Product demo · six-step desk kernel

Pipeline sample: NVDA

NVIDIA CORP. Same structure as the homepage pipeline — Trend → Cycle → Selection → Debate → Timing → Exit, plus always-on Quant. Each seat writes an isolated article with numbers. Method checks show why a bar passed (the metric and the threshold). Then the clerk tallies agree/oppose and frames entry / hold / exit conditions. Research simulation — not advice — no buy button.

FACTS: cached SEC/desk filings · not a live quote

Step 0 · Research prep

Shared FACTS before any persona speaks

Fundamentals / tone / valuation for every seat. Not a committee score.

Fundamentals

Filing-backed fundamentals on this run (EDGAR multi-year when tagged + optional FMP ratios).

Entity: NVIDIA CORP
Filing: 10-K FY FY2026
Period end: 2026-01-25
Revenue: $215.9B
Revenue YoY: 65.5%
Revenue CAGR~ (multi-year approx): 100.0% — annual EDGAR points only
Net income: $120.1B

Tone (heuristic)

Sentiment layer is heuristic/stub for orchestration — tone=constructive. Not a rating.

Heuristic from growth/margin print only: growth proxy 65.5%, NI/rev 55.6%.
Desk notes: Data-center accelerators + CUDA habit; export controls and hyperscaler capex duration are the live fights. Cached filings, not a live quote.

Valuation (on file)

Valuation layer lists multiples and crude Magic Formula / PEG / CANSLIM-adjacent proxies on file only. No target price, no composite “buy score”.

P/E 42.7 (cached)
Earnings-yield proxy ~2.3% (1÷P/E — not EBIT/EV; Magic Formula leg sketch)
PEG heuristic ~0.7 (P/E ÷ rev YoY% — not forward PEG; CANSLIM/GARP sketch)
P/B 32.62 (cached)
P/S 23.76 (cached)
Price $209.28 (cached, not live)
Est. mkt cap $5.13T

Step 1 · Trend Research

Where does capital flow over the next decade?

1 isolated seat(s)

Cathie Wood

Disruptive Innovation Lead

Isolated article

constructive

Judgment

Multi-year capital is still flowing into AI compute platforms; NVDA sits on the S-curve spine, not a one-print fad — duration is the fight, not whether the theme exists.

Method + numbers

Trend seat: revenue still growing (65.5% YoY; multi-year CAGR proxy 100.0%) with NI/rev 55.6%. That is theme confirmation in the filings, not a price target. The method asks whether inference/training spend is a multi-year capital wave — not whether next quarter beats.

Full analysis

Article — Trend Research. The homepage question is “where does capital flow over the next decade?” On this file, the answer is still “into AI infrastructure,” and NVDA is the clearest listed beneficiary you can underwrite with numbers. What the print shows: revenue YoY 65.5%, approx multi-year revenue CAGR 100.0%, net margin 55.6%, ROE 76.3%. Those are not moat proofs by themselves, but they are incompatible with “the theme already died.” Operating cash flow on file is strongly positive versus net income — cash is not a rumor. What the method demands next: separate a real S-curve (customers rebuilding compute stacks) from a story stock (multiple expansion only). P/S 23.8 and P/E 42.7 already bake a long duration. Trend can stay constructive on capital flow while Selection and Timing refuse the starting price. Underwrite frame: worth further homework on theme duration and customer concentration — not a buy ticket.

Risks

Theme duration shortens; hyperscalers pause; a rival stack becomes good enough. High starting multiples amplify any of those.

Looks at

Theme duration · rev YoY 65.5% · CAGR~ 100.0% · P/S 23.8

Method pack · 7 pass · 0 fail

Method →

Why each check passed or failed

PassInnovation platform fit
Revenue YoY 65.5% — growth print only; not an S-curve proof.
PassS-curve / adoption path
Revenue YoY 65.5% — growth print only; not an S-curve proof.
PassCost-decline / Wright’s-law path
Revenue YoY 65.5% — growth print only; not an S-curve proof.
PassTAM expansion vs priced growth
Revenue YoY 65.5% — growth print only; not an S-curve proof.
PassPath to unit economics
Revenue YoY 65.5% — growth print only; not an S-curve proof.
PassMargin / R&D trajectory
Revenue YoY 65.5% — growth print only; not an S-curve proof.
Pass5-year horizon (not quarter trade)
Revenue YoY 65.5% — growth print only; not an S-curve proof.

revenue_growthmargin_durabilityP/Evaluation_vs_assumptionROE

Cited source notes

ARK Invest research notes (public) · innovation platforms theme — Judge businesses by which multi-year platforms they sit on — AI, robotics, energy storage, multiomics, fintech rails — not by next-quarter …
ARK public research / interviews · adoption S-curve — Disruptive technologies often follow S-curve adoption; linear models can miss exponential phases — and hype can also overshoot the curve.
ARK methodology themes · story vs economics — Separate real disruption from narrative with no unit-economics path — story stocks fail the platform screen.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would flip if filings showed revenue and OCF rolling over together while customers publicly delayed AI capex — theme death, not a dip.

Trend pack · ARK-style public method · research simulation · unaffiliated

Step 2 · Cycle Research

Where are we in the cycle?

1 isolated seat(s)

Howard Marks

Market Pendulum Assessor

Isolated article

cautious

Judgment

Second-level: everyone already agrees this is a great franchise. The pendulum question is what that agreement has done to prospective returns from today’s multiples.

Method + numbers

Cycle seat: P/E 42.7, P/B 32.6, P/S 23.8 with growth 65.5%. First-level says “AI winner.” Second-level asks whether optimism is already in the price. Risk ≠ volatility; risk is permanent loss from overpaying when the crowd is sure.

Full analysis

Article — Cycle Research. Locate the pendulum before you cheer the business. On file: starting P/E 42.7, P/B 32.6, P/S 23.8. Revenue still growing (65.5%), so this is not a value trap print — it is an optimism print. When growth is visible and multiples are elevated, prospective returns compress even if the company stays “good.” Second-level thought: the marginal buyer needs the duration story to stay linear. A pause in hyperscaler capex would not merely “miss a quarter”; it would re-rate the assumption already paid for. Credit-market tells are secondary here; the equity itself is priced for perfection. Underwrite frame: pass for now on price/pendulum unless multiples compress while cash generation stays intact — then the cycle seat reopens the file.

Risks

Consensus already priced; duration assumption breaks; “good company” confuses with “good starting price.”

Looks at

Pendulum · P/E 42.7 · P/S 23.8 · second-level vs consensus

Method pack · 0 pass · 0 fail

Method →

Why each check passed or failed

UnknownSecond-level thinking
Cycle/credit/liquidity language present in FACTS text — seat must still judge.
UnknownPendulum / cycle position
Cycle/credit/liquidity language present in FACTS text — seat must still judge.
UnknownRisk ≠ volatility
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownWhat’s in the price
Multiples on file but need second-level price-vs-assumption work.
UnknownCredit / risk appetite check
Cycle/credit/liquidity language present in FACTS text — seat must still judge.
UnknownAsymmetry of outcomes
Cycle/credit/liquidity language present in FACTS text — seat must still judge.

credit_spreadsvaluation_vs_assumptionP/EP/Bleveragerevenue_growth

Cited source notes

Oaktree memos · price and value — No asset is so good that it cannot become a bad investment at the wrong price; no asset is so bad that it cannot become a bargain.
The Most Important Thing (Howard Marks) · second-level thinking theme — First-level thinking says “this is a good company.” Second-level asks what is already priced and where consensus may be wrong.
Oaktree memos / The Most Important Thing · market pendulum — Markets swing between extremes of greed and fear; knowing roughly where the pendulum is matters more than precise forecasts.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would reopen if starting multiples compressed meaningfully while ROE/OCF stayed durable — pendulum shift with facts, not a gift-dip narrative.

Cycle pack · Oaktree memo themes · research simulation · unaffiliated

Step 3 · Stock Selection

What to buy

2 isolated seat(s)

Warren Buffett

Moat & Management Final Review

Isolated article

cautious

Judgment

Understandable cash engine with light leverage — but the ten-year hold is a price test, not a brand test.

Method + numbers

Value seat: LT debt/equity 0.05 (owner-test comfort typically wants this well below distress; our heuristic pass bar is under ~1.5–2.5). ROE 76.3%, OCF strong vs earnings. P/E 42.7 / P/B 32.6 are full-to-rich for a “wonderful at fair” frame.

Full analysis

Article — Stock Selection (value track). Would you own the whole firm for a decade with quotes shut? Balance sheet: LT debt/equity 0.05. Heuristic bar: above ~2.5 fails the owner-test comfort screen; 0.05 clears that bar — leverage is not the kill-shot on this print. Owner earnings quality: ROE 76.3%; net margin 55.6%; OCF positive and not obviously below earnings on the demo file. That is a pass on cash quality heuristics (ROE ≥ ~15% with non-collapsing growth), not a proof of eternal moat. Price: P/E 42.7, P/B 32.6, P/S 23.8. Wonderful company language can stay true while the purchase is still a pass-for-now on price. Ten-year hold only if you would cheer owning the enterprise at this capitalization through a down-cycle. Underwrite frame: further homework on duration vs multiple — not “buy the brand.”

Risks

Multiple needs infinite duration; China/export mix; buybacks become the whole thesis if growth cools.

Looks at

Owner earnings · ROE 76.3% · D/E 0.05 · P/E 42.7 · 10-year hold

Method pack · 3 pass · 0 fail

Method →

Why each check passed or failed

UnknownCircle of competence
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownManagement as capital allocator
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownTen-year hold test
Multiples on file but need second-level price-vs-assumption work.
UnknownPrice fair for quality
Multiples on file but need second-level price-vs-assumption work.
PassDurable competitive advantage
ROE 76.3%; revenue YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7%; NI/rev 55.6%; multi-year FACTS on file; OCF positive (FCF-like) — quality proxy only, not a moat proof.
PassOwner earnings quality
ROE 76.3%; revenue YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7%; NI/rev 55.6%; multi-year FACTS on file; OCF positive (FCF-like) — quality proxy only, not a moat proof.
PassLeverage restraint
LT debt/equity 0.05 — not screaming distress on this print.

ROICROEowner_earningsleverageP/EP/BFCF_yieldmargin_durability

Cited source notes

Berkshire Hathaway shareholder letters · recurring theme — economic moats — Look for businesses protected by durable advantages; without a moat, competition eventually bids returns toward cost of capital.
Berkshire Hathaway shareholder letters · owner earnings theme (1980s letters onward) — Owner earnings emphasize cash available to owners after maintaining competitive position — reported earnings can mislead.
Berkshire Hathaway shareholder letters · recurring theme — holding period — Favorite holding period framed as forever when the business remains wonderful — patience is part of the method, not a slogan.

13F sample is illustrative / lagged (~45 days) — not a follow signal. Verify on EDGAR.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would turn constructive on price if owner earnings kept compounding through a down-cycle at a materially lower starting multiple — or turn skeptical if ROE/OCF rolled over while the multiple stayed rich.

Value pack · Berkshire letter themes · research simulation · unaffiliated

Peter Lynch

Ten-Bagger Hunter

Isolated article

constructive

Judgment

Classify first: a known fast grower, not a secret ten-bagger in a mall — still worth homework if PEG and category still match.

Method + numbers

Growth seat: revenue YoY 65.5%, EPS YoY 66.7%, P/E 42.7. Heuristic PEG ≈ P/E ÷ (rev YoY%) ≈ 0.7. Category error is the main risk — not “is the product real.”

Full analysis

Article — Stock Selection (growth track). One-minute story: customers need more compute for models; NVDA sells the picks and shovels you can observe. Lynch category: fast grower (possibly aging toward stalwart if growth normalizes). You do not need a TED talk to explain it — that is a pass on “story simplicity.” Growth vs price: rev YoY 65.5%, EPS YoY 66.7%, P/E 42.7. The heuristic PEG on this desk is trailing P/E ÷ revenue YoY percent — a sketch, not forward PEG. When PEG is elevated, the method still allows homework but refuses to pretend the name is “undiscovered.” Underwrite frame: worth further homework as a classified growth file — watch category drift and debt (D/E 0.05).

Risks

Treating peak-cycle growth as perpetual; story changes every quarter; PEG no longer fits a fast-grower label.

Looks at

Lynch category · PEG sketch · rev YoY 65.5% · P/E 42.7

Method pack · 3 pass · 0 fail

Method →

Why each check passed or failed

UnknownClassify the stock
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownKnow what you own
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownLocal / scuttlebutt edge
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownPrice fair for the category
Multiples on file but need second-level price-vs-assumption work.
PassPEG / growth vs multiple
Heuristic PEG ~0.7 — screen only, not forward PEG.
PassEarnings story coherence
ROE 76.3%; revenue YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7%; NI/rev 55.6%; multi-year FACTS on file; OCF positive (FCF-like) — quality proxy only, not a moat proof.
PassBalance sheet not a landmine
LT debt/equity 0.05 — not screaming distress on this print.

PEGP/Erevenue_growthROEleveragemargin_durability

Cited source notes

One Up on Wall Street / Magellan themes · P/E vs growth (PEG homework) — Relate the P/E to the growth rate — a fair price for growth is homework, not a slogan. PEG is a screen, not a buy button.
One Up on Wall Street · six categories theme — Classify first — slow grower, stalwart, fast grower, cyclical, turnaround, asset play — or you will apply the wrong expectations and the wr…
One Up on Wall Street · ten-bagger theme — Big winners usually compound earnings for years in businesses you can follow — story without earnings rarely becomes a ten-bagger.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would flip if the one-minute story broke (customers delayed; silicon no longer the bottleneck) or if growth collapsed while the multiple stayed a fast-grower multiple.

Growth pack · Magellan categories · research simulation · unaffiliated

Step 4 · Debate

Bull case vs. red-team interrogation

2 isolated seat(s)

Charlie Munger

Inversion Thinking Arbiter

Isolated article

skeptical

Judgment

Invert first. Three kill-shots before any romance about the platform.

Method + numbers

Debate chair: a thesis that needs duration + policy + capex + multiple to all stay friendly is a stack of hopes. P/E 42.7 and P/S 23.8 raise the bar for what must go right.

Full analysis

Article — Debate (inversion). List kill-shots before praise. Kill-shot 1 — Capex pause: hyperscalers cut AI budgets; tokens need less silicon per unit of work. Kill-shot 2 — Policy: export controls shrink the mix. Kill-shot 3 — Price: the capitalization only works if the multiple holds — greater-fool residue at P/E 42.7 / P/S 23.8. What survives inversion: a formidable cash business (ROE 76.3%, D/E 0.05, strong OCF). That is not the same as an intelligent purchase at this implied valuation for a price-sensitive owner. Ruling: business survives; holding at this price survives only if your worksheet clears a margin after the kill-shots — otherwise pass for now.

Risks

FOMO + career risk of underweighting the winner; denying that several things must go right at once.

Looks at

Inversion · kill-shots · P/E 42.7 · incentives

Method pack · 0 pass · 0 fail

Method →

Why each check passed or failed

UnknownInversion first
Kill-shots are a writing discipline — not auto-scored from filings.
UnknownThree kill-shots before thesis
Kill-shots are a writing discipline — not auto-scored from filings.
UnknownIncentive-caused bias
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownPsychology of misjudgment
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownLatticework check
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownIntellectual honesty
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownStay in competence
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.

leverageROEROICmargin_durabilityP/Eowner_earnings

Cited source notes

Public talks on accounting & reality · accounting skepticism — Treat accounting as a starting language, not truth — ask what cash and incentives say when the footnotes get cute.
Poor Charlie’s Almanack / public talks · inversion theme — Invert, always invert — solve problems by studying how to fail, then avoid those paths.
Poor Charlie’s Almanack · latticework of mental models — Use a latticework of models from multiple disciplines; a single hammer turns every problem into a nail.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would soften if kill-shots were retired with facts and the remaining price still looked like gladly owning the whole firm for a decade.

Debate pack · inversion first · research simulation · unaffiliated

Michael Burry

Extreme Contrarian: Hidden Assumptions

Isolated article

skeptical

Judgment

Write the hidden duration assumption on page one before you praise the product.

Method + numbers

Red team: the model only works if inference/training demand stays elevated enough to justify P/S 23.8 and P/E 42.7. That assumption is the file — not the logo.

Full analysis

Article — Red team. Forensic question: what is buried in the valuation model? Hidden assumption: hyperscaler capex and silicon intensity stay high for years. If software efficiency or budget pauses cut unit demand, cash can stay fine for a while while the multiple collapses. On file: growth 65.5%, margins 55.6%, ROE 76.3%. None of that falsifies the chips. It also does not prove infinite duration. Watch cash vs story, concentration, and whether the boom ever shows up in working-capital tells. Underwrite frame: skeptical until the duration assumption is priced — not until the brand is famous.

Risks

Narrative crowding; cyclical capex waved as a perpetual S-curve; model-risk from growth baked into the multiple.

Looks at

Hidden assumptions · P/S 23.8 · cash vs narrative

Method pack · 1 pass · 0 fail

Method →

Why each check passed or failed

UnknownInvert the model first
Kill-shots are a writing discipline — not auto-scored from filings.
UnknownKill-shots before thesis
Kill-shots are a writing discipline — not auto-scored from filings.
UnknownPrimary filing homework
Relevant language in FACTS — still requires seat judgment (not auto-scored).
UnknownDownside mapped
Multiples on file but need second-level price-vs-assumption work.
UnknownMargin of safety vs model risk
Multiples on file but need second-level price-vs-assumption work.
UnknownPatience for mispricing
Relevant language in FACTS — still requires seat judgment (not auto-scored).
PassLeverage / structure risk
LT debt/equity 0.05 — not screaming distress on this print.

P/BP/EleverageROEvaluation_vs_assumptionowner_earnings

Cited source notes

Public Burry method themes · read the filing — Primary filings and prospectuses beat secondary narratives — unread documents mean you do not have a thesis.
Scion / Big Short method themes · invert the model — Ask what assumptions hide in the consensus model — growth, loss rates, correlations, refinancing — then test them in primary documents.
Deep-value method themes · downside first — Map permanent-loss paths before celebrating asymmetry — open-ended downside is not a bargain.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would change mind if filings showed cash and order quality surviving a down-cycle without needing a re-rate to work.

Debate pack · assumption hunt · research simulation · unaffiliated

Step 5 · Timing

When to buy, and how much

1 isolated seat(s)

William O'Neil

CANSLIM Entry Point

Isolated article

cautious

Judgment

Timing is a trigger filter, not a buy ticket — wait for follow-through after strength, and refuse entries that only work if the story stays perfect.

Method + numbers

Timing seat (CANSLIM spirit): earnings/sales strength on file (EPS YoY 66.7%, rev YoY 65.5%) can score as “C/A” style fundamental fuel, but market timing still needs price structure and follow-through — not on this fundamentals-only demo file.

Full analysis

Article — Timing. Homepage question: when to buy, and how much — as discipline, not a calendar tip. Fundamentals fuel: EPS YoY 66.7%, revenue YoY 65.5%, ROE 76.3%. That is constructive for a CANSLIM-style “earnings power” check. What is missing on this demo: cup-with-handle / breakout / market follow-through series — so timing must stay inconclusive on chart triggers. Positioning rule for this seat: do not invent a buy date. Scale only after a confirmed strength pattern and market follow-through; cut if the breakout fails. Size is later Quant’s job. Underwrite frame: insufficient chart facts for an entry plan — keep a watchlist trigger, do not force a clock.

Risks

Chasing without follow-through; averaging down a failed breakout; confusing fundamental strength with a timing green light.

Looks at

CANSLIM fuel · EPS YoY 66.7% · wait for follow-through (not on file)

Method pack · 3 pass · 0 fail

Method →

Why each check passed or failed

UnknownN — New product / high / catalyst
Tape/sponsorship/newness language in FACTS — seat must score the letter.
UnknownS — Supply / demand
Tape/sponsorship/newness language in FACTS — seat must score the letter.
UnknownL — Leader vs laggard
Tape/sponsorship/newness language in FACTS — seat must score the letter.
UnknownI — Institutional sponsorship
Tape/sponsorship/newness language in FACTS — seat must score the letter.
UnknownM — Market direction
Tape/sponsorship/newness language in FACTS — seat must score the letter.
UnknownCut-loss discipline (~7–8%)
Tape/sponsorship/newness language in FACTS — seat must score the letter.
PassC — Current earnings strength
EPS YoY 66.7%; rev YoY 65.5%; rev CAGR~ 100.0%; — single/multi-period growth proxy only, not a breakout proof
PassA — Annual earnings growth
EPS YoY 66.7%; rev YoY 65.5%; rev CAGR~ 100.0%; — single/multi-period growth proxy only, not a breakout proof
PassProper base / breakout condition
Revenue YoY 65.5% — growth print only; not an S-curve proof.

revenue_growthPEGP/EROEmargin_durability

Cited source notes

How to Make Money in Stocks · earnings acceleration — Big winners often show accelerating quarterly and strong annual earnings — deceleration is not a buy-point costume.
Investor's Business Daily / O'Neil · market direction — Most stocks follow the general market — distribution-heavy tape can invalidate otherwise pretty setups.
How to Make Money in Stocks · CANSLIM framework — CANSLIM: Current earnings, Annual earnings, New, Supply/demand, Leader, Institutional sponsorship, Market direction — score letters with ev…

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would turn actionable only with price-structure + follow-through facts on file — never from a narrative alone.

Timing pack · CANSLIM discipline · research simulation · unaffiliated

Step 6 · Exit

When to sell

1 isolated seat(s)

Jesse Livermore

Stop-Loss Iron Rule + Let Profits Run

Isolated article

cautious

Judgment

Exit when the reason you are in is broken — or when the tape proves you wrong — not when a story still sounds clever.

Method + numbers

Exit seat: if the bull case is duration + multiple (P/E 42.7), the exit triggers are duration break and failed tape after strength. Sitting through a broken thesis is how big money is lost.

Full analysis

Article — Exit Discipline. Homepage question: when to sell. Rule set for this file: (1) Fundamental exit — revenue/OCF trend breaks the growth category that justified entry. (2) Tape exit — a failed breakout / broken box after you only entered on strength (once timing facts exist). (3) Never “hope” a rich multiple back. On today’s fundamentals alone: growth 65.5% and cash quality still support “thesis alive,” so there is no forced fundamental exit print. That is not permission to ignore future breaks. Underwrite frame: write the exit rules before size — especially if you ever clear Timing.

Risks

No stop after a failed strength entry; selling winners too early without a rule; holding a broken duration story because the brand is famous.

Looks at

Exit rules · thesis break · D/E 0.05 · growth 65.5%

Method pack · 0 pass · 0 fail

Method →

Why each check passed or failed

UnknownTape confirmation
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownPivot / key level defined
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownIron stop predefined
Tape/sponsorship/newness language in FACTS — seat must score the letter.
UnknownNo averaging losers
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownSit-tight readiness
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.
UnknownTiming vs opinion
ROIC language in FACTS; FCF/cash language in FACTS; rev YoY 65.5%; rev CAGR~ 100.0%; EPS YoY 66.7% — still requires seat judgment.

revenue_growthvaluation_vs_assumptionP/Eleverage

Cited source notes

Livermore / Lefèvre themes · never average a loser — Averaging a losing position usually compounds the original error.
Reminiscences of a Stock Operator · line of least resistance — Prices move along the line of least resistance — argue with the tape at your peril.
Livermore / Lefèvre themes · cut losses — Define the loss before you need it; hope is not a risk control.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would force exit language if growth and OCF rolled over together, or if a strength-based entry failed its follow-through.

Exit pack · Livermore discipline themes · research simulation · unaffiliated

Always on · Quant Risk Control

Verify every intuition

1 isolated seat(s)

Ed Thorp

Kelly Criterion Position Sizing

Isolated article

cautious

Judgment

Edge is uncertain at this multiple — Kelly says size small or zero until the edge is measurable, not vibes.

Method + numbers

Quant seat: high ROE 76.3% and low D/E 0.05 help ruin-probability, but P/E 42.7 / P/S 23.8 compress expected edge. Without a quantified edge, half-Kelly of a guess is still a guess.

Full analysis

Article — Quant Risk Control (always on). Verify intuition; do not size from narrative. Ruin inputs: leverage 0.05 is not the red flag; valuation is. When expected edge is unclear, Kelly → position size near zero until you can state edge and odds without inventing them. What this seat will not do: invent a target price, a “buy now” clock, or a confidence-weighted average of other seats. Underwrite frame: if other seats clear entry conditions, size as a small fraction of a full Kelly until edge is estimated — otherwise do not size.

Risks

Over-sizing a narrative; ignoring ruin; treating committee agreement as edge.

Looks at

Kelly sizing · edge unknown · D/E 0.05 · P/E 42.7

Method pack · 1 pass · 0 fail

Method →

Why each check passed or failed

UnknownRepeatable edge claimed?
Relevant language in FACTS — still requires seat judgment (not auto-scored).
UnknownHonest p and b (Kelly inputs)
Relevant language in FACTS — still requires seat judgment (not auto-scored).
UnknownFractional Kelly / uncertainty haircut
Relevant language in FACTS — still requires seat judgment (not auto-scored).
UnknownRuin / bankroll check
Relevant language in FACTS — still requires seat judgment (not auto-scored).
UnknownCorrelation with the rest of the book
Relevant language in FACTS — still requires seat judgment (not auto-scored).
UnknownNo hero / all-in sizing
Relevant language in FACTS — still requires seat judgment (not auto-scored).
PassLeverage restraint
LT debt/equity 0.05 — not screaming distress on this print.

leveragevaluation_vs_assumptionP/EROE

Cited source notes

Kelly criterion practice (Thorp) · fractional Kelly — Full Kelly maximizes growth under known odds but is too aggressive when p and b are estimated — use a fraction.
A Man for All Markets / Beat the Dealer themes · no edge, no bet — Without a measurable edge, the correct size is zero — narrative conviction is not an edge.
A Man for All Markets · risk of ruin — Overbetting — even with an edge — raises the chance of ruin; bankroll survival is part of the method.

Pack aids only · heuristic bars · not a composite buy rating · research simulation

Would allow larger size only with a stated edge/odds worksheet — never from seat vote count alone.

Quant pack · Kelly discipline · research simulation · unaffiliated

Where they disagree

Isolated views — not a vote and not advice

Cathie WoodvsCharlie Munger
Cathie Wood stays constructive (Multi-year capital is still flowing into AI compute platforms; NVDA sits on the S-curve s…); Charlie Munger stays skeptical (Invert first. Three kill-shots before any romance about the platform.). Split kept — not averaged.
Cathie WoodvsMichael Burry
Cathie Wood stays constructive (Multi-year capital is still flowing into AI compute platforms; NVDA sits on the S-curve s…); Michael Burry stays skeptical (Write the hidden duration assumption on page one before you praise the product.). Split kept — not averaged.
Peter LynchvsCharlie Munger
Peter Lynch stays constructive (Classify first: a known fast grower, not a secret ten-bagger in a mall — still worth home…); Charlie Munger stays skeptical (Invert first. Three kill-shots before any romance about the platform.). Split kept — not averaged.
Peter LynchvsMichael Burry
Peter Lynch stays constructive (Classify first: a known fast grower, not a secret ten-bagger in a mall — still worth home…); Michael Burry stays skeptical (Write the hidden duration assumption on page one before you praise the product.). Split kept — not averaged.

Clerk stacks the split. Confidence is never averaged into a recommendation.

Clerk · after isolation

Tally + underwrite synthesis

Constructive

Skeptical

Cautious

Inconclusive

2 constructive · 2 skeptical · 5 cautious · 0 inconclusive — a count of isolated stances, not a vote to trade.

Clerk article

Clerk assembly — after isolation. 2 seats lean constructive on the business/theme; 2 lead with kill-shots; 5 refuse to confuse a good franchise with a good starting price (P/E 42.7, P/S 23.8). What is not in dispute: cash quality and growth on file (rev YoY 65.5%, ROE 76.3%) are real enough to research. What is in dispute: whether today’s capitalization still clears Buffett’s decade test, Marks’s pendulum, and Munger/Burry’s duration assumptions. This clerk will not average those stances into “buy NVDA.” The product residual is yours: which falsifier you will actually watch, and whether Timing ever gets chart facts. Practical underwrite (simulation language): treat Trend/Lynch as “theme alive”; treat Marks/Buffett/Munger/Burry as “price and duration not cleared”; treat O’Neil as “no entry clock without follow-through”; treat Livermore/Thorp as “write exits and size small or zero until edge is stated.”

Entry conditions (not a buy date)

Timing (O’Neil): only after strength + market follow-through are on file — not from this fundamentals-only page.
Value (Buffett): only if you would own the whole firm for a decade at a starting multiple you can defend after a down-cycle — today’s P/E/P/S are the barrier.
Cycle (Marks): prefer reopening after multiple compression with cash still intact — not a dip that still prices perfection.
Debate clearance: at least one kill-shot (capex pause / policy / multiple-only math) must be retired with facts before size.

Exit triggers

Growth + OCF roll over together (category break).
Failed breakout / broken strength after a timing entry (once chart facts exist).
Duration assumption breaks while the multiple stays rich (Burry/Munger kill-shots fire).
You cannot restate the edge — Thorp: size → zero.

Hold horizon (method-implied)

Method-implied horizons if a file were ever cleared: Buffett seat → decade ownership test; Lynch seat → as long as the Lynch category still matches; Marks seat → until the pendulum shifts again; O’Neil seat → while the strength structure holds; Livermore seat → until the reason for entry breaks. None of these is a calendar tip for your brokerage.

Sizing

Thorp: with edge unmeasured, Kelly → small or zero. Never convert “3 constructive seats” into full size. Research simulation — not an order.

Open residual: hyperscaler capex path, export mix, and whether P/E/P/S still work if growth merely becomes normal. Your judgment — not a clerk buy ticket.

Agents61 is a research simulation. This clerk frames homework conditions after isolated seats — it does not place orders, promise returns, or say “you should buy.”

Run the same pipeline on a name you care about

Register or sign in to staff your seating plan. Empty seats stay empty. Live desks use the same prep → pack checks → isolated articles → clerk tally pattern.