Skip to content

bquirog/biotech-valuation-workflow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Biotech Valuation Workflow for Claude Code

A structured workflow for building and maintaining biotech company valuations using Claude Code. Designed for equity research analysts covering pre-commercial and clinical-stage biotech companies.

What It Does

This skill turns Claude Code into a biotech equity research analyst that can:

  • Initialize a structured company workspace with knowledge base, raw data store, and valuation models
  • Ingest sources (corporate presentations, SEC filings, press releases, analyst notes, scientific papers) into a compounding knowledge base
  • Build bottom-up disease revenue models with epidemiology-driven assumptions
  • Run Pre-Commercial valuations (risk-adjusted DCF + per-program attribution + scenario-weighted price target) with formula-driven Excel models
  • Maintain global disease reference files to ensure consistency across multiple company valuations
  • Track catalysts, risks, competitive positioning, and valuation history over time

Example Output

After initializing a company and ingesting a corporate presentation, you get:

kymera-therapeutics/
  CLAUDE.md                           # Company schema
  todo.md                             # Progress tracker
  raw/
    other/KYMR_Corporate_Pres.pdf     # Immutable source
  kb/
    pipeline.md                       # KT-621 STAT6 (Ph2b), KT-579 IRF5 (Ph1), KT-485 IRAK4...
    financials.md                     # Cash $1.6B, burn $87M/qtr, runway to 2029
    competitors.md                    # NX-3911 (Nurix/Sanofi), dupilumab landscape
    catalysts.md                      # BROADEN2 data mid-2027, IRF5 Ph1 data 2H26
    ...
  valuations/
    KYMR_Valuation_Model.xlsx         # Formula-driven: KT-621 AD, Asthma, IRF5, Partnerships

The Excel model uses all formulas (no hardcoded calculations):

  • Blue font = editable inputs (PoS, penetration, pricing)
  • Black font = formulas
  • Green font = cross-sheet links

Quick Start

1. Install the skill

# Option A: Clone and install
git clone https://github.com/bquirog/biotech-valuation-workflow.git
cp biotech-valuation-workflow/biotech-valuation.md ~/.claude/skills/

# Option B: Direct download
curl -o ~/.claude/skills/biotech-valuation.md \
  https://raw.githubusercontent.com/bquirog/biotech-valuation-workflow/main/biotech-valuation.md

2. Initialize a company

In Claude Code, navigate to your working directory and run:

/biotech-valuation init Nurix Therapeutics https://www.nurixtx.com/

3. Ingest data

/biotech-valuation ingest /path/to/corporate-presentation.pdf
/biotech-valuation ingest https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=NRIX&type=10-Q

The skill will extract key data, update the KB, flag valuation impact, and update global disease files for cross-company consistency.

4. Run a valuation

/biotech-valuation value pre-commercial

Produces disease-level revenue models, a formula-driven Excel workbook, risk-adjusted DCF with per-program attribution table, three valuation anchors (base DCF / cash floor / scenario-weighted PT), and sensitivity analysis.

5. Query the KB

/biotech-valuation query What is the ORR for bexobrutideg in CLL?

Architecture

Company Workspace

Each company gets a self-contained workspace:

Directory Purpose
raw/ Immutable source documents (SEC filings, presentations, notes)
kb/ Digested knowledge base (pipeline, financials, catalysts, risks, etc.)
kb/sources/ One summary per ingested source
kb/clinical-updates/ Per-asset clinical development tracking
valuations/ Valuation models, assumptions, Excel workbooks
todo.md Progress tracker

Global Reference Documents

At the root level, shared across all companies:

File Purpose
portfolio.md Master asset tracker — all companies, assets, phases, catalysts, competitive overlaps
diseases/*.md Canonical disease reference — epidemiology, drug landscape, valuation parameters

Why global docs matter: When you cover multiple companies targeting the same disease (e.g., two companies with STAT6 degraders in atopic dermatitis), the disease file ensures you use the same prevalence numbers, competitive landscape, and pricing assumptions. No discrepancies across models.

Disease Files

Each disease file contains:

# Disease Name

## Epidemiology
- Prevalence, incidence, addressable populations (canonical numbers)

## Drug Landscape  
- Approved drugs with sales
- Pipeline competitors with phase and key data

## Portfolio Assets
- Which of your covered companies target this disease
- Links back to company KB pages

## Valuation Parameters (Canonical)
- Standard pricing, penetration, EU/US ratios
- Use these numbers across ALL company models

Valuation Methodology

Pre-Commercial Protocol

Designed for pre-revenue biotech with identifiable pipeline programs:

  1. Disease-level revenue models — Bottom-up: epidemiology → addressable → penetration → pricing → revenue
  2. Risk-adjusted roll-up — Program-level probability of success (PoS)
  3. Expense build — R&D and SG&A with growth rate formulas; expense risk-adjustment applied only to R&D + SG&A, not COGS (COGS is variable cost tied to already-risk-adjusted revenue — double-counting the risk factor inflates peak-year margins)
  4. Risk-adjusted DCF — FCF discounted at WACC (12-13%); terminal value via perpetuity growth
  5. Per-program attribution — Decomposition of the DCF by program, showing $/share contribution and % of total. This is NOT a second valuation method — it reconciles to the DCF total by construction. Its purpose is to answer "where is the value coming from?"
  6. Three valuation anchors — (a) Base DCF, (b) Cash floor per share, (c) Scenario-weighted PT (25% bull / 50% base / 25% bear, bear floored at cash). These are genuinely independent lenses with different inputs. Do NOT compute a "blended DCF + SOTP" — that averages a number with itself.

Four Analytical Tools (keep them distinct)

Tool Purpose Question
DCF / rNPV Valuation How much is the equity worth?
Per-program SOTP Attribution Where is that value coming from?
Sensitivity table Risk framing Which assumptions move the value?
Scenario analysis Distribution What's the range of outcomes?

Excel Model Standards

All Excel models are formula-driven:

Convention Meaning Example
Blue font Hardcoded inputs PoS, penetration rates, pricing, WACC
Black font Formulas Revenue, FCF, PV calculations
Green font Cross-sheet links ='Disease Sheet'!D24
Red font External links or overrides Negative adjustments

Revenue formula: Revenue ($M) = Treated Patients (K) x Net Price ($K/yr) — units cancel naturally (K x $K = $M).

Typical PoS Assumptions

Stage PoS Range
Phase 2 pivotal (strong data) 35-50%
Phase 1b (with efficacy signal) 15-25%
Phase 1 (early) 10-15%
Preclinical (named target) 5-10%
Discovery 3-5%

Penetration Rate Guidelines

Market Type Peak Penetration
Large competitive (AD, asthma, RA) 3-8%
Medium competitive (CLL 2L+, MS) 15-35%
Niche/orphan (wAIHA, rare) 25-40%

Templates

The templates/ directory contains starter scaffolds:

# Copy company template to start a new workspace
cp -r templates/company/ my-new-company/

# Copy disease template to create a new disease reference
cp templates/diseases/disease-template.md diseases/my-disease.md

Workflow Tips

  1. Start with the corporate presentation — best single overview of pipeline, clinical data, strategy, financials
  2. Then the 10-K/10-Q — ground truth for detailed financials, IP, risk factors
  3. Ingest analyst notes — calibrate your model against street consensus
  4. Check disease files first when building revenue models — ensures consistency
  5. Update portfolio.md after every init/ingest that changes pipeline status

Data Sources Priority

Priority Source What You Get
1 Corporate/Investor Presentation Pipeline overview, clinical data, strategy, market sizing
2 10-K / 10-Q Full financials, IP portfolio, risk factors, share count
3 Earnings PR Latest quarterly financials, cash position, guidance
4 Clinical data presentations (ASH, ASCO) Detailed efficacy/safety data
5 Analyst notes Street consensus, valuation benchmarks

Requirements

  • Claude Code (CLI, desktop app, or IDE extension)
  • Python 3 with openpyxl (for Excel model generation)
  • Optional: LibreOffice (for automated formula recalculation)

License

MIT

Contributing

Pull requests welcome. Key areas for contribution:

  • Additional disease templates (oncology, rare disease, neurology)
  • Valuation methodology improvements
  • Integration with financial data APIs (EDGAR, Bloomberg)
  • Additional Excel model templates

About

Biotech company valuation skill for Claude Code. Pre-commercial equity research workflow with disease models, DCF/SOTP valuations, and formula-driven Excel models.

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages