Seasonal Gold Demand Patterns: What Diwali and Lunar New Year Actually Do to Premiums

The Claim That Won't Die

"Buy gold before Diwali; the premium spikes."

"Chinese New Year drives up Shanghai premiums every year."

Both statements are repeated in retail investment commentary the way weather forecasts talk about summer rain: as if they were obvious facts. Both statements contain some truth. Both are also more nuanced than the folklore version suggests, and the difference matters if you are actually watching the data for signal.

This piece works through what MCX and SGE premium data over the past decade actually show around India's two major gold festivals and around Chinese Lunar New Year. It separates the pattern from the noise, quantifies the effect where there is one, and flags the years the pattern broke down.

India: Diwali, Akshaya Tritiya, and Wedding Season

Indian gold demand concentrates around three windows: Akshaya Tritiya in April or May, Diwali in October or November, and the two extended wedding seasons in autumn and spring. Diwali alone accounts for roughly 25% of annual Indian gold jewelry sales in typical years. Combined with the wedding-season demand that overlaps it, the October-through-November window is the single largest physical buying period.

What the MCX premium data shows

Looking at monthly average MCX premiums against COMEX over 2015-2024, October and November premiums have averaged 0.7-1.1 percentage points wider than the calendar-year mean. That is a real effect, but it is smaller than most retail commentary suggests. A retail buyer expecting the Diwali premium to be dramatic — say, 3 or 4 percentage points wider than normal — will usually be disappointed.

The effect is also less consistent than folklore suggests. In four of the past ten years, the November average premium was actually below the annual mean, because either the rupee strengthened, or the pre-Diwali import quota was raised, or the government cut the duty within a few weeks of the festival. Any of these can dominate the seasonal effect entirely.

The pre-Akshaya-Tritiya window

Akshaya Tritiya, in late April or early May, is often marketed by Indian jewelers as an auspicious buying day. The corresponding effect on the MCX premium is genuinely small — typically 0.3-0.5 percentage points on the peak day, essentially gone within a week. This one is closer to marketing than macro.

China: Lunar New Year and the Post-Holiday Restocking

Chinese physical gold demand concentrates in the six weeks before Lunar New Year (which falls in late January or February) and again for a shorter window after the holiday when jewelry retailers restock. The pattern is driven by gift-giving culture during the New Year period and by inventory management by retailers.

What the SGE premium data shows

SGE premiums against COMEX in the four weeks before Lunar New Year have averaged 0.8-1.3 percentage points wider than the calendar-year mean over 2015-2024. This is a slightly larger and more consistent effect than the Indian Diwali pattern, but still smaller than dramatic retail-commentary versions.

The post-holiday restocking window is more distinctive. In the two weeks after Lunar New Year, SGE premiums have typically been very close to zero or even briefly negative before returning to normal ranges as retailer inventories rebuild. This creates a short-lived but visible dip that a data-watching reader can identify.

The 2020 and 2022 exceptions

Two recent Lunar New Year seasons broke the pattern entirely. In February 2020, the COVID lockdown hit exactly during the post-holiday restocking window, and SGE premiums collapsed into deep discount (as covered in our COVID retrospective piece). In February 2022, Chinese import quota tightening kept premiums elevated well beyond the normal seasonal window. Neither of these was a failure of the seasonal pattern to exist; both were larger forces overriding the pattern.

Cross-Market Comparison

Why the Effect Is Smaller Than Folklore Suggests

Three structural reasons keep the visible seasonal premium effect smaller than retail commentary implies.

Anticipatory imports

Both Indian jewelers and Chinese refineries know the seasonal demand is coming. They begin importing metal weeks or months before the demand peaks. By the time the retail buyer is looking at the premium chart, the peak importing has already happened. What the chart shows is the aftermath — the residual pressure that anticipatory imports could not fully absorb.

Currency and macro drift

Over any given seasonal window, currency and macro-driven noise can be larger than the seasonal signal. As covered in our currency-effect piece, a 1% rupee move against USD is enough to swamp a 0.7 percentage point Diwali premium effect entirely.

Policy interventions

Governments know festival demand is coming too. India has repeatedly used pre-Diwali windows to announce duty changes designed to influence import flows. When a duty cut lands three weeks before Diwali, the mechanical duty effect (a premium narrowing) can overwhelm the seasonal demand effect (a premium widening). This is not a market failure; it is policy responding to seasonality just as private actors do.

How to Use Seasonal Patterns

Given the modest, inconsistent, and easily-overwhelmed nature of these effects, we recommend treating them as follows:

What This Site Enables

The historical premium chart on our /market/sge and /market/mcx pages goes back multiple years. Readers who want to verify the seasonal patterns described above can navigate to October-November of past years for MCX or the January-February window for SGE and observe the actual behavior. Because the data is displayed unsmoothed, the seasonal patterns appear as they actually manifest, with the year-to-year variation intact.

For readers building longer-term views on Indian or Chinese gold demand, plotting the year-over-year change in the November MCX premium or the pre-Lunar-New-Year SGE premium against absolute retail sales data (from the World Gold Council and CGA) gives a more informative picture than either dataset alone.