How we arrive at every number

Metals

Gold, silver, platinum and palladium are traded on a live market, so we quote real prices. They are fetched on demand from api.gold-api.com, with Yahoo Finance futures as a fallback, and cached for sixty seconds. Every figure is checked against a plausible range before publication, so a bad response from a source is discarded rather than shown. Prices are converted from troy ounces at 31.1034768 g. If the live feed fails, the page falls back to a daily snapshot and says so rather than passing a stale number off as current.

Scrap value is the metal weight multiplied by its purity and the spot price. Real buyers pay 70–90% of that, because they carry the cost of refining and take a margin.

Diamonds

Diamonds are different, and it matters: there is no exchange, no public tape and no live price for a diamond. Wholesale list pricing exists but is a paid, trade-only product. So every diamond figure on this site is modelled, and we label it that way everywhere it appears.

The model works in five steps:

1. A base price per carat that steps at each recognised size threshold. Price per carat is not smooth — it jumps at half a carat, one carat, two carats and so on, because those sizes are in demand.
2. Multipliers for colour, clarity, cut and shape, applied against a G colour, VS2, Very Good cut round brilliant baseline.
3. A lab-grown factor. Lab-grown stones are real diamonds, but production scaled enormously and prices collapsed, so they are priced at a fraction of an equivalent mined stone.
4. Side stones priced individually. Each melee stone is valued at its own tiny per-carat rate, never as one combined weight. Treating a 38-stone halo as a single 0.3 ct stone would overstate it several times over.
5. A certification discount. An uncertified stone carries grading risk for the buyer and trades below a GIA-graded equivalent.

Resale — the number nobody publishes

Retail prices are everywhere. What a diamond actually fetches secondhand is not published by anyone, because the trade has no particular reason to publish it. Our resale figures currently sit at 25–40% of retail for natural stones and 5–12% for lab-grown, with melee lower again since nobody buys used melee on its own.

Those bands are the weakest part of the model, and we would rather say so than pretend otherwise. They are being replaced, gradually, by something better: every visitor who is offered a price for a piece can tell us what it was. Those reports accumulate into the only public dataset of real secondhand jewellery offers we are aware of. As it grows, the bands stop being an assumption and start being an observation. Report an offer here if you have one.

What these figures cannot tell you

This is a price model, not an inspection. It cannot see your stone. Specifically it does not know about:

• Damage, chips, abrasion or a poor repair
• Fluorescence, which can help or hurt depending on the stone
• Brand. A signed Cartier or Tiffany piece can be worth multiples of its materials, and this model values materials
• Antique and period value. Melting an Art Deco setting for scrap destroys most of what it is worth
• Coloured stones, pearls and watches, which are not yet covered
• Your local market. Prices vary considerably by country and by season

For insurance, probate or a legal matter you will need a certified appraisal from a qualified appraiser. Use these figures to know roughly where you stand and to tell whether an offer or an appraisal is sensible — not as a substitute for one.

Corrections

If you work in the trade and something here is wrong, we would genuinely like to know. The model is only as good as its assumptions, and the fastest way to improve it is to be told where it is off.