Startup Valuation MCP Server

Startup Valuation Engine

A comprehensive startup valuation library implementing 80+ formulas from the Startup Valuation textbook — Python library, MCP server, and AI-agent skills.

Overview

A production-grade Python library for startup valuation, implementing every formula from the Startup Valuation textbook by Simon Mak (Valuation in Practice Series, Ascent Partners). Designed for developers, financial analysts, and AI agents who need auditable, structured valuation computations.

Three-layer architecture:

graph TB
    subgraph Library["Python Library"]
        MOD["14 Modules<br/>80+ Functions"] --> VR["ValuationResult"]
    end
    subgraph MCP["MCP Server"]
        VR --> SVR["FastMCP Server<br/>14 Tools"]
    end
    subgraph Skills["AI-agent skills"]
        SVR --> CORE["Core"]
        SVR --> ADV["Advanced"]
        SVR --> IND["Industry"]
        SVR --> STAKE["Stakeholder"]
        SVR --> EMER["Emerging"]
    end
    style Library fill:#0083AB,color:#fff
    style MCP fill:#4CAF50,color:#fff
    style Skills fill:#9C27B0,color:#fff

1. Python Library — 14 modules, 80+ typed functions, all returning ValuationResult (value + assumptions + sensitivity)
2. MCP Server — 14 folded tools (80+ formulas) for AI agents via stdio and hosted Streamable HTTP
3. AI-agent skills — 6 skill definitions with workflow guidance for valuation domains

Installation

pip install startup-valuation          # library only
pip install startup-valuation[mcp]     # + MCP server
pip install startup-valuation[dev]     # + pytest, ruff, mypy

Quick Start (≤ 5 minutes)

Fastest path: one line — pip install "startup-valuation[mcp]" — then call any *_valuation tool (Scorecard, VC method, Black-Scholes, scenario analysis).

Python Library

from startup_valuation.core import scorecard_valuation, vc_method_post_money
from startup_valuation.advanced import black_scholes, scenario_analysis
from startup_valuation.types import Scenario

# Scorecard Method (pre-revenue startups)
result = scorecard_valuation(
    average_valuation=1_500_000,
    weights=[0.30, 0.25, 0.15, 0.10, 0.10, 0.05, 0.05],
    scores=[1.25, 1.50, 1.20, 0.75, 1.00, 0.90, 1.00],
)
print(f"Scorecard: ${result.value:,.0f}")  # $1,800,000

# Black-Scholes for real options (startup equity)
result = black_scholes(
    underlying=20_000_000, strike=5_000_000,
    risk_free_rate=0.05, volatility=0.40, time_to_maturity=1.0,
)
print(f"Option value: ${result.value:,.0f}")  # $15,240,000

# Scenario Analysis
scenarios = [
    Scenario("bull", 0.20, 10_000_000),
    Scenario("base", 0.60, 5_000_000),
    Scenario("bear", 0.20, 1_000_000),
]
result = scenario_analysis(scenarios)
print(f"Expected value: ${result.value:,.0f}")  # $5,200,000

MCP Server (for AI Agents)

The server exposes 14 tools, each folding a family of formulas behind a method argument — probability, time value, CAPM, core pre-revenue methods, options, comparables, SaaS, marketplaces, fintech, biotech, hardware, international, stakeholder equity, emerging methods, and a triangulated full analysis.

Local (stdio):

pip install "startup-valuation[mcp]"
startup-valuation-mcp          # console script installed with the [mcp] extra

**Prompts and resources.** Besides the 14 tools, the server offers three guided
prompts (`value_pre_revenue_startup`, `value_saas_startup`, `model_funding_round`)
and a machine-readable method catalog at `startup-valuation://methods`, so agents
can see every method's required parameters before calling a tool.

# or: python -m startup_valuation.mcp
# or ephemeral, no clone: uvx --from startup-valuation startup-valuation-mcp

Hosted (Streamable HTTP) — no install, no API key:

https://startup-valuation.ascent-partners.com/api

OpenCode — add to opencode.json:

"startup-valuation": {
  "type": "remote",
  "url": "https://startup-valuation.ascent-partners.com/api",
  "timeout": 60000
}

Claude Desktop / Cursor — add the HTTP URL https://startup-valuation.ascent-partners.com/api as an MCP server, or run the stdio entrypoint above.

MCP Registry — published as com.ascent-partners/startup-valuation (manifest: server.json) and listed on Glama and the Official MCP Registry. The glama.json file holds the Glama maintainer entry.

AI-agent skills

Copy the skills/ directory to your agent's skills folder:

• valuation-core — Scorecard, Berkus, VC Method, Risk Factor Summation
• valuation-foundations — Probability, time value, CAPM, comparables
• valuation-advanced — Black-Scholes, Binomial, Monte Carlo, Scenario Analysis
• valuation-industry — SaaS, Biotech, Fintech, Marketplace, Hardware
• valuation-stakeholder — Dilution, OPM, PWERM, Liquidation Preference
• valuation-emerging — SAFE, Crypto (MV=PQ), ESG, Metcalfe's Law

Valuation Methods by Category

Category	Methods	Chapter
Probability	Expected value, joint probability, Poisson	2
Time Value	PV, NPV, annuity	2
CAPM	CAPM, portfolio beta, startup-adjusted	2
Core	Scorecard, Berkus, Risk Factor, VC Method	3
Advanced	Black-Scholes, Binomial, Monte Carlo, Scenario	4
Comparables	P/E, P/S, EV/EBITDA, regression-adjusted	5
SaaS	LTV, CAC, NRR, Magic Number, Rule of 40	11
Biotech	rNPV, decision tree, peak sales, pipeline	11
Fintech	Payment revenue, lending, neobank, network effects	11
Marketplace	GMV, take rate, liquidity, network density	11
Hardware	TRL-adjusted, break-even, P-weighted DCF	11
International	PPP, CRP, currency-adjusted DCF, Damodaran	12
Stakeholders	Dilution, OPM, PWERM, liquidation, synergies	13
Emerging	SAFE, MV=PQ, ESG, Metcalfe's, data moat	14

The valuation ladder

flowchart LR
    IDEA["Idea / pre-revenue"] --> SC["Scorecard · Berkus · Risk Factor"]
    SC --> VC["VC Method · OPM · PWERM"]
    VC --> ADV["Black-Scholes · Binomial · Monte Carlo · Scenario"]
    ADV --> COMP["Comparables: P/E · P/S · EV/EBITDA"]
    COMP --> IND["Industry: SaaS · Biotech · Fintech · Marketplace · Hardware"]
    IND --> R["ValuationResult (value · assumptions · sensitivity)"]

Distribution

flowchart LR
    PKG["startup-valuation<br/>14 modules · 80+ functions"] --> L["pip install startup-valuation"]
    PKG --> MP["pip install startup-valuation[mcp]<br/>→ 14-tool MCP server"]
    MP --> T1["stdio (local)"]
    MP --> T2["hosted Streamable HTTP"]

Why This Library?

• Auditable — Every function returns ValuationResult with value, method, inputs, assumptions, and sensitivity analysis
• Textbook-accurate — All formulas verified against book example values with unit tests
• AI-ready — MCP server and Skills for seamless AI agent integration
• Industry-specific — Dedicated modules for SaaS, biotech, fintech, marketplace, and hardware startups
• Open source — MIT license, extensible, well-documented

Development

# Install dev dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run with coverage
pytest --cov=startup_valuation --cov-report=term-missing

# Lint
ruff check .

# Type check
mypy src/startup_valuation --ignore-missing-imports

Documentation

• API Reference: GitHub Pages
• Wiki (Theory & Derivations): GitHub Wiki
• PyPI: pypi.org/project/startup-valuation
• Chapter Index: Maps every function to its textbook chapter
• Examples: Interactive code snippets for each valuation category

Companion Textbook

Startup Valuation: A Comprehensive Guide to Valuing Fast-Growing Pre-Revenue Companies
Theory, Methods, Regulation, and Practice — Valuation in Practice Series by Ascent Partners
By Simon Mak · 338 pages · 15 chapters · 300+ exercises · 20+ real-world cases

Citing This Project

@software{startup_valuation_engine,
  author = {Mak, Simon},
  title = {Startup Valuation Engine},
  year = {2026},
  url = {https://github.com/simonmak-ascent/startup-valuation},
  license = {MIT},
}

Based on formulas from the Startup Valuation textbook.

Use with Context7

Up-to-date Startup Valuation Engine documentation is indexed on Context7, so coding agents can pull it into context on demand. With the Context7 MCP server or ctx7 CLI installed, name the library in your prompt:

use library /simonmak-ascent/startup-valuation for API and docs

License

MIT — see LICENSE.

---

By Ascent Partners — part of the Valuation in Practice Series.

If this saves you time, a ⭐ on GitHub helps others find it.