monarchjuno Numeric Data QC
Use this procedure when numeric data, valuation inputs, market anchors, portfolio exposure, or risk limits affect readiness.
Required output shape:
- source date and retrieval date
- period alignment
- units and currency
- share count and per-share consistency when relevant
- source versus derived versus assumption labels
- current price or market anchor with as-of posture when relevant
- formula sanity checks
- field-level gaps and the claims or actions they limit
Quality floor:
- Do not mix periods, units, currencies, or share bases silently.
- Label derived calculations and assumptions.
- Do not fabricate prices, metrics, formulas, or retrieval timestamps.
- Judge freshness against the requested as-of and the source's expected observation or publication cadence; an unavailable future period is not stale data.
- Preserve valid fields and downgrade only the conclusions or actions that depend on missing, stale, conflicting, or ambiguous inputs.
Put the data-quality result near the top of the artifact so synthesis can see whether numeric claims are decision-ready.