What Is Actually in Our Three-Year Premium Dataset?

A chart labelled “three years” can still contain very different amounts of information for each market. We therefore queried the public history endpoints on October 14, 2026 and counted the values that were actually present, rather than treating the selected date range as proof of complete coverage.

Coverage at a glance

Both commodity responses ran from October 15, 2023 through October 14, 2026 and contained 1,083 dated rows. The benchmark appears on each row, but local-market observations are sparser because exchanges close on weekends and holidays, source archives differ, and some series only began being retained recently.

Why the lines do not have equal length

SGE and MCX have hundreds of usable observations because historical exchange or market records can be aligned with the benchmark and exchange rate for those dates. A missing point is left missing when the required inputs cannot be matched safely. The chart may connect neighbouring valid points for readability, but the API does not manufacture a price for the gap.

The LBMA-related series is different. Only 11 gold and 12 silver observations were retained at the time of this audit. That is not a three-year LBMA history, and it should not be read as one. A continuous official benchmark history is subject to source and licensing constraints, so this site shows only the observations it can document and does not fill the past with a proxy labelled as LBMA.

What a row represents

Each premium is a normalized comparison: the local-market value is converted to US dollars per troy ounce and then compared with the benchmark for the stored date. The result is useful for comparing the retained series, but it is not a quote for an executable cross-border trade. Differences in timestamps, contracts, purity, taxes, delivery and foreign-exchange costs remain outside the percentage.

How to read the three-year button

The button controls the requested window; it does not promise that every market has one observation for every calendar day. Check the visible start and end dates, the individual line coverage and the source label together. In particular, a short LBMA-related line and a long SGE line are not directly comparable as equally complete three-year samples.

Reproducibility

The counts came from /api/premium/history?commodity=gold&days=1095 and the equivalent silver request. We counted a market only when its premium object was present. Because the daily job continues to append observations and source corrections can be applied, later counts may differ. The audit date is part of the result, not a footnote.