile, the subscription machine delivers 92% retention, $12.4B FCF, and 12pp margin expansion post-activist. $49B goodwill with zero impairment. Microsoft Dynamics growing 2.4x faster. Seven lenses, 11 signals, unanimous consensus — all converging on the same question: are investors paying for the proven machine or the unproven narrative? Q4 earnings Feb 25.
Block (XYZ): 64% Earnings Growth, Stock Down 20%, $22B in Opaque Lending — Which Story Is Right?
$22B in lending originations with zero vintage data. $255M in fines settled, SEC/DOJ still open 3+ years. GP accelerating 9% to 18% while stock drops 20%. Six lenses found a paradox: strong execution on top of concentrated structural risk. Q4 2025 earnings Feb 26 — the most immediate catalyst to test or confirm the assessment framework.
Beyond Meat: From $230 to $0.70 — What Went Wrong and Can It Recover?
Stock down 99.7% from peak. $1.2B debt restructured into ~5x dilution. Revenue declining 15%+ with no stabilization. Zero insider purchases below $1. Related-party lender with secured priority and repriced warrants. Five lenses, 11 signals, zero favorable findings — all converging on the same question: has the restructuring merely delayed the inevitable, or does any path to equity recovery exist?
Carvana Q4 2025: $1.9B GAAP Profit on a Pre-Tax Loss, SEC Investigation, and the Most Complex Name in US Auto Retail
CVNA beat EPS by 274% ($4.22 vs $1.13) — then dropped 15-20%. The headline profit was a $2.8B non-cash DTA release on a pre-tax loss of $890M. SEC investigating $600M+ in Garcia family RPTs. Two short sellers allege fraud. Management gave a blanket denial. 8 lenses, 15 signals, 25 debates — all converging on the same question: does the SEC investigation resolve the tension between genuine operational strength and captured governance?
Visa: Dominant Moat, Unprecedented Regulatory Test — What 6 AI Models Found About DOJ + CCCA Dual-Front Pressure
Stock down 11% while revenue grew 15%. DOJ antitrust (MTD denied) targets debit exclusivity. CCCA with Trump endorsement targets credit routing. Dual-front regulatory pressure on both products simultaneously — unprecedented in Visa's history. $21.6B FCF, 0.09x leverage, VAS at 28% constant-currency growth. Compound scenario (10-17% probability) could produce 30-50% equity impairment while funding remains STABLE. The market is pricing fear — the question is whether it's enough. Eight-lens deep dive.
Mastercard: Dominant Moat Meets Its Toughest Test — What 6 AI Models Found About the Most Hostile Regulatory Environment in MA History
3.7B cards, 220+ countries, revenue at 2x GDV growth — but MDL settlement, UK CAT ruling, CCCA reintroduction with presidential endorsement, and Capital One migration all hit at once. 5 of 6 lenses flagged the same risk. VAS organic growth is 15-19%, not the 22% headline. At ~38x P/E, the moat is genuine — the question is how much rent regulation allows it to generate. 20-30% correlated probability of meaningful adverse scenario within 3-5 years. Seven-lens deep dive.
Berkshire Hathaway: Post-Buffett Transition — What 6 AI Models Found About Whether the Moats Survive
Buffett is gone. Greg Abel has been CEO for 45 days with zero acquisitions and unknown BRK holdings. $176B insurance float is widening. PacifiCorp wildfire ($48B claims, BBB-) is the overlooked time bomb. Record Q3 earnings (+33.6%) prove operations work without Buffett, but $381B cash earns declining yields with zero deployment. Three compound tail scenarios at 5-15% probability. The 'Correlation Trap' — Berkshire exposed to wildfires as BOTH insurer AND utility owner. Six-model deep dive.
Lemonade: The AI Insurance Paradox — Strong Execution, Stretched Valuation, and the Five-Dimensional Success Requirement
Stock up 94%, loss ratio at record lows, cash flow positive a year early — yet ~8x P/S requires simultaneous success across 5 dimensions with 25-35% conjunction probability. AI advantage is real ($20M/year) but explains only 6% of the valuation premium. Car insurance at 76% loss ratio. Bundling at 5% vs 30-40% industry. Q4 earnings Feb 19 — the single most important data point. Eight-lens deep dive.
Deere & Company: The Right-to-Repair Nexus — FTC Targets the Same Moat That Makes DE Defensible
Revenue down 25%, $2.4B tariff+labor headwind, FTC suing to pry open proprietary diagnostics — yet 450bps higher margins and record $2.31B R&D. All 5 lenses flagged the same risk variable: the proprietary tools that create switching costs are the tools the government is suing to open. 8 signals, 22 debates, 3 compound scenarios. HIGHER SCRUTINY posture.
Booking Holdings: The Parity Paradox — EUR 8B+ in Lawsuits, Zero Revenue Damage After 14 Months
EUR 413M Spain fine, EUR 8B+ class action, DMA gatekeeper designation, $457M KAYAK impairment — yet revenue accelerated and guidance raised every quarter of 2025. Five lenses found 8 signals: strong execution today, structural risks building on divergent timelines. The market may be overweighting near-term fear while correctly anticipating structural deterioration that hasn't yet appeared in financial results.
Airbnb Q4 2025: Revenue +12%, GBV Highest in 2+ Years — All 5 Signals Confirmed
Revenue $2.8B (+12%) beat guidance by $80M. GBV +16% — highest in 2+ years. Q1 2026 guided +14-16%. Full-year 2026 'at least low double digits.' All 5 signals confirmed. Narrative-reality gap narrowing but North America still mid-single-digits and product-cycle effects (200-300bps) have uncertain durability. Moat trajectory upgraded to Stable-to-Widening. Four-lens earnings update.
Twilio Q4 2025: Record Revenue, First GAAP Profitability, Voice Renaissance — All 10 Signals Confirmed
Q4 revenue $1.4B (+12% organic), FY2025 $5.1B (+13% organic), FCF $945M (+44%), first full year of GAAP profitability. Voice hit high teens growth with Voice AI accelerating above 60%. All 10 signal classifications confirmed with strengthened evidence. Silent Churn Spiral tail risk partially defused. FY2026 organic guide 8-9%. Six-lens earnings update.
Fastly Q4 2025: Stock Surged 64-88% on a Genuine Beat — But the AI Narrative Driving It Is Unquantified
Revenue +23%, first profitable year, guidance $65-75M above street. Stock surged 64-88%. But the agentic AI traffic thesis driving the repricing is E1 evidence only — management declined to quantify it at the most favorable moment. Execution EXCEEDING, narrative DIVERGING. Five-lens deep dive.
Walmart: DOMINANT Moat, DURABLE Revenue, $681B Scale — But at 45x P/E, Is the Transformation Already Priced In?
Walmart crossed $1T market cap with a DOMINANT moat, DURABLE revenue, and EXCEEDING execution at $681B scale. Operating margin expanded from 3.34% to 4.31% in three years. The 'Second P&L' (advertising + membership) is real but earlier-stage than the narrative implies. At 45x P/E with 93% analyst buy ratings, a GLP-1 grocery spending headwind was missed by every prior lens. Five-lens deep dive.
Moderna: FDA Blocked Its Flu Vaccine, Revenue Is 95% COVID, and the Market Values the Business at Zero
FDA's CBER Director overruled career scientists on a pre-agreed trial design. International regulators accepted the same application. Revenue down 90% from $18B peak. But the market values Moderna's operating business at approximately zero above cash, cost execution beat targets by $900M, and oncology showed 49% melanoma reduction at 5 years. Five-lens deep dive inside.
Twilio: Growth Re-Accelerated to 13%, AI Revenue Is 'Still Relatively Small,' and Buybacks Exceeded Cash Flow by 3.4x
Twilio grew from 7.3% to 13% organic while the CEO says AI revenue is 'still relatively small.' Buybacks of $3B exceeded operating cash flow by 3.4x. Segment acquisition looks value-destructive. Net cash $1.4B. Q4 earnings Feb 12 — five-lens earnings preview inside. UPDATE: See our Q4 2025 earnings analysis for the latest.
Robinhood Q4 2025: Record Results, Stock Down 40% From Highs — Our Committee Changed One Signal
$4.5B revenue (+52%), 56% EBITDA margins, #1 in options, 8 straight quarters of positive net transfers — and the stock is down 40%+ from its late-2024 peak. EXPECTATIONS_PRICED moves from DEMANDING to MODEST as P/S compresses from 17.7x to ~9.9x. All other signals confirmed. Four-lens earnings update.
Datadog Q4 2025: $953M Revenue, 29% Growth — Every Signal Confirmed
Revenue accelerated to $953M (+29% YoY). Non-AI customer growth surged to 23%. All three core pillars crossed $1B ARR. Record $1.63B bookings. No signal classifications changed — moat trajectory upgraded to confirmed widening. Full committee update.
How We Assess Price vs. Value (Without a Price Target)
We classify equities as price-above-value or price-below-value but never name a target price. No DCFs. No earnings multiples. Instead: probability-weighted stress-testing of the market's implicit assumptions. Here's exactly how the prediction ensemble produces that assessment — and why we think calibrated uncertainty beats false precision.
Airbnb: Dominant Brand, Proven Moat, and a Regulatory Corridor That Keeps Narrowing
Airbnb generates 90% unpaid traffic. Market share grew 28% to 44%. $4.5B FCF at 38% margins. But NYC eliminated 90%+ of listings. Barcelona's ban was upheld. 4 active lawsuits. North America at low-single-digit growth while management says 'not close to mature.' Q4 earnings Feb 12 — four-lens preview inside.
Hims & Hers: Five Federal Proceedings, 60-70% Compounding Dependency, and 48 Hours That Proved the Risk Is Real
DOJ criminal referral. FDA enforcement naming HIMS specifically. Novo Nordisk patent lawsuit. FTC investigation. Securities class actions. Stock down 60%+. But the commonly cited 31% GLP-1 figure may understate the real exposure — our Myth Meter found total compounding dependency may be 60-70%. Five-lens deep dive inside.
Robinhood: 100% Revenue Growth, 57% Transaction-Dependent, and a 17.7x P/S Priced for Permanence
Robinhood's revenue doubled to ~$5B with 75% EBITDA margins and 11 $100M businesses. But 57% is transaction-based, 52-57% depends on specific regulatory treatment, crypto (21%) is at cycle peak, and 17.7x P/S requires ~4 independent risk dimensions to break favorably for 3+ years. Four-lens deep dive inside.
NVIDIA: $130B Revenue, 73% Margins, and a $3T Valuation That Demands Perfection
NVIDIA's revenue doubled to $130B with 73% margins, but 88% is Data Center, its own customers are building ASICs (Claude 4 trained on Trainium2), DOJ has subpoenas out, and the $3T+ valuation requires 25-30% growth at 70%+ margins for years. Six-lens deep dive inside.
Oracle's $50B AI Gamble: $523B in Contracts, 8 of 9 Signals Elevated, and a Balance Sheet at the Breaking Point
Oracle guided $50B CapEx on $67B revenue. FCF collapsed from +$10.8B to -$30B in two years. ~58% of $523B RPO tied to OpenAI. CDS spreads at 2009 levels. 7 insiders sold $46.5M, zero bought. Five-lens analysis inside.
Amazon Q4 2025: $200B CapEx, 8% Selloff, and the Largest Capital Bet in Corporate History
Amazon's $200B capex plan exceeded consensus by 37%. Stock dropped 8%. But AWS growth re-accelerated to 24% (fastest in 13 quarters) with $244B backlog. 7 signals, 12 debates, 4 lenses reveal both the bear and bull case have legitimate evidence.
AST SpaceMobile: $30B Market Cap, Zero Service Revenue, and the Most Polarizing Stock in Space
6 satellites deployed of 45-60 needed. FCC license pending while SpaceX already has one and filed an adversarial complaint. CTO sold 94.4% of holdings. $30B market cap on zero commercial service revenue. But the broadband D2D technology is real, the MNO partnerships are real, and the bull case survives scrutiny. Six-lens deep dive inside.
Compass: The Largest Residential Brokerage Merger in History Meets 4.4x Leverage on Cyclical Revenue
Compass acquired Anywhere Real Estate for ~$10B, creating 340K agents across 7 brands. Leverage jumped from 0x to 4.4x. The acquired entity nearly went bankrupt under similar leverage in 2008. 'Tech-enabled brokerage' narrative meets traditional brokerage economics. Four-lens deep dive inside.
Netflix: $82.7B WBD Deal, Record Organic Results, and a Market Pricing Only One Story
Netflix stock down ~28% while revenue hit $45.18B (+16%), margins expanded to 29.5%, and ad revenue doubled. DOJ second request. Management said 'more builders than buyers' 4 months before the deal. The organic business has never been stronger — but the market is pricing only the deal risk. Five-lens deep dive inside.
Snowflake: $1.5B in Stock Comp, 2.3% AI Revenue, and a Narrowing Path
Snowflake spends $1.5B/year on stock comp (40.8% of revenue). AI is 2.3% of product revenue despite 'AI revolution' framing. Databricks surpassed SNOW on ARR ($4.8B vs ~$4.4B). NRR fell 33pp. $400M+ insider selling, zero buys. Q4 FY2026 earnings Feb 25 — five-lens analysis inside.
Datadog: Down 46% From Peak While Every Metric Accelerates
Datadog stock is down 46% from peak. Revenue growth accelerated from 25% to 28%. Non-AI usage hit a 12-quarter high. Insiders sold $110M+ with zero purchases. Q4 earnings Feb 10 — four-lens analysis inside.
GitLab: 89% Gross Margins, Decelerating Growth, and an Unproven AI Pivot
GitLab's DBNRR fell from 122% to 119% in 3 quarters. 80% of expansion depends on seat growth — the exact model AI threatens. Duo Agent is the strategic fulcrum, but carries zero evidence. Five-lens deep dive inside.
Reddit Q4 2025: Strong Execution Meets Demanding Expectations
Revenue $726M (+70% YoY, 9.2% beat), EPS $1.24 (32% beat), DAUq 121.4M (+19%), stock +4.5% after-hours. DAUq growth challenges our traffic-decline thesis. $1B buyback contradicts $192.7M insider selling. Four-lens signal update inside.
Molina Healthcare Q4 2025: FY2026 EPS Guidance Collapses 64% — Every Signal Worsened
FY2026 EPS guided to $5.00 (down 64%). Q4 GAAP EPS: -$3.15 (first quarterly loss). Stock -35%. California retroactive premium clawback introduces new precedent risk. Revenue durability debate resolved — both lenses now FRAGILE. Posture moved to AVOID.
Reddit: 74% Revenue Growth, 55% Traffic Decline — What's Really Happening?
Reddit grew revenue 74% while Google traffic dropped 55%. Insiders sold $192.7M with zero purchases. Data licensing growth collapsed from +66% to +7%. Is this sustainable or a warning sign?
Eli Lilly Q4 2025: Beat-and-Raise Confirms GLP-1 Leadership
$19.3B Q4 revenue (+43%), $80-83B 2026 guidance. LLY +7% vs NVO -13%. Our committee confirms all signals — CONDITIONAL revenue durability, ELEVATED regulatory exposure, DEFENSIBLE competitive position. Thesis validated.
Novo Nordisk Q4 2025: From Growth to Decline — Revenue Durability Downgraded
Management guides for -5% to -13% revenue decline in 2026, reversing from +10% growth. Stock down 13%. Our committee downgrades REVENUE_DURABILITY from CONDITIONAL to FRAGILE.
PayPal Q4 2025: CEO Ousted, Branded Checkout Stalls
Board fires Alex Chriss for execution concerns. Branded checkout growth collapses from +5% to +1%. Stock down 19%. Our committee upgrades Platform Wallet Competition from SLOW_EROSION to ACCELERATING_EROSION.
Eli Lilly Q4 2025 Earnings Preview: The 57% Concentration Question
Tirzepatide is 57-63% of Lilly's revenue — highest single-molecule concentration in large-cap pharma. Q4 earnings Feb 4, orforglipron FDA decision March 2026. Is this concentration strength or vulnerability?
Novo Nordisk: Is the GLP-1 Gravy Train Ending?
Lost 9pp market share to Lilly in 12 months. Stock down 40-66%. Four regulatory pressures converging. Is this oversold or justified? Three-lens analysis inside.
Disney Q1 FY2026: Why We're Not Changing Our Assessment
No monitoring triggers fired. Streaming +12% revenue, +50% earnings YoY. Experiences hit $10B quarterly milestone. ESPN DTC remains the question mark.
The $5.3B Swing That Changed Disney's Streaming Math
Disney turned streaming profitable with a $5.3B turnaround. Now they're returning $9.7B to shareholders while carrying $37B debt. Is this confidence or hubris? Four-lens analysis inside.
The 29% Miss That Changed Resideo's Bonus Plan
Why did OCF disappear from management compensation the same year they missed targets? 4 CFOs in 7 years. Growth collapsed from 32% to 3%. We ran the pattern through 5 lenses.
What the Gotham City Research Report Taught Us About Carvana
Our LLM committee re-analyzed CVNA after new FOIA evidence. Classification upgraded from ELEVATED to HIGH. Here's what we got right, what we missed, and how we're improving.
The Backtest Illusion: Why AI Stock Platforms Show You Fantasy Returns
Any AI can beat the market from 2017-2021. You just have to build it in 2021. A guide to spotting backtest manipulation.
Auditing the Auditors: What We Learned Running 4 LLMs Through a Structured Debate
LLMs have personalities — predictable behavioral patterns that show up consistently. We ran four models through a structured debate and mapped their biases.
Advanced LLM Use: How Frameworks Can Hide Hallucinations
We built structured frameworks to reduce LLM errors. Then we discovered the frameworks were creating new errors we couldn't see.
The Fugazi Filter: When You Can't Trust the Numbers
Standard analysis assumes the numbers are real. What if they're not? A framework for fraud and fragility risk.
The LLM-isms in "LLM-isms"
We graded ourselves on our own LLM-ism warnings. Final score: 2 guilty, 3 acceptable, 2 intentional.
LLM-isms: Patterns That Should Raise Your Skepticism
A guide to recognizing rhetorical patterns that may indicate overconfidence or shallow reasoning.
Why Personas Matter More Than Models
The real insight isn't 'use multiple LLMs' — it's which roles you assign. Our persona ensemble uses specialized analysts, critics, and synthesizers.
A Taxonomy of LLM Hallucinations in Financial Analysis
Not all hallucinations are equal. Some are dangerous, some obvious, some almost impossible to catch.
The Prospectus Probe: A Framework for Newly Public Companies
IPOs are minefields. Here's how we systematically evaluate newly public companies.
More Posts Coming
New equity deep dives weekly, plus methodology posts on prompt engineering lessons, cross-LLM disagreement patterns, and forecasting calibration.