P2B-C5 · Write Your Own Thesis¶

Key Insight

Writing is learning — 6 months from now you'll have your own thesis library, worth more than any KOL recommendation.

Layer 2 · Analyzing a Single Stock — After learning the industry, now apply it to one stock

L2-C5 (5 chapters total, this is the last). After this chapter, you'll have an executable template + weekly/monthly/quarterly review rhythm, and can ship your first thesis within 24 hours.

1. The Problem: The First 4 Chapters Were All About Others' Frameworks — If You Don't Write, It's Not Yours¶

C1-C4 you followed:

C1: Wrote 1 sentence (framework template)
C2: Filled in yaml (framework template)
C3: Looked up terms (dictionary)
C4: Followed NVDA (NVDA case)

But… you'll forget all of it in 3 days. Why:

No personal archive → can't find it for review
No review rhythm → thesis expires after a week and you don't know
No personal ticker library → you only have a feel for NVDA, other stocks are still zero

Writing is learning. This chapter gives you: a complete template + 5 self-checks + weekly/monthly rhythm + 6-week roadmap.

2. The Solution: Complete Thesis Template + Review Rhythm + Library Accumulation¶

Complete Thesis Template (Copy and Use)¶

Save as one markdown file (e.g., ~/thesis/NVDA_2026-05-19.md):

# Thesis: NVDA
created: 2026-05-19
last_reviewed: 2026-05-19
my_position: starter / core / full / hedge / pass
my_time_horizon: 3-12 months

# ===== Layer 2 Main Thesis (C2 4 Dimensions) =====

view: bull                  # bull / bear / neutral / watching
confidence: medium          # high / medium / low

core_thesis: |
  1-3 sentence core view (Can you explain it to a friend in 1 minute?)

# WHAT Dimension
supports:
  - Specific evidence 1 (with numbers / sources)
  - Specific evidence 2
  - Specific evidence 3

# RISKS Dimension
red_flags:
  - text: Risk 1
    trigger: If you observe X, this risk is triggered
  - text: Risk 2
    trigger: ...

# SO WHAT Dimension
catalysts_90d:
  - date: 2026-05-20
    event: Q1 FY27 Earnings
    look_for: revenue >$43B, gross margin >74%, FY27 guide >$200B

price_outlook:
  current: $225.32
  base_90d: $215-235
  bull_90d: $245-260 (if Q1 earnings beat)
  bear_90d: $190-205 (if Q1 miss OR Samsung strike impact)

# ===== Layer 3 Multi-Perspective (RISKS Strengthened) =====

multi_pm:
  value_pm: hold/pass (Valuation expensive PE 30x but ROE 100%+ unprecedented)
  growth_pm: bull/add (TAM AI capex $725B+, NVDA core beneficiary)
  macro_pm: bull/volatile (long-duration AI stock + Fed hawkish environment)
  my_takeaway: Long-term bull, short-term volatile, no position increase (based on 3 PM consensus)

anti_thesis:
  core: Hyperscaler $725B capex is a cycle peak over-investment
  mechanism: GenAI ROI disappoints → MSFT/GOOGL cut capex → NVDA inventory + valuation double contraction
  invalidation_triggers:
    - MSFT or GOOGL FY27 capex guide < 10% YoY
    - Rubin platform delayed to H2 2027
  strongest_advocate: SemiAnalysis (Dylan Patel) - occasionally writes this angle

macro_overlay: |
  Current regime mixed, near-term neutral. Fed hawkishness suppresses valuations but liquidity improvement offsets.
  → No change to NVDA bull thesis, but need short-term volatility tolerance

technical_overlay: |
  trend=uptrend, ADX 30.9 (strong), RSI 61.7 (neutral), price $225 near BB upper band.
  → Don't chase, wait for pullback to $215-220 (20d SMA) to add

# ===== Action =====

action:
  position_size: starter (3-5% of portfolio)
  entry: $215-218 (current price $225 wait for pullback)
  exit: $245-260 (Bull case) or break below $200 (stop loss)
  rebalance_trigger:
    - Major thesis change after Q1 earnings
    - Any invalidation_trigger occurs

# ===== Review Schedule =====

next_review: 2026-05-27 (1 week)
review_cycle: Weekly (financial data updates), Monthly (full thesis review)

3. How It Works: 5 Self-Checks + 3 Review Cadences¶

5 Self-Checks (Must Pass Before Draft Completion)¶

[ ] Can I explain core_thesis to a friend in 1 minute?
[ ] Does every support have specific numbers + sources (not vague statements like "AI is the future")?
[ ] Does every red_flag have an observable trigger?
[ ] Is there at least 1 catalyst within 90 days?
[ ] Can I explicitly state the strongest opposing argument + invalidation_triggers for anti_thesis?

All 5 yes → ship. Any no → go back and fill in.

3 Review Cadences¶

Weekly Review (Every Sunday, 10 min / stock)¶

Check supports: Still valid? Any new data strengthen or weaken?
Check red_flag triggers: Any triggered?
Check catalyst progress: How far from trigger date? Any leaks on look-for numbers?
Check price_outlook: Where is price (base/bull/bear)?
Update last_reviewed field.

Any major change → trigger monthly review (rewrite thesis).

Monthly Review (1^st of each month, 1 hr / stock)¶

Rewrite core_thesis 1-2 sentences (keep if unchanged)
Rewrite supports (remove invalid old supports, add new evidence)
Rewrite red_flags (mark resolved triggers, add new risks)
Next month's catalysts (most important 1-3 events)
Update price_outlook (based on new quarterly data)
Decide position adjustment: Add / Reduce / Hold

90-Day Review (Quarterly) — Most Important¶

Check if your thesis from 90 days ago was correct:

90d Ago	Current	Interpretation
Bull, price up 15%	Now bull, price still up	Thesis aligned with price, ✅
Bull, price down 10%	Now bull, but 1-2 supports invalid	Thesis partially wrong but has support, partial credit
Bull, price down 20%+	Now neutral/bear	Thesis wrong, what did you learn?
Bear, price up 20%+	Now watching	Bear missed, what was the opposing evidence?

Key: It's not about whether price was right, but whether the overall thesis direction + supports/red_flags were validated.

4. vs C4 What You Already Know¶

C4 gave you one complete cycle — but only for NVDA, and didn't explain how to maintain 5-10 theses simultaneously.

Dimension	C4 You Can Do	C5 You Can Do More
1 stock 1 cycle	✓	✓
Template reusable for 5-10 stocks	✗	✓ (yaml template)
Personal library archive	✗	✓ (markdown folder)
Review rhythm	✗	✓ (weekly/monthly/quarterly)
90-day review mechanism	✗	✓ (most important)
Ship/no-ship decision	✗	✓ (5 self-checks)

C5 is from single-stock cycle to library system. Without it, 6 months later all your theses will be expired and unreviewed.

5. Try It: Ship Your First Thesis Within 24 Hours¶

Task:

Pick a ticker (industry you know best — recommend 1 stock related to your work/interests. Don't pick NVDA to avoid overlap with C4)
Copy the complete template (see §2) → save as ~/thesis/<TICKER>_<date>.md
Run through C1 → C4 in sequence:
- C1: 30-word 1 sentence → seed for core_thesis
- C2: 4-dimension yaml → fill in view/supports/red_flags/catalysts/price_outlook
- C3: Terminology → ensure supports/red_flags have real numbers
- C4: KPI predictions → at least 3 + verify_at dates
Run 5 self-checks (ship only if all yes)
Add to calendar — next_review: <one week later>

Ship Criteria:

All 5 self-checks yes
File ≤ 200 lines (too long means you haven't distilled)
You can explain view + 1 support + 1 red_flag to a friend over coffee (without notes)

If you can't do it in 24 hours → go back to C1-C4 to find what step you missed.

6. What's Next (Layer 2 End, Layer 3 Start)¶

Congratulations, you've completed Layer 2: Analyzing a Single Stock — all 5 chapters.

You now have:

1 self-written thesis (single stock closed loop ✓)
Complete yaml template (reusable for 5-10 stocks)
Weekly/monthly/quarterly review rhythm

But all your theses are single-perspective — you're in an echo chamber alone. What if you're systematically wrong?

→ Real Analysis Process teaches you:

8 mental models / 50 mental models library
Value / Growth / Macro 3 PM perspectives automatically run on your thesis
Anti-thesis (forced opposing argument)
Macro + technical overlay

Layer 3 = Give your thesis 3 independent reviewers.

7. Deep Dive (Optional): Don't-Do List + Tool Recommendations + 6-Month Portrait¶

Click to see 4 anti-patterns + tool stack + timeline

❌ 4 Don't-Do Items¶

Anti-Pattern	Why Not
Review daily	Too frequent, you'll be swayed by noise. Weekly is enough
Change thesis based on price moves	A 5% price change shouldn't change your view. Thesis should drive price, not the other way around
Add thesis support immediately from Twitter	Information density too low. First verify (find transcript / 13F / company IR original documents)
Copy others' theses	Your review rhythm is different from theirs. A copied thesis you don't know well will panic you on a dip

🛠 Tool Recommendations (Minimalist, No Payment Needed)¶

Storage: Obsidian / Notion / git folder (I recommend git folder — you can diff thesis evolution later)
Financial data: Yahoo Finance (free is enough)
News: SeekingAlpha (free tier) / Bloomberg (paywall) / WSJ
Deep dive: SemiAnalysis (paid, but many freebies too) / Stratechery / Motley Fool Transcripts (completely free!)
Alerts: IFTTT / Zapier monitor RSS + ticker price

Paid (Bloomberg / FactSet) not necessary. Using only free sources for the first 6 months already puts you ahead of 90% of retail.

📅 What You Should Have After 6 Months¶

Time	What You Should Have
1 week	1 self-written thesis (C5 task)
1 month	3-5 theses + 1 monthly review experience
3 months	5-10 active theses + first 90-day review (see which were right/wrong)
6 months	5-10 active theses + each reviewed at least 6 times + 2-3 90-day reviews

This is your personal thesis library — 6 months from now you won't need any KOL recommendations, because you know every detail of your theses + your judgment record (which types of theses you're right on, which you're wrong on).

This is the end of Layer 2. Layers 3-5 are amplifiers, but the library itself is the core asset.

🎉 Layer 2 Complete — Keep building your thesis library. A year from now, you'll be amazed at how much you've grown.