Investment Philosophy - Beechwood Capital

"If you can't explain it without jargon so anyone understands, you don't understand it well enough."

Understanding is not memorizing facts. Understanding means you can trace every claim back to first principles, predict consequences of new information, identify what would change your mind, and explain ALL of this to someone with no domain expertise without losing a single nuance. Simplification is the OUTPUT of deep understanding, not a shortcut to avoid doing the work.

Depth First, Then Simplify

Go absurdly deep first. Read the actual papers. Understand the biology at the molecular level. THEN simplify. The simplification proves you understood. If your "simple" explanation has gaps, you didn't understand the complex version well enough. Never start with simple — that's just learning the headline without the substance.

Two Audiences, One Understanding

Expert level: To communicate in real-time with KOLs, scientists, company executives — understanding every word they say and its implications.

Non-expert level: To explain to a portfolio manager, family, friends who might buy the stock — without losing ANY nuance. If you can do both, you truly understand. If you can only do one, you don't.

Theory vs Data Reconciliation

Every mechanistic claim about a drug MUST be checked against actual clinical data. If they match, explain why biology predicted the data. If they DON'T match — that's a KEY INSIGHT. Explore possible explanations. Was the theory wrong? Is the data confounded? Never gloss over mismatches between theory and data. They are often the most valuable insights.

Target Location Determines Drug Format

WHY a drug is oral vs injectable is usually determined by WHERE the target sits. This single principle predicts the convenience profile, commercial uptake, formulary positioning, and competitive dynamics of every drug in the portfolio.

Extracellular Targets

Proteins floating between cells (TNF, IL-17, IL-23, PCSK9). Need a large antibody to grab them → injectable biologic. Too big for a pill — would be digested. Antibodies cannot cross cell membranes.

Intracellular Targets

Proteins inside cells (JAK, TYK2, PDE4, BTK, NLRP3). Need a small molecule that can pass through the cell membrane → can be oral pill. Antibodies can't reach them.

Modality	Route	Dosing	Target Access	LOE Risk	Key Tradeoff
Small molecule	Oral	Daily	Intracellular	Generic	Convenience vs selectivity
mAb	SC/IV	Q2W-Q12W	Extracellular only	Biosimilar (slow)	Efficacy vs convenience
Peptide	SC (or oral+enhancer)	Daily-weekly	Extracellular	Varies	Intermediate properties
siRNA	SC	Q6M	Intracellular (gene)	Complex	Very long-lasting vs liver concerns
Degrader	Oral	Daily	Intracellular (destroy)	Generic	Eliminates protein vs novel/unproven
ADC	IV	Q3W	Both (antibody delivers payload)	Complex	Targeted delivery vs toxicity

Replacement

Best drug wins almost all share. Homogeneous disease with one dominant pathway. Example: Keytruda in 1L NSCLC.

Split

Different patients need different drugs based on biology/subtype. Multiple drugs coexist. Example: psoriasis (IL-17 vs IL-23 vs oral suits different patients).

Sequential / Switch

Try drug A first, switch to drug B on failure. Market stratified by line of therapy. Example: LDL-lowering (statin → ezetimibe → PCSK9i).

Expansion

New drug brings in patients who weren't previously treated. Oral format for injection-averse patients. Example: oral PCSK9 (enlicitide expanding the PCSK9 market beyond injection-only).

What Actually Drives Doctors to Choose Drug A Over Drug B

1. First-mover advantage and clinical comfort — doctors prescribe what they've seen work in their patients
2. Formulary positioning and payer rebates — what insurance covers most favorably
3. Guideline inclusion — when a drug makes it into treatment algorithms
4. Patient preference — oral vs injectable, frequency, side effect profile
5. Safety monitoring requirements — JAK drugs require lab monitoring = more office visits = less convenient
6. Switching inertia — "my patient is stable on drug X, why rock the boat?"

Understanding these dynamics is as important as understanding the biology. A clinically superior drug that's behind a wall of prior authorizations will lose to a slightly inferior drug that's easy to prescribe.

BLANK > WRONG

If you don't know the real number, leave it null. A blank field signals "needs research." A wrong number (e.g., defaulting ownership to 50/50, applying class averages to individual drugs) signals "verified" and nobody catches it. Every placeholder that looks like data is a liability.

Drug-Specific, Never Class-Level

Every drug gets its own numbers from its own trials. Never apply class-level meta-analytic averages uniformly across drugs. If all statins show "myalgia 5-10%, diabetes 9%" in your comparison table, you haven't done the work. Rosuvastatin has the highest diabetes risk; pitavastatin has none. These differences drive prescribing and are investment-relevant.

Full Interconnection

Every data change must be fully interconnected. Adding a drug means: drug record + company link + program + indication chain + mechanism + study. Adding a study result means: update program stage, drug stage, related catalysts. Never split data addition and reconciliation.