Skip to content

Thesis 4-Dimensional Framework

For each stock, we use the same 4-dimensional + 2 auxiliary framework to break it down. This differs from textbook DCF/rating models — it's a structure designed to enable multiple people to discuss the same stock.

4 Core Fields

Field Purpose Example (NVDA)
view Current directional judgment bull / bear / neutral / watching
confidence Your conviction in the view high / medium / low
supports 3-5 specific pieces of evidence "Q1 FY27 earnings 5/20, Data Center segment YoY +75%", "CRWV $99.4B backlog"
red_flags 2-3 potential anti-thesis trigger conditions "Samsung strike disrupts HBM supply", "H200 China export license revenue not recognized"

2 Auxiliary Fields

Field Purpose
catalysts (next 90d) Time-anchored observable events — earnings/conferences/product launches
price_outlook 90d base / bull / bear price range (must be anchored to current price)

Why This Structure (Not Others)

1. View forces a single-point judgment There's no "I'm not sure" option — you must choose bull/bear/neutral/watching. But the confidence field allows you to say "I chose it but lack conviction."

Counterexample: "NVDA is definitely good long-term but uncertain short-term" → This is vague judgment, not a thesis. In our framework, write it as: view=bull, confidence=low, red_flag="Short-term valuation compression"

2. Supports must be specific + verifiable You can't say "AI is the future" — that's an opinion, not a support. It must be:

  • ✅ "NVDA Data Center segment Q3 FY26 revenue $30.8B YoY +112%"
  • ❌ "AI industry is growing fast"

3. Red_flags must have trigger conditions You can't say "Valuation is high, there's risk" — that's always true. It must be:

  • ✅ "If NVDA fwd PE > 35x AND hyperscaler capex guidance is cut, valuation compression risk triggers"
  • ❌ "Valuation is too high"

A trigger lets you know when the thesis breaks — this is the core difference between a thesis and a "feeling."

4. Catalysts are time-anchored Each catalyst must have a date + observable outcome:

  • "2026-05-20: NVDA Q1 FY27 earnings. Look for: Data Center segment YoY +60%+, FY27 guide >$200B"

If a thesis has no catalyst within 90 days, it's nearly impossible to verify — you should set confidence=low or view=watching.

NVDA v91 Real Thesis Example

ticker: NVDA
view: bull
confidence: medium
core_thesis: |
  NVDA benefits from continued growth in AI infrastructure capex, Q1 FY27 earnings will validate demand strength,
  but needs to watch customer quality risk (CoreWeave circular financing).

supports:
  - "2026 hyperscaler AI capex expected to reach $725B+, NVDA is the primary beneficiary"
  - "CoreWeave order backlog approaching $100B, an all-time high"
  - "327 large institutions' 13F filings show NVDA remains the AI core holding"
  - "H200 China export license approved for approximately 10 companies"

red_flags:
  - text: "Samsung strike resumes 5/18, if 18-day strike materializes, could cut 3-4% of global DRAM supply"
    trigger: "Strike starts + lasts >7 days"
  - text: "CRWV-NVDA capital/commercial circular financing concerns (NVDA $36B stake in CRWV)"
    trigger: "CRWV customer concentration worsens OR next disclosure shows NVDA's CRWV exposure further expands"

catalysts_90d:
  - date: 2026-05-20
    event: "Q1 FY27 earnings"
    look_for: "Data Center segment >$30B, FY27 guide >$200B, gross margin >74%"
  - date: 2026-06-19
    event: "GTC Asia"
    look_for: "Rubin platform shipment timeline, Blackwell Ultra production update"

price_outlook_90d:
  base: "$215-235"
  bull: "$245-260 (Q1 earnings beat + China revenue confirm)"
  bear: "$190-205 (Q1 earnings miss OR Samsung strike impact OR continued 13F selling)"

How to Write Your Own Thesis

  1. Pick 1 ticker you're interested in (recommend the industry you know best)
  2. Fill in the blank template: 
    view: ?
confidence: ?
core_thesis: (1-2 sentence core view)
supports: (3-5 specific pieces of evidence)
red_flags: (2-3 + triggers)
catalysts: (1-3 events in the next 90 days)
price_outlook: (3 ranges + trigger conditions)
  3. Review every 7 days — Are supports still valid? Has a red_flag been triggered? Has a catalyst occurred?
  4. Review every 90 days — Was the thesis correct? Which support/red_flag was wrong?

For a detailed template, see Layer 5 · How to Write Your Thesis.

Next → Valuation/Earnings Terminology Glossary