Historical data do not identify one universal “best month” to buy gold. Here we compare 50 observations for each calendar month from 1975 through 2024. The mean, median, and frequency of gains answer different questions: seasonality helps describe the spread of past outcomes, but it does not become a ready-made buying signal.

The Metric We Measure

This analysis uses a series of 601 monthly observations from December 1974 to December 2024, in nominal U.S. dollars per troy ounce World Bank Commodity Price Data, 4 August 2026 release. December 1974 serves as the baseline for the first comparison. Over 50 full years, from 1975 to 2024, we get 600 month-over-month changes—exactly 50 for each calendar month.

In June 2025, the Gold series changed from the average London afternoon fixing to an average spot price based on daily values. The sample's end date in 2024 leaves this known definitional break outside the scope of our calculation; however, this does not guarantee the absence of any other changes over the half-century.

For each month, the change in the average price relative to the previous month was calculated as 100 × (P_t / P_(t-1) − 1), where P_t is the average price for the current month and P_(t-1) is for the previous month editorial methodology based on World Bank data. This is not a return based on end-of-month prices nor the result of a trading strategy: the monthly average does not represent the price of any single transaction. Costs, taxes, and specific entry and exit prices were not modeled here. The source values in the file are rounded to the nearest dollar.

The Mean Does Not Describe a Typical Year

Over the 1975–2024 period, January shows the highest mean change among the 12 months: +2.24%. Its median, however, is +0.95%. There were 31 positive changes out of 50 (62%) and 17 negative ones; in two other cases, the average prices for consecutive months were identical in the rounded source data editorial calculations based on World Bank data. The median describes the middle of the distribution, not a result that is guaranteed to repeat.

Below are all 12 months, showing the mean, median, and the number of positive changes in the 50-year sample.

Source: World Bank Pink Sheet, 4 August 2026; original editorial calculations and design. December 1974 is the baseline for 600 changes in monthly averages during 1975–2024, with 50 for each calendar month. q25–q75 use type 7 linear interpolation and are given in the CSV summary: the central half of observations, not a confidence interval or the full range. The chart shows the seasonal index: monthly average changes compounded from a baseline of 100, with the 0.55% average monthly drift removed so the within-year path is visible. All means, medians, min/max and counts are in the 12-row CSV summary. These are not month-end returns or a trading rule. World Bank terms include a third-party-rights caveat. Data CSV.

March illustrates why these metrics should not be conflated. The mean change is negative (−0.49%), the median is slightly positive (+0.05%), but there were exactly 25 Marches with gains out of 50. A positive median here does not mean that gains were more frequent than losses: the signs and magnitudes of individual changes describe different aspects of the distribution.

One Extreme Year Changes the Picture

The high mean value is significantly influenced by one unusual January. In January 1980, the average monthly price of gold rose by 48.35% compared to December 1979 editorial calculations based on World Bank data. This year remains in the main calculation as it is part of the historical record.

If we exclude only January 1980, 49 observations remain, and the mean drops from +2.24% to +1.30%. This is a sensitivity check, not a replacement for the main calculation. The median of the full sample, +0.95%, is less affected by the magnitude of a single outlier, but it also does not predict the outcome for the next January.

Historical Periods Are Not Alike

The picture also depends on the historical window. Let's compare two equal periods of 25 years each.

From 1975 to 1999, the mean January change was +2.12%, but the median was 0.00%. The price rose in only 11 out of 25 cases. In the 2000–2024 period, the mean was similar (+2.36%), but the median increased to +1.80%, and the number of positive Januaries grew to 20 out of 25 editorial calculations based on World Bank data. Here, similar means are accompanied by different medians and frequencies of positive changes.

The Second Sample: Seasonality in the 1971–2026 Study

The editorially approved study of seasonality covers a broader period—from the early 1970s to 2026—and utilizes a different set of price series and sources. In it, gold's 56-year track record is characterized by an annualized growth rate of 7.81%, a standard deviation of 18.74%, and a Sharpe ratio of 0.457. A positive annual return was recorded in 35 out of 56 years, and a positive monthly return in 352 out of 679 observations (Curvo historical sample, annual performance comparison).

These figures do not replace the World Bank calculation in the chart above. The time horizon, price series, and aggregation method differ. For example, the expanded study yields an average return of +3.67% and 70% positive observations for January, whereas the calculation based on World Bank monthly averages for 1975–2024 shows +2.24% and 31 positive changes out of 50. For June, the expanded sample indicates an average of −1.68% and predominantly weak performance; the World Bank reports −0.09% and 23 positive changes out of 50. This is not a rounding error, but a reminder that conclusions regarding seasonality depend on the chosen series, period boundaries, and calculation methodology (12-month pattern review, XAU/USD seasonal sample).

Both samples support a single, cautious conclusion: the calendar effect is a probabilistic backdrop, not an autonomous price driver. Winter momentum and autumn demand appear noticeably in history, but they do not mechanically repeat every year.

What Can Create a Seasonal Backdrop

Seasonality has both physical and financial dimensions. In India, demand strengthens ahead of Diwali, Dhanteras, and the wedding season; in China, buying can rise ahead of the Lunar New Year. These periods are capable of supporting the physical bullion market, particularly when prices and consumer incomes permit volume expansion (holiday demand review).

At the same time, prices are influenced by institutional rebalancing at the start of the year, exchange liquidity depth, and ETF and futures flows. Consequently, the same calendar month can yield opposite outcomes in different years. Physical demand forms a foundation, but it does not negate the impact of interest rates, currencies, and large-scale capital flows.

When Macroeconomics Overwhelmed the Calendar

Historical episodes clearly illustrate the limitations of a seasonality-based approach:

• In 1981, gold lost 32.60% amid tight Federal Reserve policy and rising real rates. Traditional physical demand failed to reverse the bearish trend.
• In 2013, the tapering of quantitative easing expectations and institutional capital outflows from gold ETFs were accompanied by a 28.30% decline over the year.
• In 2022, gold declined for seven consecutive months from March to September, even though the end of this period is typically associated with a revival in Asian demand. Rate hikes and rising real yields proved stronger than the calendar factor.
• In 2020, the pandemic and monetary stimulus shifted the price peak toward summer: the macro shock disrupted the standard sequence of seasonal phases.
• In 2024–2026, easing financial conditions, central bank purchases, and sovereign demand reinforced autumn-winter momentum. In the editorially approved study, two estimates of annual growth are provided for 2025—+65.10% and +66.50%—across different price series; these must not be conflated into a single universal figure.

The point of these examples is not to search for exceptions, but to establish the proper hierarchy of factors. Real interest rates, the US dollar, central bank policy, fiscal risks, and institutional liquidity flows typically dictate the market regime. Seasonal demand can amplify an existing momentum or cushion a decline, but it rarely determines direction on its own (World Gold Council data, study on nonlinear dynamics between gold and the dollar).

Timing or Regular Purchases

Attempting to wait for the "ideal month" requires not only accurately assessing the seasonal backdrop, but also correctly anticipating rates, the dollar, geopolitics, and central bank actions. During a strong trend, the anticipated summer dip may fail to materialize; during a policy tightening cycle, autumn demand may not halt a decline.

A dollar-cost averaging (DCA) strategy addresses a different objective: a fixed amount is invested at regular intervals regardless of the current calendar month. At lower prices, it acquires more metal; at higher prices, less. This approach neither guarantees a profit nor protects against an extended downturn, but it reduces the dependence of the outcome on a single entry date.

Approach	Potential Advantage	Primary Risk
Waiting for a seasonal window	Allows for increased buying if a drawdown actually occurs	The drawdown may not materialize, causing the investor to miss part of the trend
Lump-sum purchase	Immediately establishes the desired allocation of metal in the portfolio	The outcome is heavily dependent on a single entry price
Regular DCA purchases	Distributes entry-point risk over time	Does not eliminate market risk and may underperform a well-timed lump-sum entry
Combining a baseline schedule with a reserve	Maintains discipline while preserving capital for notable corrections	Requires pre-defining rules and refraining from emotional alterations

The editorially approved study considers a 15–20% allocation to physical gold as one possible option for portfolio diversification. This is not a universal benchmark: an appropriate allocation depends on the horizon, liquidity, expenditure currency, risk tolerance, and other assets of the investor.

How to Use the Calendar

In our sample, January and September have the two highest mean changes, but this does not constitute a buying rule. Another analysis, published by the World Gold Council on 20 December 2023, found a January mean of 1.79% for the LBMA Gold Price PM in USD from January 1971 to November 2023 WGC research. Its data series, period, and calculation method differ from ours; the results shown above are independent calculations based on World Bank monthly averages, not a reproduction of the WGC's chart.

These observations describe the past. The frequency of positive changes should not be taken as a reliably established probability of a gain in the coming year. The interquartile range for January over the 1975–2024 period stretches from approximately −1.77% to +4.30% editorial calculations based on World Bank data. This represents the central half of the distribution, not the full range or a confidence interval for the mean.

Seasonal analysis helps in understanding an asset's historical risks and volatility, but it does not provide ready-made signals for buying or selling.