Stooq — XAUUSD Historical Gold Data

One row per trading session, seven columns, zero commentary. Stooq's XAUUSD series records gold spot against the US dollar as open, high, low, close and change — the classic OHLC shape that every charting library, backtest loop and risk model already speaks. At review the numbered table ran past 15,300 daily rows, and paging through the archive reached entries stamped March 1793, which puts this among the longest nominal gold price runs in Datadory's 1,744-dataset index.

Inside the gold industry catalog it plays a specific role: the time-series backbone. Ticker pages give you now; mineral statistics give you last year. This dataset gives you every session in between, measured the same way throughout — which is precisely what a chart axis or a drawdown calculation needs and rarely gets.

Get a sample of this dataset

Captured during the August 2026 review. Two consecutive sessions carry most of what you need to judge fit: the date format, the three-decimal dollar quoting, and how the change column packs both percentage and absolute movement into a single pipe-separated cell.

No.    Date         Open      High      Low       Close     Change
15315  20 Aug 2026  4522.985  4540.275  4450.995  4519.375  -0.08% | -3.66
15314  19 Aug 2026  4336.765  4524.295  4325.405  4523.035  +4.35% | +188.73

Reading them: the 19th was a violent session — gold added $188.73, just over four percent, and still closed a hair below its $4,540-style intraday ceiling set the following day. The 20th opened essentially flat, stretched to 4540.275, sagged to 4450.995, and settled down less than a tenth of a percent. Two days, one trend, two completely different volatility signatures — the kind of texture a close-only series flattens away.

Rows arrive newest-first: the reverse-chronological No. counter counts down toward the archive rather than up from it, so the oldest row in any retrieved window carries the highest number.

Seven fields, all verified against live rows during review:

• No. — reverse-chronological row number within the series; a stable handle for citing a specific session.
• Date — trading date in DD Mon YYYY form (20 Aug 2026).
• Open — session opening price of XAUUSD in US dollars (4522.985).
• High — session highest traded price (4540.275).
• Low — session lowest traded price (4450.995).
• Close — session closing price (4519.375).
• Change — movement versus the prior close, delivered as percent and absolute value in one cell (-0.08% | -3.66), so momentum math needs no derivation on your side.

The source interface also exposes a weekly rendering of the same symbol and a pre-modern archive segment whose quoting conventions predate modern spot practice. Those sit folded under additional fields on request rather than the base dictionary — ask for them with your sample and they arrive shaped like the rest of the table.

Coverage runs narrow and deep rather than broad and shallow:

• Geography: Global — one world gold spot series priced in US dollars, with no per-market or per-vault splits.
• Temporal: Latest row at review dated 20 August 2026; the modern continuous run spans roughly 15,300 sessions, and paging deeper reaches rows stamped March 1793 — nominal prices drawn from historical archives rather than electronic quotes, so treat the deep tail as context, not backtest fuel.
• Granularity: Daily, one row per trading session; a weekly view exists as the folded additional option described above.

API, files, or your warehouse. Daily, weekly, or hourly.

The series lands however your stack wants it — pushed to storage, served over an endpoint, or synced straight into your database. Name the fields and the cadence when you request the sample; the sample ships first either way.

• Investors and quants treat the OHLC spine as the default input for gold factor work: drawdown ladders, volatility regimes, correlation windows against equities and rates. Fifteen thousand sessions is enough history to test whether a signal survives more than one monetary era.
• Developers and builders wire it into charting components and portfolio dashboards because the seven-column schema never drifts — a renderer written against today's rows reads the archive unchanged.
• Journalists and academics reach for the deep tail when a story or paper needs centuries of context: gold through wars, suspensions and Bretton Woods, quoted on one consistent daily grid.
• Market researchers and product teams benchmark pricing pages, hedge clauses and commodity-linked contracts against the same close everyone else sees, keeping internal models honest.

Persona-specific breakdowns live on the gold pages for investors and quants, data scientists, journalists and academics and developers and builders.

Resolution and reach pull in opposite directions, and this dataset sits deliberately at the reach end.

A sub-second feed answers execution questions: where is liquidity right now, what did the last tick print. A two-century daily series answers survival questions: how deep have gold drawdowns actually run across regimes, how long did the 1970s re-rating take end to end, what does a decade of flat real returns look like on the calendar your CFO plans against. Merged multi-source histories answer these too, but every stitch point imports a basis error; a single publisher maintaining one symbol removes them by construction.

The honest trade-off is definitional: the pre-modern rows are nominal archive prices, not electronic spot, so cross-era comparisons deserve a footnote in anything client-facing. Keep the modern run for modeling and the archive for narrative, and the dataset over-delivers on both.

Provenance note — Stooq is a Poland-based financial data service that has kept the XAUUSD quote table as one of its longest-running public series; Datadory catalogs the result under the gold industry vertical.

Structure note — the change column's dual percent-and-absolute layout means a single parse pass yields both return and delta series; split on the pipe and both land typed correctly.

Comparability note — rows before the modern electronic era are nominal archival prints. They are invaluable for long-horizon charts and unreliable for high-frequency inference; label them accordingly in anything you publish.

Where to go next — for a second opinion on the same metal, Kitco's gold price and charts page pairs naturally; Kaggle's gold price data offers a modeling-ready extract; and the USGS Mineral Commodity Summaries add the supply-and-production fundamentals the price alone never shows.