← Files CrunchbaseARCHIVED FILE
skills/crunchbase-market-mapper/references/calculation-spec.md
1.94 KB · Sep 30, 2026 · 22:50 UTC
# Calculation Specification Use these definitions for recency, market, and monitoring metrics. The optional `scripts/derive_metrics.py` helper implements the same rules. ## General rules - Fix one `as_of` date at the start of the task in the user's timezone. - Deduplicate organizations and rounds by UUID before calculating. - Use `value_usd` for money calculations. - Exclude `—` amounts from sums, medians, and concentration calculations; count those rounds separately. - Reject future event dates from calculations and surface them for review. - Keep raw retrieved values separate from derived labels. ## Full months since an event For event date `E` and as-of date `D`: `months = ((D.year - E.year) * 12 + (D.month - E.month)) - (1 if D.day < E.day else 0)` Do not divide elapsed days by 30. Funding-recency labels: - Fewer than 12 full months: `recent` - 12 through 17 full months: `intermediate` - 18 or more full months: `extended interval` - No returned event date: `—` These labels describe timing only. They do not establish fundraising intent, company health, or outreach priority. ## Trailing capital windows Using calendar-month subtraction: - Current period: `[D minus 12 months, D]` - Prior period: `[D minus 24 months, D minus 12 months)` For each period report: - Total round count - Rounds with a numeric USD amount - Rounds shown as `—` - Total disclosed USD capital - Median numeric round size - Funding-type mix by count When the prior value is zero, show percent change as `n/a` rather than dividing by zero. ## Concentration Calculate top-three concentration over the trailing 24-month window as the sum of the three largest numeric round amounts divided by total numeric capital in that window. Report `n/a` when total numeric capital is zero. ## Formation Bucket confirmed organizations by the year component of `founded_on`. Label the result “formation within the confirmed universe.” Do not extrapolate it to the whole market.
SHA-256: a28a4c44ac55bfab030249fd7a1a33583f934b5758991e0d2eecbbb8cd691777