Gold Trading Sessions Explained: Asia, London, NY (Data)

Everything below comes from our own studies on XAUUSD spot data (Dukascopy, M5, January 2023 to July 2026): an hour-by-hour volume study across 653 normal trading days, and a session-direction study across 821 sessions. When we make a claim, the number and the sample size come with it.

The Session Clock

Gold trades nearly 24 hours a day, five days a week, but it’s useful to break the day into the three classic blocks. Times below are the common conventions; London and Paris shift together with European DST, so the ET column moves by an hour for a few weeks each year when US and EU clocks change on different dates.

Three structural anchors inside that clock matter more than the session boundaries themselves:

• London open, 09:00 Paris / 3 AM ET — the first genuine liquidity injection of the day.
• US data window, 8:30 AM ET / 14:30 Paris — where the major releases (NFP, CPI) land.
• NY morning block, 8–11 AM ET / 14:00–17:00 Paris — COMEX floor hours, US desks fully staffed, London still open. Remember this one.

The Volume Reality: New York Is the Main Event

Here is Gold’s actual hour-by-hour participation, from 653 normal days (tick-volume proxy, Paris time):

• 15:00 Paris (9 AM ET): 8.4% of the day’s volume
• 16:00 Paris (10 AM ET): 8.5% — the single busiest hour
• 14:00 Paris (8 AM ET): 6.5%

Those three hours — the NY morning block — carry roughly 23% of the entire day’s volume on their own.

Now the London open, the hour the folklore crowns as king: 09:00 Paris gets 4.5% of daily volume. Busy, absolutely — it’s one of the better hours of the European morning. But it’s about half of what a single NY-morning hour does. The “London is the big session for Gold” story appears to be imported from FX, where London genuinely dominates. On Gold, the center of gravity is New York — which makes sense, since that’s when COMEX futures, US macro data, and US institutional flow all converge.

At the other end, the dead zone is real: the hours starting 22:00, 23:00 and 00:00 Paris carry 1.5%, 1.0% and 1.4% of daily volume respectively. Spreads are wider, moves are thinner, and stop-hunting a book that empty is a professional’s game, not yours.

Practical consequence: if you can only trade one window per day, the data says it should be 8–11 AM ET (14:00–17:00 Paris) — not the London open you set an alarm for because a YouTube video told you to.

Data days rewrite the schedule

On the 119 days in our sample with a major US release, the profile concentrates violently: the 14:00 Paris hour jumps from 6.5% to 10.7% of daily volume and 15:00 jumps to 10.5%. The 14:30 Paris (8:30 ET) release compresses everything toward itself. Know your calendar before you know your setup — we cover release behavior in detail in the news-trading pillar, and one number worth internalizing now: on mornings before a major release, breakouts in the 09:00–14:25 Paris window were fake 73.1% of the time on normal days and 85.7% before NFP (n=28 breaks). Pre-data morning breakouts are traps more often than not.

Does Asia Predict London? We Checked 821 Sessions

The most common session “strategy” on the internet: read Asia’s direction, trade London in the same (or opposite) direction. We tested it on 821 sessions.

• Asia closed up on 53.2% of days — a mild upward drift, nothing more.
• London continued Asia’s direction 49.3% of the time [95% CI: 45.9–52.7].

That’s a coin flip with the confidence interval sitting squarely on 50%. Asia does not predict London. Full stop. Any strategy built on “Asia bullish, therefore London bullish” (or the contrarian version) is trading noise.

The extended results are just as sobering:

• NY continues Asia: 51.6% [48.2–55.0]. Nothing there either.
• Strong Asia sessions tilt NY slightly — around 55.7% continuation — but the confidence interval is wide and the sample thin. We report it because we found it; we would not trade it. Treat it as fragile, not as an edge.

Full write-up: Asia Session Doesn’t Predict London — We Checked 821 Sessions.

The 74.8% number that fools people

One correlation in the study looks spectacular: NY matches London’s direction 74.8% of the time [71.7–77.6]. Before you build a strategy on it, understand why it’s mostly an artifact: the NY and London sessions overlap for hours. A chunk of the price action being scored is literally the same price action, counted in both sessions. Shared hours produce shared direction. It’s a measurement overlap, not a tradable relationship — and it’s a perfect example of why session stats need to be read with their definitions attached.

Sweeps by Session: Timing Changes the Odds

Session structure also shows up in how liquidity sweeps resolve. From our sweep study (2,434 sweeps over 807 days): sweeps that occurred during the London morning reversed by end of day 44% of the time (n=304), and sweeps during the NY morning 49% (n=459) — both hovering around the coin-flip baseline that the full sweep study documents.

The session-specific standout: a sweep of the Asia session low, taken during a transition regime, reversed 65% of the time (n=63, CI excluding 50%). The Asia range isn’t a directional signal — but its boundaries are real liquidity, and what happens when they break is measurable.

A Session-by-Session Playbook

Asia (7 PM–3 AM ET / 01:00–09:00 Paris). Thin, rotational. Its real job for a day trader is to build the levels — the Asia high and low become the first liquidity pools of the coming day. Trading it live from Europe or the US is usually a sleep-deprivation strategy, not an edge.

London (3 AM–11:30 AM ET / 09:00–17:30 Paris). The 09:00 Paris open brings the first real flows and often the first sweep of Asia’s range. Respect it — but size your expectations to its 4.5% volume share, and remember the pre-data fake-break numbers before chasing a morning breakout on a release day.

New York (8:30 AM–5 PM ET / 14:30–23:00 Paris). The main event. Data at 8:30 ET, COMEX flow, the day’s two busiest hours at 9 and 10 AM ET. If your process only has room for one session, this is the one the volume data endorses. After the US morning, participation decays into the 22:00–00:00 Paris dead zone — a natural stopping point.

One scheduling caveat: on FOMC days the whole map inverts — roughly 79% of the day’s range happens after the 2 PM ET announcement (vs 20% in that window on normal days). Session habits built on normal days don’t apply; see the FOMC study.

Tracking all of this by hand — session boxes, Asia/London/NY highs and lows, sweep status, regime — is exactly the grunt work we automated in our session dashboard, and its Lite version is free.

What to Remember

• NY morning (8–11 AM ET) ≈ 23% of Gold’s daily volume. London open: 4.5%. Plan your day around the first number, not the folklore.
• Asia does not predict London (49.3%, n=821). Stop trading that.
• The NY–London 74.8% agreement is an overlap artifact, not an edge.
• Data days concentrate volume into 14:00–16:00 Paris, and pre-release morning breakouts fail 73–86% of the time.
• Session boundaries create the levels; regime decides what a break of them means.

For how sessions fit into a complete trading process, read How to Day Trade Gold: A Systematic Approach.

Delta Terminal ResearchStudies run on 3.5 years of XAU/USD spot data, developed hand-in-hand with a full-time Gold day trader. We publish the negative results too.