hedgehog-finance/hogagent/company-valuation

Company Valuation Engine

Overview

Pure calculation engine — no external API calls. Financial data is queried by the caller (agent) via skills like hedgehog-company-index-data and passed as JSON params.

23 valuation methods (DCF includes 5 sub-methods):

Invocation

Use the same CLI parameter rule on every Agent and operating system:

1. When every business value is a non-empty, single-line string | finite number | boolean, pass it as a named argument (--key value or --key=value). Names are case-sensitive and are not normalized.
2. When any value is an object, array, null, multiline text, a numeric/boolean-looking string that must remain a string, or contains difficult quoting, write the complete parameter object as UTF-8 JSON and pass --params-file.
3. Agent-created parameter files must have a unique basename matching tmp-<skill-name>-<unique-id>.json, must not use the reserved .hedgehog/ directory, and must be removed after the call when no longer needed. UTF-8 BOM is accepted.
4. Do not inline nested JSON or combine flat arguments with a JSON/file payload. Create JSON with the Agent's file-writing capability, not echo, a shell heredoc, or PowerShell string assembly.

POSIX/Git Bash form: node '<script>' <method> --key 'single-line value' or node '<script>' <method> --params-file '<workspace>/tmp-<skill-name>-<id>.json'.

PowerShell form: node "<script>" <method> --key "single-line value" or node "<script>" <method> --params-file "<workspace>\\tmp-<skill-name>-<id>.json".

On Windows, use PowerShell or a verified Git for Windows Bash; cmd.exe is unsupported. Keep each command on one physical line. HogAgent marks its Windows shell as UNSANDBOXED; other Agents apply their own native permissions and sandbox.

List methods with node '<skill_dir>/scripts/<script>.mjs' --help. For example, a flat call can use:

node '<skill_dir>/scripts/relative.mjs' pe --marketCap 1000000000 --netIncome 80000000

When a method needs an array or object, write its payload to a unique file such as <workspace>/tmp-company-valuation-<id>.json, call it with --params-file, then delete the file. The payload blocks below describe that file's contents. Legacy positional JSON and --params remain compatibility-only.

Method Classification

Relative Valuation

Compare with industry peers to determine value.

Absolute Valuation

Forecast future earnings and discount to present value (intrinsic value).

Strategic Valuation

For disruptive tech companies or special scenarios.

Premium/Discount Factors

Adjust industry multiples based on company traits:

Premium factors (raise multiples):

• Market leadership / brand moat
• Core patents / high tech barriers
• High switching costs / strong stickiness
• Blue-chip customer endorsements

Discount factors (lower multiples):

• Single-customer revenue concentration
• Low R&D investment vs. peers
• Regulatory / policy risk

Defaults: low factor 0.8 (−20%); high factor 1.1 (+10%) for pe/pe-ttm/pb/ps/ps-ttm, 1.2 (+20%) for arr/p-active-user/p-gmv/ev-fcf. Customizable via *LowFactor / *HighFactor params.

Usage Recommendations

Use two valuation models for cross-validation (e.g. 1 relative + 1 absolute):

Data Preparation

No API calls. Caller must query data via other skills and pass as params.

Field Reference

TTM Calculation (PE-TTM / PS-TTM)

Important: Quarterly/semi-annual/9-month/annual financial reports contain year-to-date cumulative values (not single-quarter). TTM calculation accounts for this.

Two modes:

1. Direct: pass pre-computed ttmNetProfit or ttmRevenue
2. Auto: pass reports[] (with end_date + financial field), script computes TTM automatically

TTM rules (all based on cumulative values, subtraction removes overlap):

• end_date ends with 1231 → Annual: TTM = annual value
• end_date ends with 0331 → Q1: TTM = Q1_cum + (lastYear_annual − lastYear_Q1_cum) → this year Q1 + last year Q2-Q4
• end_date ends with 0630 → H1: TTM = H1_cum + (lastYear_annual − lastYear_H1_cum) → this year H1 + last year H2
• end_date ends with 0930 → Q3: TTM = Q3_cum + (lastYear_annual − lastYear_Q3_cum) → this year Q1-Q3 + last year Q4

API Reference

Relative Valuation (relative.mjs)

Common Optional Params (shared across methods)

1. pe — Static P/E

{"marketCap":300000,"netIncome":20000,"industryPE":18,"totalShare":50000}

2. pe-ttm — Trailing P/E (TTM)

// Direct mode
{"marketCap":300000,"ttmNetProfit":30000,"totalShare":50000,"industryPE":15}
// Auto mode (reports array)
{"marketCap":300000,"reports":[{"end_date":"20250331","n_income_attr_p":8000},{"end_date":"20241231","n_income_attr_p":25000},{"end_date":"20240331","n_income_attr_p":7000}],"totalShare":50000,"industryPE":15}

3. pb — Price/Book

{"marketCap":300000,"equityToParent":200000,"totalShare":50000,"industryPB":1.5}

4. ps — Static P/S

{"marketCap":300000,"revenue":50000,"industryPS":8,"totalShare":50000}

5. ps-ttm — Trailing P/S (TTM)

{"marketCap":300000,"ttmRevenue":120000,"totalShare":50000,"peerPS":2.5}

6. ev-ebitda

{"marketCap":300000,"totalDebt":50000,"cash":30000,"ebitda":40000}

7. ev-revenue

{"marketCap":300000,"totalDebt":50000,"cash":30000,"revenue":100000}

Same as ev-ebitda but with revenue instead of ebitda.

8. peg — PEG Ratio

{"pe":20,"earningsGrowthRate":0.25,"targetPEG":1.0,"netIncome":20000,"totalShare":50000}

9. arr — ARR Multiples

{"marketCap":500000,"totalDebt":50000,"cash":30000,"arr":60000,"industryARRMultiple":12}

10. p-active-user — Per-User Value (MAU/DAU)

{"marketCap":500000,"activeUsers":80000000,"userType":"MAU","industryValuePerUser":0.01}