Feynman-Inspired Deep Dive
Bottom-up, disease/drug landscape analysis. Biology first, then drugs, trials, competitive landscape, valuation. The ultimate test: can you take any expert statement about this disease, fully understand every word and its implications, AND explain it to someone with no science background without losing a single nuance?
Explanation Standards (Apply to All Steps)
Three Layers — Expert, Everyday, Analogy
1. Expert: Full technical terminology, receptor biology, signaling pathways. Sufficient for real-time conversation with a KOL or scientist.
2. Everyday Language: Same depth, same nuances, same number of distinct points — but in plain words. NOT a summary. NOT analogies. If the expert layer makes 6 points, the everyday layer covers the same 6. Could someone reading ONLY this layer understand ALL implications?
3. Analogy (optional, last): Only after layers 1 and 2 are solid. Must be vivid, logical, memorable. Bad analogies are worse than none. Analogies reinforce — they never substitute.
Full Functional Treatment of Every Entity
When introducing any biological entity, cover all 11 points: (1) What is it literally? (2) What does it do? (3) Where is it? (4) How is it activated? (5) Why does it matter for this disease? (6) What happens if you block it? (7) What drugs target it? (8) What does targeting predict for clinical profile? (9) What does actual data show? (10) Is the drug oral or injectable, and why? (11) How does it compare to other targets?
The 10 Steps
Step 0
What Do We Already Know?
Review related deep dives, biology, and knowledge that already exist. What biological foundations are already laid? What diseases, targets, drugs share biology? Start from existing knowledge — never restart from zero on covered biology.
Step 1
Define the Real Question
Define the key question BEFORE any analysis. Frame at three levels: the science question, the investment question, and the elevator pitch. What sub-questions must the deep dive answer?
Step 2
Exhaustive Jargon Map
Define EVERY non-everyday term. Not just clinical headlines — all biological entities, cell types, enzymes, signaling molecules, clinical measurements. Expert definition AND everyday language for each. Living section: update as later steps introduce new terms.
Step 3a
Healthy Biology (First Principles)
Understand the biological system on its own terms BEFORE disease or drugs. Build step by step, sequentially. Three Layers + Full Functional Treatment throughout. Goal: disease mechanisms and drug targets become OBVIOUS rather than memorized.
Step 3b
Disease Biology & Target Landscape
What goes wrong in disease, traced through the healthy pathway. Map ALL druggable intervention points. For each: what would blocking address, what would it miss, predicted safety, which drugs target it, predicted vs actual clinical profile.
Step 3c
Connections & Mental Model
Map how this deep dive connects to all prior knowledge. Cross-disease biology, overlapping targets, cross-disease drug positioning. Every new deep dive should make previous deep dives richer.
Step 3d
Bottom Line Distillation
Distill 3a-3c into streamlined bottom-line that PROVES understanding. Not a shortcut — the OUTPUT of deep understanding. Everyday version must have equal depth to expert version.
Step 4
Clinical Trial Analysis
Analyze key trial data from scratch. For each trial: design, endpoints (expert + everyday), results, what numbers mean. Flag cross-trial comparison pitfalls. Apply Theory vs. Data Reconciliation — does data match Step 3b predictions? Mismatches are key insights.
Step 5
Competitive Landscape
Comprehensive map: all drugs by mechanism and format. 5 required deliverables: efficacy table, dosing/convenience table, commercial landscape, catalyst map, doctor decision framework. Market structure analysis: replacement vs split vs switch vs expansion.
Step 6
Valuation
Translate biology/trial insights into valuation. Market size, penetration, probability of success, pricing — all clearly justified. All assumptions explicit. Could someone without a finance background follow the model?
Step 7
Scenario Model
Bull/Base/Bear scenarios with probabilities. Expected value calculation. Identify the 2-3 factors that genuinely move the needle. Sensitivity analysis: what changes the thesis most?
Step 8
Downstream Impact Chain
Per-scenario downstream impacts on other companies/drugs in portfolio. Second-order effects (payer behavior, pipeline reprioritization, biosimilar dynamics). Trading playbook: if X happens, do Y.
Step 9
Post-Mortem & Calibration
Track predictions vs outcomes. When wrong: biology gap or market irrationality? Update framework based on lessons learned. Calibrate confidence levels for future predictions.
Company Teardown Framework
Top-down, company-level investment analysis. Business overview, drug mapping, value drivers, revenue bridge, competitive positioning, pipeline, catalysts, valuation, market perceptions, variant perception, and the final call. Distinct from the Feynman Deep Dive — the teardown synthesizes disease-level work into a company-level investment decision.
The 11 Steps
Step 0
What Is This Company?
Business overview, revenue base, key franchises, corporate structure, management track record, balance sheet, BD/M&A philosophy, geographic footprint. What do they do, where, and how?
Step 1
Map Every Drug That Matters
For each material drug: disease, mechanism, competitive comparison, clinical evidence, prescribing dynamics, revenue trajectory, regional dynamics, consensus expectations, and where YOU differ from consensus. Link to Feynman disease/landscape deep dives where they exist.
Step 2
Key Value Drivers
The 3-5 things the stock trades on. Growth story, value story, turnaround, or SOTP discount? What questions does every investor ask first? Rank by magnitude of P&L and stock impact.
Step 3
Revenue Bridge
Today's revenue → 3-5 year forward view. Growth drivers (ramps), growth brakes (declines), patent cliffs (with erosion curve analysis, lifecycle management, historical comps), pipeline fills (probability-adjusted). Does the math work?
Step 4
Competitive Landscape (per franchise)
For each key franchise: current competitors, new entrants, prescribing hierarchy, competitive moat durability, paradigm shifts, market share shift dynamics, historical pattern recognition. Link to Feynman deep dives.
Step 5
Pipeline & BD Strategy
Full pipeline inventory with probability-weighting. Which assets is the market paying for vs free options? Implied pipeline value. Key binary readouts. BD/M&A outlook: gaps to fill, potential targets, balance sheet capacity, takeout potential.
Step 6
Catalysts & Event Path
Map every catalyst over 12-18 months. Classify stock-moving vs noise. For each: market expectations, your expectations, upside/downside skew, magnitude. Quarterly earnings focus. Front-loaded vs back-loaded catalyst path. Regulatory/pricing/political risks (IRA, HTA, VBP).
Step 7
Valuation
SOTP, DCF, relative comps, implied expectations. Scenario analysis with probability weights and price targets. Historical valuation context — what drove past re-ratings or de-ratings?
Step 8
Perceptions & Positioning
Sell-side consensus, buy-side positioning, management credibility, KOL/physician sentiment, short interest, recent stock action and why. Is this consensus long, consensus short, or under-owned?
Step 9
Debates & Variant Perception
Every active debate. Your view vs consensus. The variant perception: what do you see that the market doesn't? Why does this mispricing exist? What would cause re-rating? What would change YOUR mind?
Step 10
The Call
Long / Short / Do Nothing — and WHY. Upside/downside magnitude, timeline, conviction level. Position sizing. Tactical (3-month) vs fundamental (3-year). What's the first thing you'd look at tomorrow morning to check your thesis?
Maintenance Workflows
Systematic processes for keeping research current. Core principle: always show diff first, never auto-execute. User approves every change.
Quick Commands
Earnings Update
[TICKER] reported earnings, update
Fetch latest earnings data, show diff of what changed vs current tracker state, user approves before any database writes.
Catalyst Triage
catalyst triage (weekly)
Review overdue catalysts, unclassified events, missing links. Flag anything that needs resolution or reclassification.
Readout Resolution
[TICKER] [DRUG] read out, [OUTCOME]
Resolve a catalyst event. Update downstream entities: program stage, drug stage, related catalysts. Reconcile state across all connected records.
Staleness Check
staleness check (monthly)
Identify stale financials (>90 days), overdue catalysts, outdated debates. Flag everything that needs refreshing.
Catalyst Update
update [TICKER] catalysts
Refresh catalyst timing from latest IR presentations, earnings calls, and press releases for a specific company.
Landscape Update
landscape update [DISEASE]
Refresh competitive landscape for a disease area. Check for new approvals, trial readouts, market share shifts.
Deep Dive
deep dive [TICKER] [DRUG] [TRIAL]
Launch a full Feynman-inspired deep dive following the 10-step process. Outputs an interactive HTML dashboard.
Company Teardown
teardown [TICKER]
Launch a company-level teardown following the 11-step framework. Top-down investment analysis.