MCP server for the Startup Valuation Engine — 45+ valuation tools implementing 80+ formulas from the Startup Valuation textbook. Includes probability, time value, CAPM, core valuation, advanced methods, comparables, SaaS, biotech, fintech, marketplace, hardware, and international valuation methods.
28 tools with consistent naming (valuation_* prefix) and mathematical descriptions. All tools have input schemas with typed arrays/numbers and descriptions. However, descriptions are terse (avg ~80 chars, well below the 194-char baseline for A+ tools) and lack actionable context for LLM selection. No output schemas documented. All tools are READ_ONLY (good), but error handling and recovery guidance are absent. Parameter descriptions are minimal, e.g., 'List of probabilities' does not explain valid ranges (0-1?), whether they must sum to 1, or what happens if constraints are violated. No examples of when to use each tool vs. alternatives.
Berkus valuation. V = Σ vᵢ (max $500K per factor).
Black-Scholes call option value. C = N(d₁)S - N(d₂)Ke^(-rT)
Calculate expected return using CAPM. E(R) = Rf + β(E(Rm) - Rf)
Decision tree EV. EV = Π pᵢ × Terminal Value
Calculate expected value for a discrete random variable. E[X] = Σ xᵢ × P(X=xᵢ)
GMV multiple valuation. Valuation = GMV × Multiple
Calculate joint probability of sequential events. P(total) = Π Pᵢ
Parameter descriptions lack constraint documentation. E.g., 'List of probabilities' does not specify valid range (0 - 1), whether they must sum to 1, or behavior if violated. LLMs cannot infer constraints from names alone.
Tool descriptions are terse (avg ~80 chars vs. 194-char baseline). They state WHAT the tool does but not WHEN to use it or how it differs from similar tools. E.g., 'valuation_expected_value' vs 'valuation_probability_weighted', both compute weighted sums but descriptions do not clarify the distinction.
No output schemas documented. Tools return dicts with 'value', 'method', 'inputs' fields, but LLMs cannot see this structure in the tool definition. Downstream tools cannot reliably extract or chain results.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
Lending fintech valuation. V = Loan Book × ROE × P/E - NPL Reserves
Calculate net present value. NPV = Σ Cₜ / (1+r)^t
Payment revenue. Revenue = Volume × Take Rate
Peak sales. Peak = Population × Penetration × Price × Compliance
Biotech pipeline valuation. V = Σ(Peak Sales × Multiple × P_success) / (1+r)ⁿ
Calculate portfolio beta. βp = Σ wᵢβᵢ
Calculate VC portfolio expected return. E[R] = Σ pᵢ × Rᵢ
Calculate present value. PV = C / (1+r)^t
Calculate probability-weighted expected value. E[V] = Σ pᵢ × Vᵢ
Regression-adjusted multiple. Multiple = β₀ + β₁g + β₂M + β₃S + β₄G
Risk Factor Summation. V = V_base + Σ(rᵢ × $250K)
Rule of 40. Score = Growth Rate + Profit Margin
SaaS LTV. LTV = (ARPU × Gross Margin) / Churn Rate
SaaS revenue multiple valuation. Valuation = ARR × Multiple
Scenario analysis expected value. E[V] = Σ pᵢ × Vᵢ. scenarios: [{name, probability, value}]
Scorecard valuation. V = V_avg × Σ(wᵢ × sᵢ)
Calculate startup-adjusted CAPM with additional premiums.
Marketplace take rate. Take Rate = Revenue / GMV
TRL-adjusted valuation. V = Market Size × Share × Margin × Multiple × (1 - TRL Discount)
VC Method post-money. Post = Terminal / ROI
VC Method pre-money. Pre = Post - Investment
No error handling or recovery guidance. Tools do not validate inputs (e.g., probabilities outside [0,1], mismatched array lengths, division by zero in discount rates). Failures will return raw exceptions instead of actionable error messages.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are READ_ONLY, but this is not declared in the schema. Modern MCP clients use annotations to optimize caching and retry logic.