MEASURED · 3,077 DAYS · 12 YEARS
Is gold more volatile on Mondays or Fridays?
Fridays, over the long run, for two reasons: the jobs report lifts Friday, and Monday is gold’s quietest weekday. Across 3,077 weekdays of XAUUSD daily bars from 2014 to 2025, the median Friday’s high-to-low range was 1.05 times an ordinary day for its month and the median Monday’s 0.89 times. Leave out the 133 Fridays with a US jobs report and the other Fridays ranged 0.98 times, level with Tuesday to Thursday and still above Monday.
Monday’s quiet is the steadier result: it was the quietest weekday in 10 of 12 years, and still the quietest with US holidays taken out. The latest year followed neither pattern. From September 2025 to August 2026, on the US clock, Monday ranged 1.05 times an ordinary day and Friday 1.02, and weekday labels shuffled within weeks matched the real Friday-Monday gap 47% of the time.
How a Monday was compared with a Friday
Twelve years of daily bars. The broker’s XAUUSDm daily bars from Exness, February 17, 2014 to December 31, 2025, the file behind our Thanksgiving study, which matched them to the five-minute bars: they are UTC days. So the Monday bar starts at 8 p.m. ET on Sunday (7 p.m. in winter), the first hour or two after the Sunday open sits in a separate Sunday bar, and the Friday bar ends at the 5 p.m. ET close, one to two hours shorter than the others.
One year on the US clock. 71,402 five-minute bars from the same account, 2025-08-01 to 2026-08-04, cut into gold days that run from 6 p.m. ET (5 p.m. CT; 22:00 GMT in US summer, 23:00 in winter) the evening before to 5 p.m. ET (4 p.m. CT; 21:00 or 22:00 GMT). Monday therefore includes the Sunday-evening open: the first bar printed at 6:00 p.m. ET on 33 of 49 Mondays and at 6:05 on the rest. Every Friday ended at 5 p.m. ET. Days missing more than an hour of bars were dropped (11: US holidays, early closes and the ends of the archive), and August 2025 serves only as the first baseline, which leaves 230 days from September 2, 2025.
Gold went from about $1,328 to $4,319 over the daily sample, so each day’s range is divided by the median range of the 20 full days before the Monday of its week. All five days share that baseline; 1.00 means an ordinary day for that month. The control shuffles each full week’s five values across the weekday labels, 10,000 times, keeping the week’s volatility and removing only which day it fell on.
Twelve years: Monday low, Friday high
| Weekday | Days | Median range, % of price | Median vs ordinary day | Same, no jobs-report days | Big days (1.5x+) | Quiet days (0.67x or less) | Widest day of its week |
|---|---|---|---|---|---|---|---|
| Monday | 616 | 1.02% | 0.89 | 0.89 | 14.3% | 24.0% | 15.9% |
| Tuesday | 618 | 1.14% | 1.00 | 1.00 | 15.7% | 14.9% | 19.9% |
| Wednesday | 618 | 1.17% | 1.01 | 1.01 | 18.3% | 14.9% | 19.9% |
| Thursday | 618 | 1.18% | 1.02 | 1.02 | 18.9% | 13.9% | 20.6% |
| Friday | 607 | 1.20% | 1.05 | 0.98 | 18.6% | 14.2% | 23.7% |
Median range as a share of the day’s opening price. “Widest day” counts the 598 weeks with all five days present; chance is 20%.
The order runs almost in a line from Monday to Friday. Monday had the most quiet days, 24.0% against about 14-15% for the rest, and was the quietest day of its week in 28.1% of weeks. Friday was the widest day of its week in 23.7%. None of the 10,000 shuffles produced a Friday-Monday gap as large as the real 0.16; 95% of them stayed under 0.05. In twelve years of data, that gap is not luck.
Friday’s lead over midweek is the jobs report
137 of the 143 Employment Situation releases from January 2014 to December 2025 came out on a Friday (BLS archive). The other 6: 4 on a Thursday in early July, and Thursday, Nov 20 and Tuesday, Dec 16, 2025, after the federal funding lapse moved the schedule (see our NFP release study).
On the 133 jobs-report Fridays in the sample, gold ranged a median 1.29 times an ordinary day, and 36.1% were big days. On the other 474 Fridays it ranged 0.98 times, with 13.7% big days. That is the whole of Friday’s lead over midweek: without jobs-report days, Friday sits with Tuesday (1.00), Wednesday (1.01) and Thursday (1.02).
Monday stays at the bottom either way. With every jobs-report day removed, the Friday-Monday gap in full weeks shrank to 0.10, and 5 of 10,000 shuffles matched it across 463 weeks.
Monday stays quiet with the holidays taken out
Five US public holidays always fall on a Monday (5 U.S.C. 6103). Gold trades on all of them, but on this feed four were short sessions in 2025-26: Martin Luther King Jr. Day, Washington’s Birthday, Memorial Day and Labor Day. Those four are the ones taken out here. The 47 in the sample ranged a median 0.56 times an ordinary day. Without them, the other 569 Mondays ranged 0.92 times, still below every other weekday, and 19.9% of them were quiet days.
The hour or two after the Sunday-evening open is not in these Monday bars. Those Sunday bars ranged a median 0.18 of an ordinary day. The one-year sample below does count them as Monday.
| Year | Monday | Friday | Loudest day | Quietest day |
|---|---|---|---|---|
| 2014 | 0.94 | 1.12 | Friday | Monday |
| 2015 | 0.97 | 1.03 | Wednesday | Thursday |
| 2016 | 0.83 | 1.10 | Friday | Monday |
| 2017 | 0.95 | 1.00 | Wednesday | Monday |
| 2018 | 0.90 | 1.01 | Tuesday | Monday |
| 2019 | 0.85 | 1.07 | Thursday | Monday |
| 2020 | 0.94 | 0.99 | Tuesday | Monday |
| 2021 | 0.82 | 1.15 | Friday | Monday |
| 2022 | 0.89 | 0.99 | Thursday | Monday |
| 2023 | 0.81 | 1.09 | Tuesday | Monday |
| 2024 | 0.82 | 1.08 | Friday | Monday |
| 2025 | 1.10 | 0.97 | Monday | Wednesday |
Friday ranged above Monday in 11 of 12 years. The exception was 2025, when Monday was the loudest day of the week. Each year has about 50 of each weekday, so single years are noisy; 2014 starts in mid-February.
The last year, on the US clock
| Weekday | Days | Median range | Median vs ordinary day | Big days (1.5x+) |
|---|---|---|---|---|
| Monday | 45 | $100.18 | 1.05 | 26.7% |
| Tuesday | 49 | $88.46 | 0.98 | 14.3% |
| Wednesday | 47 | $99.05 | 0.99 | 25.5% |
| Thursday | 45 | $91.93 | 1.02 | 20.0% |
| Friday | 44 | $87.24 | 1.02 | 20.5% |
All five weekdays sat within 0.07 of each other. In the 38 full weeks, shuffled labels gave a Friday-Monday gap at least as large 47% of the time, and a spread between the loudest and quietest day at least as large 89% of the time. In this year the weekday made no difference the test could detect.
The day boundary does not explain the change. Measured as UTC days, as the twelve-year file is, the same days gave Monday 1.07 and Friday 0.97. Switching clocks moved single weekdays by up to 0.10 in a sample this size, but it did not make Monday quiet. The jobs report also stopped being a Friday event: only 5 of the 9 reports in this sample fell on a Friday, and those 5 Fridays ranged a median 0.98. Five days is too few to read.
Where in the day: Monday’s evening, Friday’s morning
The five-minute bars can be cut by session. Each cell is the session’s median high-to-low span on that weekday, as a multiple of an ordinary day’s range.
| Session (ET / CT) | GMT, US summer | Mon | Tue | Wed | Thu | Fri |
|---|---|---|---|---|---|---|
| Asia6 p.m. to 3 a.m. ET / 5 p.m. to 2 a.m. CT | 22:00-07:00 | 0.63 | 0.61 | 0.54 | 0.60 | 0.50 |
| London3 a.m. to 8 a.m. ET / 2 a.m. to 7 a.m. CT | 07:00-12:00 | 0.33 | 0.37 | 0.37 | 0.36 | 0.33 |
| New York morning8 a.m. to noon ET / 7 a.m. to 11 a.m. CT | 12:00-16:00 | 0.52 | 0.55 | 0.56 | 0.63 | 0.69 |
| New York afternoonnoon to 5 p.m. ET / 11 a.m. to 4 p.m. CT | 16:00-21:00 | 0.31 | 0.37 | 0.43 | 0.39 | 0.35 |
Monday’s Asian session, which opens on Sunday at 6 p.m. ET, was the widest Asian session of the week at 0.63; Friday’s, which starts on Thursday evening, was the narrowest at 0.50. Friday’s New York morning was the widest of the week at 0.69, against 0.52 on Monday. Friday afternoon was not the quietest afternoon: 0.35, against 0.31 on Monday. Each cell rests on about 45 days of one year. In winter add an hour to the GMT times.
What to do with this if you run a bot
Do not switch a bot off because it is Monday or Friday. Over twelve years the median gap between the quietest and the loudest weekday was 0.16 of an ordinary day. A jobs-report Friday against any other Friday was 0.31. If a strategy should stand aside for news, key the filter to the release calendar, not to the weekday: 6 of 143 jobs reports in the sample were not on a Friday, and in the last year 4 of 9 were not.
Expect Monday to be a little smaller than other days. The median Monday ranged about a tenth less than an ordinary day (0.89; 0.92 without holidays), and 14.3% of Mondays were still big days. US Monday holidays are the real quiet: a median 0.56.
Do not hard-code a weekday rule. Monday was the quietest day in 10 of the 11 years to 2024, the loudest in 2025, and no weekday stood out in the year to August 2026. A rule fitted on the older data was fitted to a pattern that then stopped. Re-measure on your own broker’s bars before acting on it.
Friday’s real risk is the weekend gap. A position held past the 5 p.m. ET close meets the Sunday reopen (see the weekend gap study). Check whether your EA’s news and weekend settings use dates or weekdays, the Tech Kick Trading Bot included. If you trade from the US, read whether US traders can use MT5 trading bots first: the instrument measured here is a spot gold CFD.
Method and limits
One broker, one spot CFD, UTC days for the twelve years. Range is high minus low: how far price travelled in either direction. CPI and Fed days were not split out of the twelve years; in the one-year sample 11 of its 27 CPI, jobs-report and Fed days fell on a Wednesday. The shuffle tests use full weeks only. Every figure comes from one script, research-gold-monday-friday-volatility.py, run on the same archive as our other studies. For the hour-by-hour view, see spread and volatility by hour.
Common questions
What is the quietest day of the week for gold?
Monday, over the long run: a median 0.89 times an ordinary day from 2014 to 2025, 0.92 without US holidays, and the quietest weekday in 10 of 12 years. Not in the latest year: from September 2025 to August 2026 Monday was the widest day by median.
Is gold more volatile on NFP Fridays?
Yes. On 133 jobs-report Fridays from 2014 to 2025, gold ranged a median 1.29 times an ordinary day, and 36.1% were at least 1.5 times. On the other 474 Fridays: 0.98 times, and 13.7%.
Should my bot stop trading on Friday afternoon?
Friday afternoon is not unusually quiet, so quiet is no reason to stop. In the year measured, noon to 5 p.m. ET (11 a.m. to 4 p.m. CT) on Fridays spanned a median 0.35 of an ordinary day, against 0.31 on Mondays and 0.43 on Wednesdays. The question that matters on Friday is whether to hold over the weekend, and that is a gap risk.
Does the weekday matter more than the news calendar?
No. Over 12 years the quietest and loudest weekday were 0.16 of an ordinary day apart at the median; a jobs-report Friday and any other Friday were 0.31 apart, about twice as much.