How Currency Movements Actually Affect Gold Premiums (And How To Read Them Correctly)

The Math That Trips People Up

Every regional gold premium you see quoted — SGE against COMEX, MCX against COMEX, LBMA against COMEX — is a ratio that includes at least one currency conversion step. This means the number moves not just when the metal price changes in the local market, but also when the local currency moves against the U.S. dollar.

The trap is that these two causes look identical in the final number. A 2% widening of the MCX premium looks the same whether it came from Indian jewelers paying more per gram in rupees or from the rupee depreciating 2% against the dollar. Getting this distinction wrong leads to reading physical demand into what is actually pure currency arithmetic.

This guide walks through the arithmetic, gives you a decomposition method you can apply to any daily premium move, and works a real 2024 example to make it concrete.

The Underlying Formula

The premium of a local market M against COMEX is computed by first normalizing the local price to USD per troy ounce, then comparing to the COMEX price.

The "Local_M in USD/oz" step is where currency enters. For MCX, quoted in INR per 10 grams, the conversion is:

If the USD/INR rate moves from 83.0 to 84.7 (roughly 2% rupee weakness), and the underlying INR-per-gram price stays exactly the same, the MCX price in USD/oz drops by about 2% — which widens the MCX premium against COMEX by close to 2%.

Nothing has changed about the actual level of Indian gold buying. The Indian market is pricing gold at the same rupee level. All that has happened is that the reference frame we express the premium in — the U.S. dollar — has strengthened.

A Decomposition Method You Can Apply

Any daily change in a regional premium can be decomposed into three parts:

The residual is the number that has physical-market meaning. Everything else is arithmetic pass-through from other markets.

A Worked Example: MCX in March 2024

Consider a hypothetical but representative sequence of daily readings from March 2024.

Between Day 1 and Day 2, the MCX premium widened by 1.2 percentage points. A headline reader might interpret this as strengthening Indian demand. Decomposing:

• COMEX component: +0.69% higher COMEX tightens the premium by that much. Adjusted change contribution: -0.69pp.
• Currency component: 0.60% INR weakness widens the premium by that much. Adjusted change contribution: +0.60pp.
• Residual: 1.2pp actual - (-0.69 + 0.60) = 1.2 - (-0.09) = 1.29pp.

So of the 1.2pp headline widening, about 0.9pp was mechanical (COMEX and currency) and about 1.29pp was the local demand residual. In this case the direction reading is confirmed — the Indian market did tighten in real terms — but the magnitude of the "real" move was different from the headline suggests.

Now Day 2 to Day 3. Headline MCX premium changed by -0.2pp (tightening). Decomposing:

• COMEX component: -0.91% lower COMEX widens the premium mechanically by that much. Contribution: +0.91pp.
• Currency component: 0.12% INR weakness widens by that much. Contribution: +0.12pp.
• Residual: -0.2pp actual - (0.91 + 0.12) = -0.2 - 1.03 = -1.23pp.

The headline says premium barely moved. The decomposition says Indian retail actually softened significantly — the mechanical forces would have widened the premium by more than 1pp, and only strong local softening kept the headline number near flat. This is the kind of insight the decomposition unlocks.

What This Means For Reading the Dashboard

The premium numbers on this site's dashboard include both the mechanical components and the residual. We do not decompose them for you because the raw premium is the number that actually matters if you are transacting in that market: an Indian buyer really does pay more when either the INR weakens or Indian demand tightens.

But if your question is "is Indian physical demand actually tightening?" the raw MCX premium is not the number to look at. The residual is. To compute it quickly:

Currency Impact by Market

MCX (India, INR)

Highest currency sensitivity of the markets we track. The rupee has depreciated against the dollar in most calendar years since 2010, meaning MCX premiums have a persistent mechanical widening tailwind independent of any change in Indian demand. A long-run rising MCX premium chart is partly a chart of rupee weakness.

SGE (China, CNY)

Lower currency sensitivity because CNY is managed. Day-to-day CNY moves against USD are usually under 0.3%, and the PBOC keeps a fairly narrow trading band. This means SGE premium residuals track local demand more cleanly than MCX residuals do — the currency noise is smaller.

LBMA (UK, GBP)

Moderate currency sensitivity. GBP can move 0.5-1.0% against USD on macro-driven days, which is enough to show up in LBMA premium readings. On UK CPI day or Bank of England day, expect part of any LBMA premium move to be currency.

What the Site Does For You Automatically

We do not display the decomposition on the dashboard, but the methodology page explains our FX handling: whenever possible, we use the FX rate captured in the same upstream response as the metal price. This eliminates a common form of phantom premium — the kind that comes from mixing a stale FX rate with a fresh metal price. If you see a premium reading on the site, the FX component in it is properly time-synced with the metal price component. What is left for the reader to interpret is the currency-driven versus demand-driven split described above.