REFERENCE · 50 STATISTICS · 14 STUDIES
Gold trading statistics
50 facts about how gold actually trades, every one measured from our own XAUUSD data. Each carries its sample size and links to the study that produced it, so you can check the number before you use it.
Free to cite · CC BY 4.0
Use any figure here, including commercially. Credit and a link are all that is asked.
Tech Kick (2026). Gold Trading Statistics. https://techkick.me/research/gold-trading-statisticsSections: How far gold moves · Gold and silver · Gold price crashes · Buying and holding gold · What moves the gold price · Gold and stocks · Gold and inflation · Gold and the dollar · What it costs to trade · Stop losses · Gaps and news · How gold falls · Popular signals, measured · Risk, reward and evidence
How far gold moves
Range is the starting point for every stop, target and position size.
$76.12
The median daily range of gold was $76.12 per ounce — 1.77% of price.
Source: Spread and volatility by hour · 312 sessions, 2025-08-01 to 2026-07-31
6.3×
A wide day is 6.3 times a quiet one: the 10th percentile daily range was $25.84 and the 90th $162.62.
Source: Spread and volatility by hour · 312 sessions
$768
The widest single day in the sample covered $768.39 per ounce.
Source: Spread and volatility by hour · 312 sessions
42%
March is the most volatile month for gold and September the calmest: March's average daily range runs 42% wider. Unlike return seasonality, this difference is statistically real.
Source: Gold seasonality · 3,871 days, 2014–2026
$2.08
The median 15-minute ATR(14) on gold is $2.08, ranging from $0.90 in the calmest tenth of conditions to $7.04 in the fastest.
Source: ATR stop loss on gold · 223,208 M15 bars
Gold and silver
The gold-silver ratio, every month since August 1971.
61.3
The gold-silver ratio averaged 61.3 from August 1971, with eight months in ten between 34 and 85. Its daily high was 123.5 in March 2020 and its low 14.0 in January 1980.
Source: The gold-silver ratio, tested · 661 months, 1971-2026
90%
When the ratio was 80 or above, silver beat gold over the next three years in 90% of months, against 44% for all months. Those months come from only 5 episodes.
Source: The gold-silver ratio, tested · 82 months at 80+
6 of 7
The gold-silver ratio rose in 6 of the 7 US recessions since 1973: silver did worse than gold in 6 of them.
Source: The gold-silver ratio, tested · NBER recessions, 1973-2020
Gold price crashes
Every fall of 20% or more in the London afternoon gold price since August 1971.
16
Gold has fallen 20% or more from a peak 16 times since August 1971, counting the fall under way: about once every 3.4 years.
Source: Gold price history · Daily prices, August 1971 to 2026
3.4 years
Every completed fall that stopped short of 30% was back above its peak within 3.4 years; the 7 deeper ones took 3.6 to 28 years.
Source: Gold price history · 15 completed falls
8 of 15
Buying gold on its first close 20% below a peak left the buyer ahead a year later in 8 of 15 falls: a coin flip, not a signal.
Source: Gold price history · LBMA PM prices, 1971-2026
Buying and holding gold
Every month since August 1971 treated as a purchase, held for one to twenty years.
61%
A gold purchase in any month since August 1971 was worth more one year later 61% of the time, and twenty years later 88% of the time. It beat inflation over twenty years 63% of the time.
Source: Is gold a good investment? · 661 monthly averages, 1971-2026
26 years
The longest wait for a gold buyer to get back to their purchase price was 26 years, from January 1980. 85% of purchases were back to their cost or higher within a year.
Source: Is gold a good investment? · 660 purchase months
60%
Putting money into gold all at once beat spreading it over the next twelve months 60% of the time, judged one year after the first purchase. Spreading it softened the worst result (-26.1% against -36.4%).
Source: Is gold a good investment? · 649 one-year periods
What moves the gold price
Gold against real yields, the dollar and the Fed, month by month.
-0.44
Correlation of gold’s monthly change with the change in the 10-year real (TIPS) yield since 2022, against -0.49 in 2003-2021. Neither the real yield nor the dollar predicted the next month’s gold move.
Source: Why is gold going up? · 56 months, FRED and World Bank
2.67x
In 2026 gold averaged $4,411, 2.67 times the $1,655 implied by its 2006-2021 relationship with real yields and the dollar. None of 2,000 resampled histories missed by that much.
Source: Why is gold going up? · 191 months fitted, 56 run forward
Gold and stocks
Gold against the S&P 500 with dividends reinvested, every month since August 1971.
27%
Gold beat the S&P 500 with dividends in 27% of ten-year holding periods since August 1971, and in 19% of twenty-year periods.
Source: Gold vs the S&P 500 · 541 ten-year periods
11.2%
The S&P 500 with dividends returned 11.2% a year from August 1971, gold 8.8%, and the S&P 500's price alone 8.3%: gold beat the price index, not the total return.
Source: Gold vs the S&P 500 · 661 months, 1971-2026
5 of 7
Gold rose during 5 of the 7 S&P 500 declines of 15% or more since 1971, but month to month the two were uncorrelated, and gold fell in the two worst months for stocks.
Source: Gold vs the S&P 500 · Monthly averages, 1971-2026
Gold and inflation
Gold against the US consumer price index, every month since the $35 peg ended in 1971.
55.2%
Gold beat US inflation in 55.2% of five-year holding periods since August 1971, and in 56.9% of ten-year ones.
Source: Is gold an inflation hedge? · 601 five-year periods
-15.1%
During the 2021-22 inflation surge, the highest US inflation since 1981, gold lost 15.1% after inflation: prices rose 13.9% while gold's average price fell.
Source: Is gold an inflation hedge? · December 2020 to December 2022
45 years
Adjusted for inflation, gold's January 1980 average took 45 years to regain, until February 2025. In between it lost 83% of its real value.
Source: Is gold an inflation hedge? · 661 months, 1971-2026
Gold and the dollar
The relationship traders watch most, measured over twelve years of daily data.
1.04%
Gold moves 1.04% in the opposite direction for every 1% move in the US Dollar Index — almost exactly one-for-one — with a daily correlation of -0.43.
Source: Gold vs the dollar · 3,905 trading days
19%
The dollar explains 19% of gold's daily movement, and it cannot predict it: today's dollar move against tomorrow's gold move correlates at 0.013.
Source: Gold vs the dollar · 3,905 trading days
-0.465
In 2024 gold and the dollar both rose, yet their daily correlation was -0.465 — the relationship held. It was negative in every calendar year since 2014.
Source: Gold vs the dollar · 13 calendar years
What it costs to trade
Every measured edge has to clear these numbers first.
$0.16
The median gold spread was $0.16 per ounce on a raw-spread account; 99% of the time it stayed at or below $0.36.
Source: Spread and volatility by hour · 70,546 M5 bars
$3.35
The widest spread recorded was $3.35 — about 21 times the median.
Source: Spread and volatility by hour · 70,546 M5 bars
0.0113%
Gold's median spread is 0.0113% of price, so any trading edge smaller than that is consumed by the cost of entering.
Source: Does RSI work on gold · 223,301 M15 bars
Stop losses
How often ordinary price movement reaches a stop, before any view on direction.
81.0%
A $2 stop on gold is reached by ordinary intraday movement 81.0% of the time.
Source: Spread and volatility by hour · 312 sessions
58.3%
A $5 stop on gold is reached by ordinary intraday movement 58.3% of the time.
Source: Spread and volatility by hour · 312 sessions
34.2%
A $10 stop on gold is reached by ordinary intraday movement 34.2% of the time.
Source: Spread and volatility by hour · 312 sessions
13.5%
A $20 stop on gold is reached by ordinary intraday movement 13.5% of the time.
Source: Spread and volatility by hour · 312 sessions
44.5%
In the fastest tenth of market conditions, a fixed-dollar stop won 44.5% of 1:1 trades while a volatility-scaled stop risking the same average distance won 50.0%.
Source: ATR stop loss on gold · 55,678 trades
Gaps and news
The moments when a stop may not fill at its price.
$7.60
The median weekend gap in gold was $7.60; 25 of 52 weekends gapped by more than $10 and the largest was $97.98.
Source: Gold weekend gap · 52 weekends
2.7×
The median US Non-Farm Payrolls release minute moved gold $4.81 — 2.7 times an ordinary minute. The largest moved $40.48.
Source: News trading on gold · 11 releases
2 min
Elevated volatility after an NFP release lasted a median of 2 minutes, and at most 24.
Source: News trading on gold · 11 releases
81.8%
The direction of the NFP release minute was still intact an hour later in 81.8% of releases, against 54.4% for randomly chosen minutes — on a sample of 11.
Source: News trading on gold · 11 releases vs a 30,000-minute control
How gold falls
The shape any drawdown rule is set against.
45.9%
Gold closed lower on 45.9% of trading sessions over a year in which it rose strongly.
Source: Gold losing streaks · 257 sessions
9 days
The longest run of consecutive down days was 9, across 61 separate losing streaks.
Source: Gold losing streaks · 257 sessions
84.5%
Gold closed below its own running high on 84.5% of sessions — even in a year it rose — which is why a trailing drawdown rule is live almost all the time.
Source: Trailing drawdown · 258 sessions
28.49%
The median distance below the running high was 9.47%; the deepest was 28.49%, or $1570.89 an ounce.
Source: Trailing drawdown · 258 sessions
Popular signals, measured
Each tested against a control, not against zero.
1.00
Retracements ended at the 61.8% Fibonacci level exactly as often as at the levels either side of it — a ratio of 1.00. 0 of 25 Fibonacci tests found a level pullbacks preferred.
Source: Do Fibonacci retracements work · 32,592 retracements
56.6%
Round-number price levels turned gold back 56.6% of the time. Levels chosen precisely because nobody watches them did so 56.1% of the time — the difference is statistically indistinguishable from zero.
Source: Round numbers as support and resistance · 35,002 decided approaches, $10 to $100 levels
9
The 50/200 daily golden cross occurred 9 times on gold between 2014 and 2026, and underperformed a randomly chosen day at every horizon tested.
Source: Does the golden cross work · 36 tests, 2 beat chance (1.8 expected)
7 of 10
RSI carried real information on gold — 31 of 75 tests beat the base rate against 3.8 expected by chance — but 7 of the 10 strongest results pointed to momentum, not the mean reversion RSI is taught as.
Source: Does RSI work on gold · 75 tests, 3 timeframes
51.7%
After breaking out of its overnight range, gold followed through 51.7% of the time and fully reversed 12.0% of the time.
Source: Session breakout · 234 breakouts
+3.61%
January was gold's strongest month on average at +3.61%, but the effect does not survive correction for testing twelve months at once (p = 0.073).
Source: Gold seasonality · 151 months
Risk, reward and evidence
What a result needs before it means anything.
0.021 pp
With no edge at entry, gold's win rate fell to offset every increase in risk-reward ratio almost exactly — a mean deviation of 0.021 percentage points from the arithmetic break-even line.
Source: Risk-reward ratio tested · 66,572 trades per ratio
$0.2165
Random entries at 1:1 lost $0.2165 per ounce per trade, against a median spread of $0.200: without an edge, you lose the cost.
Source: Risk-reward ratio tested · 2,000,000 M1 bars
416
A system winning 38% of trades at 2:1 needs about 416 trades before its record separates from luck with 95% confidence.
Source: How many trades to judge an EA · 200,000 simulations per point
How these were measured
Every figure comes from XAUUSD price data taken from a broker feed — not surveys, forecasts or third-party estimates. Most studies compare a result against a control rather than against zero, because gold rose strongly across most of the period and almost anything looks profitable in a rising market without one.
Several of the most useful figures are null results. That Fibonacci levels and round numbers attract no more reversals than arbitrary levels is exactly the kind of statistic that is hard to find anywhere, because it does not sell anything.
Some trading statistics that circulate widely — the share of retail accounts that lose money, for example — are deliberately absent. We did not measure them, and publishing them here would repeat the unsourced figures this page exists to replace. Full methods and raw data for every study are on the research page.
Common questions
How volatile is gold per day?
The median daily range of XAUUSD was $76.12 per ounce, or 1.77% of price, measured across 312 sessions. The spread between quiet and fast days is large: the 10th percentile was $25.84 and the 90th $162.62.
What is the typical spread on gold?
On a raw-spread account the median XAUUSD spread was $0.16 per ounce, and 99% of the time it was at or below $0.36. The widest recorded was $3.35. Spread varies by account type and broker, so treat these as one broker's measured figures rather than a universal constant.
How big are gold weekend gaps?
Across 52 weekends the median gap was $7.60. 25 weekends gapped by more than $10 and the largest was $97.98. A stop cannot fill inside a gap, so these are the moves a stop loss offers no protection against.
Can I use these statistics in my own article?
Yes. Every figure here is published under CC BY 4.0, which permits reuse including commercially, provided you credit the source. Linking to this page or to the individual study is the simplest way to do that, and each study publishes its full method and raw data so the number can be checked.
Where do these numbers come from?
From our own measurements of XAUUSD price data taken from a broker feed, not from surveys or third-party estimates. Each statistic links to the study that produced it, which states the sample size, date range, method and the controls used. Several are null results, included because they are the findings least likely to be found anywhere else.