Gold and Silver Premiums After a Falling 30-Day Benchmark

Gold and silver benchmarks both ended below their starting levels in our latest 30-day request, but the recorded local-market premiums did not move as one group. This article describes the values stored by the site; it does not claim that the benchmark decline caused any particular premium.

The window and the benchmark move

The history endpoint returned 27 dated rows from September 16 through October 16, 2026. Gold moved from $4,263.855 to $4,156.37 per troy ounce, a change of −2.52%. Silver moved from $62.985 to $57.098, a larger decline of −9.35%.

Those percentages compare the first and last benchmark values in the returned window. They are not total returns and do not include financing, spreads or transaction costs.

Gold premium observations

The gold SGE observations remained close to the benchmark: five of 13 were below zero and none exceeded 5%. The MCX-related observations were much higher throughout this sample, with all 20 above 5%. That gap should not be read as a ready-made arbitrage opportunity because the normalized series can differ in contract, timestamp, taxes, delivery and currency conversion.

Silver premium observations

Both SGE and MCX silver observations stayed above 5% whenever a value was available. Even so, their levels were not interchangeable: MCX had the higher median and latest value, while the two series also covered different numbers of dates.

Why median and sample count matter

A single latest value can exaggerate or hide what happened during the rest of the month. The median shows the middle recorded observation and is less sensitive to an isolated extreme. The sample count is equally important: 8 LBMA-related gold values cannot support the same conclusions as 20 MCX values.

Missing observations were not converted into zero. The chart may connect adjacent valid points visually, but the calculations above use only rows where a market premium was present.

What can and cannot be concluded

The sample shows that a shared benchmark decline can coexist with very different local premium levels. It does not establish why those differences occurred. Market hours, stale closes, exchange rates, contract specifications, taxes, liquidity and local demand may all matter, and the site does not observe all of them directly.

To reproduce the figures, request /api/premium/history?commodity=gold&days=30 and the equivalent silver endpoint. Keep finite market premiums, report their count, median, minimum, maximum and latest dated value, and calculate the benchmark change from the first and last returned rows. Results will change as the rolling window advances.