How the Retail Premium Index is made

The affiliation

SilverTrade is affiliated with SD Bullion, a precious-metals retailer, and this phase of the index is built from SD Bullion's own records, published with its consent. That is the disclosure, and it is also the point: this history exists nowhere else. Figures are computed by SilverTrade under the methodology on this page, verified by independent recomputation before publication. See our Editorial Policy.

What we measure

The Retail Premium Index publishes the average premium paid on retail orders at SD Bullion, month by month, for each benchmark bullion product. A premium is what a buyer pays above the metal value of the piece, as a share of that value; a dollar figure is shown only where the record carries it. It includes the mint's and the distributor's charges as well as the dealer's, so it is not a measure of dealer margin.

Where the data comes from

SD Bullion's own completed retail transaction records, going back to January 2013. Cancelled and refunded orders excluded as of each publish; non-retail pricing tiers excluded where they can be identified: this is a retail index. The most recent months are refreshed at each publish; earlier months stand as recorded.

No customer information of any kind enters the pipeline.

The reference price

From March 2019, each transaction's premium is measured against the metal value SD Bullion recorded on that transaction at the time it was priced: a business record made in the ordinary course, not a quote pulled after the fact. For earlier years, see The years before March 2019 below. The index publishes premiums only, at monthly resolution, never the underlying metal prices themselves.

The years before March 2019

SD Bullion has sold bullion online for far longer than this index has measured it, and its order records go back that far. What those early records do not hold is the metal price at the moment of each sale: that was added to the systems later. For the earlier years we measure each sale against the market's closing price for that day instead, so the premiums shown before March 2019 are close estimates rather than exact figures. Where the two methods can be compared side by side, the closing price sits a little below the price recorded at the moment of sale, so the estimated years are likely to overstate premiums slightly. We show them here, named as estimates, so you can judge them for yourself.

For the round and bar categories, the figures average every brand sold in that size, not one product line, because that is the fairest picture of what a buyer actually paid. Months with very few sales are drawn like the others and left out of the averages and the high. Each coin series includes every quantity sold, from single coins to monster boxes, weighted by ounces, so it reflects what buyers paid across all order sizes. Where possible, the benchmarks leave out premium product lines and collector editions, so each series reflects what a buyer paid for the ordinary form of that product.

How the averages are computed

Each month's figure is weighted by ounces: every order counts in proportion to the metal it moved, so a large order counts for its size and small ones count for theirs. A plain per-order average is computed alongside it as a cross-check and reported in the verification record; where the two diverge meaningfully, that record says why. The headline figure for each metal is a composite: the median across the benchmark products of their monthly ounce-weighted averages.

Reading the figures

These are historical statistics, not price quotes or offers, and no single order should be expected to match a monthly figure. Orders varied by payment method; by order size (larger orders transact at tighter premiums, so the ounce-weighted figure sits below a simple per-order average and a smaller order will typically have paid more than the month's figure); by product; and by the day within the month.

What moves the figures

These figures move for many reasons besides the metal price, and no single month should be read as a precise measure. Payment methods differ, and the mix of them has changed over the years. Premium product lines, collector pieces and graded coins sometimes sit in the records alongside ordinary bullion, and we leave them out where we can identify them, not always. Bulk quantities such as tubes and monster boxes trade at different prices from single pieces. Some months are thin because volume fell; others are heavy because of deals and specials, or because a handful of buyers made very large purchases. Volume and trade purchases may be priced differently from single retail orders. A premium can also be negative: when the metal price moves very fast, some products have sold for less than the metal in them, and the series shows that as it happened. All of this is part of what a real sales record looks like, and the index reports it rather than smoothing it away.

Promotions and outliers

Promotional sales are excluded from the headline series and counted. No figure is trimmed or smoothed: a transaction is either included, or excluded under a stated rule, or individually reviewed, and the counts of each are part of the verification record.

Collector editions, graded coins, pre-owned lots and premium product lines are left out where they can be identified, so each benchmark reflects, as far as possible, the ordinary form of the product.

Product mapping

SKUs change over the years. Each SKU is mapped to its benchmark product type by a curated, versioned mapping reviewed against SD Bullion's product history; unmapped volume is reported, never guessed into a bucket.

The roadmap

The index grows into an anonymized multi-dealer panel: premiums tracked daily across major US dealers, counts published, names never. When the panel launches, its series will appear alongside this history as separately labeled series, measured against a different reference, and never spliced.

Revisions

Figures can be revised. When a correction is made it is published as a new version of the dataset, and the page shows the corrected figure with the date of the capture it came from. A month the method cannot compute is left out rather than filled in. Nothing here is a commitment about future publishes: what the index covers, and how it is built, may change.

How to cite

Cite figures as “per the Retail Premium Index (SilverTrade)” with the capture date. A copyable citation line appears under every chart.

Retail Premium Index (SilverTrade), silver composite, August 2026. silvertrade.com/premiums/

The methodology on this page is versioned; material changes are noted here with their effective date.

The Retail Premium Index is a free public resource published for educational purposes only. It is compiled from real transaction records and from market data we license, using the methods described on the methodology page, and we correct it when we find errors. It is not a price quote, an offer to buy or sell, or investment advice, and it may not match figures compiled by other dealers or from other sources, whose products, pricing and methods differ from ours. The methods, the products covered and the period covered may change over time; nothing about the continuation, consistency or method of the index is guaranteed, and any figure may be revised or withdrawn. Past premiums do not indicate future premiums. The figures come from one retailer's own records of completed orders, and for the months before March 2019 from estimates of what the metal was worth at the time of each sale.

Related instruments

• Miner margins: what it costs miners to produce gold, silver and copper.
• Gold-silver ratio: how many ounces of silver one ounce of gold buys, and where that stands today.

← Back to the index