Cavport is built on a small set of defined measures. This page states what each one is, what it needs before it can be published, and its current status. Several are deliberately not live: a measure is described here as mathematics long before it is shown as a number.
Price per kilogram falls as tins get larger. The gradient is estimated within products sold in several sizes, by regressing log price per kilogram on log pack weight with a product effect:
Fitted on this snapshot across 1353 price points from 255 products, β is -0.0706, which puts a 30 g tin about 28% higher per kilogram than a 1 kg tin. This is an analytical normalisation; it never replaces the observed bucket cells on the Master Board.
For each resolved cell of species by production country by product state, value is production in kilograms times a representative producer or first-sale price per kilogram, using only lanes L0 to L2. Retail prices are never used.
internal estimate No species-by-country producer-price cells exist yet, so any current figure is a single-world-cell estimate with a known downward bias: the only producer-price anchors available are Chinese, and European first-sale prices are probably higher. A Monte Carlo over the industry total and that price range produces an internal point estimate and an uncertainty band, but Cavport does not publish the number, because its own data discipline withholds a production value until the European first-sale cells exist. The method is shown; the figure is not.
An experimental proxy: production in kilograms times the public retail reference, converted to a one-kilogram equivalent with the pack-size gradient. It measures a different thing from GCPV and is always labelled "analytical retail-equivalent proxy, not production revenue, market size or GDP." It is never called GCPV, and it sits apart from outside market-size estimates (which measure yet another concept again).
Chain-linked Törnqvist indices. A cell enters a link only in years where it has prices in both periods, so no imputation is needed, and the Fisher index is computed in parallel as a cross-check. pre-launch Thresholds before launch: at least two consecutive years, at least 60% of production value covered by matched cells, and every contributing cell at tier B or better. There is currently one snapshot, so no index line is drawn.
Value intensity for a species is its share of value divided by its share of volume. It is purely economic and is never labelled quality or prestige. Beluga is the standing illustration: a small share of volume, a larger share of value. Concentration is the Herfindahl index of production shares, reported with the effective number of producers and the top-one and top-three shares. Growth is never computed across missing or interpolated years without saying so.
The RAPI-equivalent, in three stages. Stage one is fixed-cell indices, each set to 100 at its first qualifying snapshot, controlled by a matched-seller panel so that sellers or products entering and leaving never move the index. A cell is publishable only with at least eight matched sellers. Stage two composes species from fixed, published, annually revised pack weights. Stage three is a global composite on production and economic weights. The median is never indexed directly, because that would mix composition with price. pre-launch
The future verified reference is a hedonic time-dummy model on certified classes only, with a smoothed time effect, a random effect per source, shrinkage for thin cells, a Kish effective-sample-size correction, robust estimation against outliers, a 25% contributor weight cap, and a prediction band carrying parameter and residual uncertainty. Sufficiency fails closed and is separate from public coverage. pre-launch No verified figures are published.
Eligible mature female biomass by species and country, built from biomass, sex ratio, maturation age and mortality, times a roe yield rate and a harvest probability, with mandatory uncertainty intervals. research only It is never presented as fact and is not launched without peer-reviewed parameters.
Status: internal estimate internal, not published; pre-launch defined, awaiting the data gates; research only research only. What stays withheld is listed on Data discipline.