Methodology changelog

Price basis: the price each retailer displays

• Every listing is recorded at the price the retailer displays to shoppers in its own country, in its own currency. Where a page shows both a price before tax and a price including tax, the price including tax is recorded. We do not estimate what a buyer pays at checkout: no tax the retailer does not display, no import duty and no shipping is added.
• Tax is not adjusted retailer by retailer. Consumption tax is charged where the buyer lives, not where the retailer is based, so no single adjustment per retailer makes a cross-border comparison fairer. On stones listed by more than one retailer, the typical price at individual retailers ranges from about 0.6 to 3 times the US median, far wider than the 10 to 21 percent difference that tax makes.
• One UK retailer’s feed carried prices 20 percent below the VAT-inclusive prices it displays to UK shoppers. Its listings and their price history were restated to the displayed price. A restatement of three Belgian retailers made earlier the same day was reversed exactly, because they display their prices before VAT; the displayed price stands.
• Six stores on one e-commerce platform return prices for the visitor’s country rather than their own. Collection now requests each store’s home market. At three of them, 245 listings carried another market’s price, in two cases in a different currency from the one recorded; their current prices were corrected. The other three were unaffected.
• Releases published before this date stand as published and are not restated. A comparison that spans this date inherits a one-off step for the restated retailer’s listings.
• No currency pin, weight, gate or grade band changed: the scoring specification remains version 1.0.

Currency basis: Indian rupee and South African rand

• The pinned conversion rate for the Indian rupee moves from 0.012 to 0.0105 US dollars per rupee. The pinned rate for the South African rand moves from 0.055 to 0.0619 US dollars per rand.
• The index is computed in a pinned unit of account, so that a listing’s US dollar series does not move on days the retailer never repriced. Both pins had drifted from the market. The rupee pin was 14.7 percent high, so Indian listings read as more expensive than they were. The rand pin was 11.1 percent low, so South African listings read as cheaper than they were.
• Restating the rupee lowers the US dollar figure for affected listings by about 12.5 percent. Restating the rand raises it by about 12.5 percent. The retailer’s own price in its own currency is unchanged in both cases; only our internal conversion basis moved.
• Releases published before this date stand as published and are not restated. A comparison that spans this date inherits a one-off step in the basis for listings in those two currencies.
• Price history was restated with the listings, so charts show no artificial step on the rebase date.
• No other currency pin changed. No weight, gate or grade band changed: the scoring specification remains version 1.0.

Version 1.0

• Initial public specification: four pillars weighted 35/30/20/15.
• Quality denominator fixed at the maximum weight of 100 so that missing evidence contributes zero; the 60 percent data coverage gate introduced.
• Certification integrity gate introduced: certificate claims without color and clarity are capped at 30.
• Price plausibility guard introduced for tier-priced vendor feeds: value neutralised at 50 below 300 USD per carat.
• Grade bands fixed at A+ 90 to 100, A 80 to 89, B+ 70 to 79, B 60 to 69, C 50 to 59, D 0 to 49.

Versions before 1.0 were internal iterations on Carat Hunter and predate this site. The effective date above is the date the current methodology took effect on live scoring.