Data discipline
A reference earns trust as much by what it refuses to publish as by what it shows. Cavport holds to a single standard: no fabricated quantities, no synthetic history, no retail figure dressed up as production value, and no quality grade invented from a marketing word. The list below is what Cavport deliberately does not show until it is verified. It is published openly, because the discipline is the product.
- Any species split of the 2025 output that is not resolved from a primary source (today about 64% is unresolved).
- Any country production figure based on assumption, or any figure rescaled to match a world total.
- Global Caviar Production Value as a published number. Today it is an internal estimate with a known downward bias.
- The retail-equivalent proxy without its proxy label, or placed next to production value without clear separation.
- Any “market size” figure presented as Cavport's own.
- Any price change, arrow or index line before two qualifying snapshots exist for that cell. There is currently one snapshot.
- The caviar price index, or the production and deflator indices, as live series.
- Origin-specific prices. Today about 16% of listings state an origin, all from seller text.
- Quality grades or grain classes that were not issued by the external certifier.
- Any “Cavport Reference List”, CQR or CMR naming applied to public retail data.
- Liquidity measures built from listing counts, or demand indices before real platform activity exists.
- CITES quantities presented as production or consumption, or biomass presented as caviar tonnage.
- Trade unit values presented as market prices.
- The earlier retail-regression origin premia and the earlier modelled species shares.
Each item becomes visible the moment its evidence exists: a second snapshot unlocks change and index lines, European first-sale cells unlock production value, certifier-issued grades unlock quality fields, and tier B production observations unlock species and country output. Until then, the space stays empty rather than filled with an estimate wearing a confident face.
This page is the public face of the methodology's suppression rules. See Methodology and Governance.