Real Public Cases¶
Part 4 Mission
Use the five tools from Part 3 (Hedge Fund / Buffett / Bottleneck / Multi-PM / Anti-thesis) to dissect 5 real public events. One case per chapter, all based on public data (earnings / 13F / public interviews / government announcements), 0 paid sources.
Part 4 Backbone
Part 3 gives tools, Part 4 puts them to work. Each case explicitly invokes which Part 3 tools — after learning, you can dissect the next event yourself.
5-Chapter Map¶
| # | Title | Core One-Liner | Which Part 3 Tools Used |
|---|---|---|---|
| C1 | DeepSeek Selloff Full Forensic | NVDA -17% in one day — panic or thesis broken? | All 5 tools (none break) |
| C2 | Samsung Loses HBM Market Share | 3 companies in same industry, 1-year stock divergence ±150% | P3-C3 Find Bottleneck 5 Steps |
| C3 | Stargate $500B Beneficiary Chain | Selling shovels > Panning for gold | P3-C3 + P3-C4 Multi-PM |
| C4 | Export Controls → Domestic Substitution | Regulatory lag 6-24 months + substitution effect | P3-C4 Multi-PM Perspective |
| C5 | Buffett Didn't Buy NVDA Forensic | The discipline of non-participation | P3-C2 Buffett 5 Steps |
Part 4 vs Part 3 Relationship¶
| Part 3 Tool | Used in Part 4 Chapter |
|---|---|
| P3-C1 Hedge Fund 5 Steps | C1 (Druckenmiller / Coatue held, didn't trim) |
| P3-C2 Buffett 5 Steps | C5 (NVDA fails each of the 5 steps) |
| P3-C3 Find Bottleneck 5 Steps | C2 Samsung HBM / C3 Stargate beneficiary chain |
| P3-C4 Multi-PM Perspective | C1 / C3 / C4 multiple angles |
| P3-C5 Anti-thesis | C1 (4 invalidation triggers after DeepSeek all not activated) |
Recommended Pace¶
- Day 1: C1 DeepSeek (biggest recent event, learn short-term panic vs long-term thesis)
- Day 2: C2 Samsung HBM (bottleneck-finding in practice)
- Day 3: C3 Stargate (shovel-selling framework)
- Day 4: C4 Export Controls (geopolitical lag lesson)
- Day 5: C5 Buffett (discipline of non-participation)
After Completing Part 4, You Can¶
- Take any new event (new NVDA earnings / new Stargate-like announcement / new geopolitical shock) and dissect it within 30 minutes using Part 3's 5 tools
- Distinguish short-term price panic vs long-term thesis broken
- Identify shovel-selling beneficiary chain vs single direct beneficiary
- Acknowledge circle of competence boundaries, don't participate in what you shouldn't
Any ✗ → go back to the relevant chapter to review.
5 Case Selection Forensic
These 5 cases are not random. DeepSeek + Samsung + Stargate + Export Controls + Buffett Non-Participation covers the 4 major wildcards in AI investing (algorithm / supply chain / policy / geopolitics) + 1 major discipline (non-participation). If you can dissect these 5, you can apply the same approach to most new events.