Best Time to Trade Gold (XAUUSD): 22 Years of Minute Data, Hour by Hour

00The short answer

• When does gold move the most? From 8 to 11 a.m. New York time (13:00 to 16:00 London, 15:00 to 18:00 on most MT4 server clocks): about 0.52% from high to low in an average hour, when London and New York overlap and US data come out at 8:30. The quietest hours are the evening before the daily break and late night.
• Does gold rise at night and fall by day? A little, in our data, but not in a way you can trade. Holding gold only from 13:30 to 08:20 made +0.0515% a night before costs; the same lengths of time started at other hours made nearly as much. Short gold by day lost money after costs.
• So the best time to trade gold depends on what you trade: the New York morning for movement and liquidity; no hour of the day for direction. All three time-of-day strategies we tested were buried.

01When does gold move the most? Hour by hour

The average range of each hour (high minus low, in % of the hour's open) from 2004 to 2026 and from 2020 to 2026, with the average change and its t-statistic. Hours are New York time; London is 5 hours later and most MT4 server clocks 7 hours later, except in the few weeks a year when the US and Europe change clocks on different dates.

17:00 to 18:00 New York is gold's daily break (no trading). The range columns were added for this article after the frozen test; nothing in the verdict depends on them.

The best session to trade gold for movement is the London and New York overlap: 8 a.m., 9 a.m., 10 a.m. New York are the three widest hours. Asian hours are calmer (about 0.24% an hour from 8 p.m. New York), and the hour before the break is the quietest (0.15%).

02Does gold rise at night and fall by day?

The idea has a source. Blose and Gondhalekar (Applied Economics Letters, 2014) found that COMEX gold futures from 1985 to 2012 rose on average from the day-session close to the next open and fell during the day session; a later paper with Kort (Journal of Economics and Finance, 2018) found the same in spot gold, miners and gold funds, weaker in later years. In 2026 the World Gold Council was reported describing gold's gains in Asian hours and drops in US hours.

We tested it on spot gold with the COMEX day session fixed in advance: day = 08:20 to 13:30 New York, night = 13:30 to 08:20 the next trading morning (Friday night runs to Monday).

• GN-NIGHT-L: buy at 13:30, sell at 08:20 the next trading morning, every day.
• GN-DAY-S: sell short at 08:20, buy back at 13:30, every day.
• GN-LS: both.

03How we tested it

Gold (XAUUSD), the 1-minute bid history of an IC Markets MetaTrader 4 demo server, June 2004 to 7 October 2026, on the New York clock (clock-corrected, since the rules depend on the time of day). Entries and exits at the open of the first minute at or after the stated time. Costs: $0.30 an ounce a round trip, as in every gold test on the site; no slippage. Financing (night trades cross the 5 p.m. rollover) is shown in two scenarios. Periods: 2004 to 2012, 2013 to 2019, and a holdout from 2020 that was already used by our earlier gold tests. Control: every trade kept its day, direction and length, and its start was moved to another hour, 500 times, so that gold's long rise is in both. Engine check: a self-test with hand-worked days (a weekend, a holiday, a missing minute) and an independent re-implementation that matched every trade of the three versions (gn_selftest_result.txt); it passed on its first run.

04Results

Net % of the entry price per trade, after costs, without financing.

Cumulative sum of trade returns by month (a simple sum, not compounded).

By period

* Already used before this test.

A night of gold made +0.0515% before costs and a day session -0.0093%, so gold did a little better at night. But the night is 19 hours and the day five: the night earned about 85% of the rise in about 79% of the time. The night trade lost money from 2013 to 2019 and made +110.6% from 2020, when gold rose fastest.

05Year by year and by weekday

Before costs the night was negative in 4 of 23 years (2012, 2013, 2015, 2022).

06Which hour does gold rise in?

By direction, the strongest hour was 6 p.m. New York (t 3.53), the first hour after the daily break, when Asia opens; the weakest was 9 a.m. (t -2.17), the first hour of the COMEX day session, followed by a rise from 10 a.m. (t 2.01). With 23 hours, about one would pass t = 2 by chance alone; 5 did, so gold probably does have a time-of-day pattern, but its size is a few thousandths of a percent an hour, far below a $0.30 round trip. Read this table as a description of 22 years, not as a trading rule: picking the best hours after seeing them is exactly the kind of result our gates exist to reject.

Clock check (added 23 Oct 2026): according to our clock audit (built on the 08:30 US data releases), the broker's file labels most weeks of 2004 to 2010 an hour early, and the clock correction used for this page does not move those weeks (see the data notes). We ran the whole test again on the audit's clock. The strongest hour is still 6 p.m. (t 4.38) and 9 a.m. is still among the weakest (t -2.08; 5 a.m. -2.14), but the rise from 10 a.m. does not hold (t 0.38); 11 a.m. is positive instead (t 2.10). The three widest hours are still 8 to 11 a.m. All three strategies stay buried: a night made +0.0449% before costs (+0.0515% on this page's clock), at the 71.2th percentile of other hours; the day short passes 1 gate instead of 2; and the long-short version ends at -0.0027% a trade after costs instead of +0.0045%. Details: gn_results_auditclock.txt.

The night in three parts

Average night, before costs, over 5,605 nights: +0.0083% from 13:30 to 17:00, +0.0153% across the daily break (17:00 close to 18:00 open) and +0.0301% from 18:00 to 08:20, Asian and London hours. Most of the night's rise came after the break, in line with the “Asian hours” version of the story.

07The swap: why holding gold overnight costs more than it looks

A CFD held through 5 p.m. New York pays the broker's overnight swap. At the IC Markets rate recorded on 12 October 2026 (long -57.03 points per lot per night, Wednesday counting three) applied to every night, the night trade goes from +145.2% to -248.2%. It is today's rate, not the historical one, but it shows the order of magnitude: a small nightly drift does not survive a daily swap.

08Is it the hours or chance?

The gates

The data-gap check

Night trades cross the weekend and holiday gaps of the file, so many touch a suspect one; without them every verdict stays the same.

09Verdict by variant

No proven edge (GN-NIGHT-L): positive, but not better than the same holding time at other hours, and negative in the middle period. Costs (GN-DAY-S): positive before costs (+0.0093% a trade), negative after. Fragile (GN-LS): positive overall, not without its ten best trades.

TWhat this shows, and what to take into your own tests

What the results show

• Gold moves the most from 8 to 11 a.m. New York time (1 to 4 p.m. in London), about 0.52% from high to low in an average hour, twice the late-evening hours. If you want movement, that is the best time to trade gold; if you want quiet, the hour before the daily break.
• Gold did rise more at night than by day from 2004 to 2026, but in proportion to the time: the night is about 19 of the 24 hours and earned about 85% of the rise. Holding it only at night was no better than holding it for the same length of time from other hours.
• The published "rises at night, falls by day" effect barely showed: by day gold slipped -0.0093% a session on average before costs, not enough to pay a $0.30 round trip. Overnight financing at today's swap turned the night trade from +145.2% into -248.2%.

What remains unknown

• Whether the effect is stronger in gold futures with their real session open and close, as in the original study (1985 to 2012). We tested spot gold, around the clock.
• Other session definitions, such as the London fixes or the World Gold Council's regional hours. We fixed one definition in advance and kept it.
• News days. The 8:30 a.m. US data explain much of the morning range; we did not split by calendar.

Why it was buried

None of the three versions picked better hours than chance: each sat between the 87th and the 91st percentile of trades moved to other hours of the same days (95 needed). The night trade lost money from 2013 to 2019 and made most of its gains from 2020, when gold rose fastest; the day short could not pay its costs; trading both every day was positive only because of its ten best trades.

What you can apply to your own tests

• Separate size from direction. Gold moves most in the New York morning, but that tells you nothing about which way.
• Compare a time-of-day rule with the same holding time at other hours, not with zero. A market that rises makes every long window look good.
• Price in the overnight swap before you hold a CFD through the 5 p.m. rollover every day.

A note on the numbers: drawdown figures are measured on closed trades unless a table says it includes open losses. Open losses inside a trade are not counted in a closed-trade drawdown, so real account risk is higher, especially for systems with no stop. Backtests are not live trading.

FAQCommon questions

What is the best time to trade gold?

If you want movement and liquidity, the London and New York overlap: from 8 to 11 a.m. New York time gold moves about 0.52% from high to low in an average hour (2004 to 2026). No hour of the day gave a tradable direction after costs in our test.

What time does gold move the most?

Between 8 and 11 a.m. New York time (13:00 to 16:00 London, 15:00 to 18:00 on most MT4 server clocks), around the 8:30 a.m. US data releases and the COMEX open. The quietest hour is the one before the 5 p.m. daily break.

Is the Asian session good for trading gold?

It is calmer (about 0.24% an hour from 8 p.m. New York) and, in our data, the hour after the daily break was gold's strongest rising hour on average, but the effect is a few thousandths of a percent an hour, far below trading costs.

Does gold rise at night and fall during the day?

A little: from 2004 to 2026 a night of spot gold (13:30 to 08:20 New York) made +0.0515% before costs and a day session -0.0093%. But the night is 19 hours long; the same holding time at other hours made nearly as much, and the published 1985-2012 effect was much stronger than what we found.

How was the best time to trade gold tested?

On 22 years of 1-minute spot gold, with rules written and dated before the data was run, costs included on every trade, results split into development, validation and holdout periods, a random-entry control as the benchmark, and every variant reported. Read the protocol.

10Check our arithmetic

Code provenance: our own code, written from the rules in the specification.

11Data notes

Clock. The rules depend on the time of day, so the run uses the clock correction of our EMA 9/20 gold test, which moves 20 weeks of 2020 to 2026 in which the broker's server followed the European clock change. It does not move the weeks of 2004 to 2010 that our clock audit of 9 October 2026 found labelled an hour early (until 23 October 2026 this note said it did); section 06 shows the results on the audit's clock. Source. A broker's demo history of spot gold, bid prices, not COMEX futures; the original study used futures with a real session close and open. Time zones. New York hours throughout; London and server columns assume the usual 5- and 7-hour differences.

12Corrections

23 Oct 2026, correction: the data notes said the clock was corrected for the weeks of 2004 to 2010 that the file labels an hour early; the correction used moves 20 clock-change weeks of 2020 to 2026 instead. We re-ran the test on the clock of our clock audit, which does move the 2004-2010 weeks, and added the comparison to section 06: no verdict changed; on that clock the rise from 10 a.m. and the long-short version's small profit after costs do not hold. 22 Oct 2026, navigation: added links to the gold strategies guide, which now includes this test. No result changed. 22 Oct 2026, navigation: added a link to the new guide comparing trading costs across all our tests. No result changed. 22 Oct 2026, navigation: added a link to the new guide comparing trend following and mean reversion across all our tests. No result changed.

+More tests and guides

• Best gold trading strategy
• Gold opening range breakout backtest
• Noise area breakout gold
• H1 open trap strategy
• Trend following vs mean reversion: this test next to the others of its style
• Free prop firm challenge calculator
• What all the backtests found so far
• The backtesting protocol
• How to backtest a trading strategy without fooling yourself
• Backtesting glossary