recoup-catalog-estimate-value

Imported from recoupable/skills (skills/recoup-catalog-estimate-value/SKILL.md). Install upstream with npx skills add recoupable/skills --skill recoup-catalog-estimate-value. Copyright stays with the author.

Recoup Catalog — Estimate Value

Value a catalog/album/recording from public/streaming data alone — the clean-room cousin of a full valuation. Deterministic scripts ship in scripts/ (invoke relatively); methodology + API contract in references/.

Procedure

• python3 scripts/fetch_album_tracks.py — pull the catalog from public sources.
• python3 scripts/estimate.py — estimate streams→revenue→value per references/methodology.md (uses references/recoup-api.md).
• python3 scripts/build_report.py — assemble the estimate report (needs pip3 install reportlab).

Be explicit that this is a public-data estimate (wider error bars than a seller-file valuation), and state the assumptions. For a real data room with seller files, use recoup-catalog-review-deal instead.

Guardrails

• estimate ≠ valuation — public-data estimates carry wider error bars; say so.
• State assumptions — never present an estimate as an audited valuation.

References & scripts

• references/methodology.md · references/recoup-api.md
• scripts/estimate.py · scripts/fetch_album_tracks.py · scripts/build_report.py — ship alongside this skill; invoke relatively.

Coverage and reproducible output

Freeze the input population, identities, measurement period, currency and rights scope. Read references/methodology.md before using any numerical default. They are illustrative, unverified scenarios, not a refreshed market rate card. Preserve dated sources and rationale in the config. Public metadata never proves the user's controlled interest.

The helper supports master-side scenarios only. Missing annual measurements produce null money, not zero; partial coverage produces a covered-subset scenario, never an unqualified catalog-wide value. A complete input subset still does not prove all platforms or rights. The modeled net factor already includes deductions: do not subtract fees a second time. Do not convert all-time YouTube views to annual owner income using a Spotify ratio.

For supplied observations, no API/account is needed:

python3 scripts/estimate.py --observations /private/work/observations.json --config /private/work/assumptions.json --out /private/work/estimate
python3 scripts/build_report.py --estimate /private/work/estimate/estimate.json --out /private/work/estimate

The observation JSON has tracks (unique id, title, sp_ttm, ttm_source, optional sp, captured_at, data_source) and explicit population_complete. Accepted annual sources are measured_365d or runrate_Nd from inspected underlying windows; never invent these labels for missing data. Keep the original evidence with the case. Network mode uses existing credentials; missing portfolio snapshots are only created with explicit --snapshot authorization. Reads may themselves have a billing gate; honor it.

Return the JSON workpaper and bounded summary/PDF. Preserve prior public scenarios when new contracts/statements support a deal review or accounting stage. Calibrate only comparable rights, services, currency, period and measurement coverage; one statement does not eliminate residual uncertainty. Run python3 -m unittest discover -s tests from this skill to check calculation and missing-coverage behavior.

For the development-only Catalog scenario contract, read references/scenario-contract.md. scripts/scenario.py reuses the existing model for one explicit Spotify-rate sensitivity; it is a local domain reference, not browser integration or a verified rate card.

The canonical browser ES module is scripts/scenario.mjs; read references/browser-scenario.md for runtime requirements, conformance and source pinning. Shared App/Context integration remains a separate consumer responsibility.