Selling Gold Forward Is Not a Hedge: A Margin-First Structure for Gold Producers

Abstract

A 300,000 ounce-per-year gold producer. Revenue of $1.24 billion, all-in sustaining cost of $1,850/oz, margin of $687 million. It wants a floor on gold revenue and it wants to know what the floor costs.

The instinct is to sell forward. In 2026 that instinct was right about the price and still a bad hedge. Gold ran to a record $5,595 in January and traded $4,166 on 16 October — about 25% lower — while Brent went from $65.57 to $109.15, up 78.9% year over year. Over the same stretch the World Gold Council put global all-in sustaining cost at $1,785/oz in Q1’26, up 16% year over year and the 28th consecutive annual increase, with royalties and production taxes doubling from 6% of AISC in Q1’21 to 12% in Q1’26 and rising 85% year over year.

A producer that sold forward in January gave up the melt-up and left the cost line untouched — and the cost line is the one that ratcheted. A producer that sold forward in March locked a price into a cost curve that had not stopped climbing.

This note specifies a different structure: hedge the margin, not the revenue. A gold floor, an oil cost cap, and a local-currency hedge, sized by their contribution to margin variance rather than by notional, and built so the states where each leg pays are the states where the others do not. Below are the decomposition, the sizing rules, the regime switch, and the numbers on a representative mid-cap Australian operation.

1. The 2026 setup

Two risks define a gold producer’s margin: the gold price and the cost curve. In 2026 they ran opposite ways, and the cost curve never reversed.

Two details deserve emphasis. First, Q2 2026 was gold’s worst quarter since 2013. Gold fell 16% from the end of Q1, with the average price down 7.2% quarter over quarter, and the LBMA average still 37% above the prior year — a drawdown inside a record-setting year, not a bear market. Second, the cost side does not mean-revert. WGC’s Q1’26 numbers put US average diesel up 54% quarter over quarter and Perth wholesale diesel up 96%, with bunker fuel doubling and war-risk insurance rising in March.

The options market is not pricing the dispersion the spot market has produced. Gold’s 52-week range ran $3,283 to $5,595, a 70% high-to-low range, while 30-day implied volatility sat near 23%. That gap is the reason a floor is cheap and the reason it is not free.

So the round trip in price was partial and the round trip in cost was not. That asymmetry is the whole reason a revenue-only hedge is the wrong instrument.

2. Revenue risk is a margin problem

Margin per ounce is realized price minus all-in sustaining cost. A revenue hedge floors the first term and leaves the second untouched, which means it does not floor the thing the producer actually earns.

The AISC line is not a constant. It is labour, energy, royalties, sustaining capital, consumables and closure provisions. Two components behave like instruments:

Energy is a short position in Brent. Fuel and power are a high single-digit share of the cost base — OceanaGold puts diesel alone at around 6% of AISC — and the share moves with the energy price, not with the metal. Producers hedge it directly. OceanaGold reported around 80% of annual diesel consumption hedged, and WGC notes that major and intermediate producers were mostly insulated from the March price spike through fuel inventories, hedging, power purchase agreements and long-term procurement: Evolution Mining said its existing contracts prevented any disruption, and Newmont was unaffected. OceanaGold, Newmont and others still cautioned that a prolonged period of elevated fuel prices would bite harder in later quarters than in Q1.

Royalties are a short call on the gold price. Royalties and production taxes have moved from about 6% of AISC in Q1’21 to 12% in Q1’26, and royalty payments rose 24% quarter over quarter and 85% year over year in Q1’26 alone. Ghana replaced a flat 5% rate with a sliding scale in March 2026 that reaches 12% above $4,500/oz; Burkina Faso set 10% for $4,000 to $4,500/oz; Mali set 9.5% at $4,100/oz. At IAMGOLD’s Essakane mine, royalty costs surged 220% year over year to 35% of cash costs.

A price rally raises the royalty and the cost curve chases the metal up. That is embedded convexity no derivative removes, and it mechanically lowers the effective delta of a producer to the gold price relative to naive operating leverage. It also works in reverse: a revenue-linked royalty is a natural floor under the downside, which is why the cost line does not fall as fast as the price does.

A put on revenue floors the price and leaves the cost line running. A call on fuel caps the cost line and leaves the revenue running. The two exposures are not the same exposure, and the states that hurt are different states.

3. The three legs

Sizing rule first: hedge each leg in proportion to its contribution to margin variance, not in proportion to its notional. Gold dominates the variance and its options are the cheapest in the complex. Oil is a modest variance contributor with expensive options. Currency is a minor variance contributor with the most expensive options in the complex, and is best done with forwards. Hedge rate is set by variance, not by intuition about which risk “feels” largest.

The gold leg is the fixed-exit structure with the direction reversed. The put spread buys the floor; the call that would fund it is the producer’s own pre-approved sale level, and it should only be sold against production the board has already decided to sell at that price. Selling calls to fund the floor on production you would have kept is a forward sale with extra steps and worse disclosure.

Note what the oil leg is not doing. It is not a directional oil view. It caps the fuel bill at $130/bbl — above the ~$118 Brent printed in March 2026 and below the $147.50 cycle high — and it is the only leg in the package that pays when nothing else does.

Note what the FX leg is not doing either. At roughly 3.5% of notional for a twelve-month at-the-money option, an options-based FX hedge is the most expensive thing in the package for the least variance reduction. A rolling forward book locks the rate at close to zero premium and is the right answer for 70% of the cost book. The other 30% takes the rate it gets, not the option premium, unless the client is specifically buying a floor on a currency.

4. Why the legs are cheaper together

The two option legs pay in different states, which is the reason the package costs less than the sum of its parts and covers a band neither leg covers alone.

The gold put pays when demand collapses and the oil call does not. The oil call pays when supply is disrupted and the gold put does not. Their payoffs are negatively correlated, so the joint premium sits below the standalone sum.

This is where the gold/oil ratio earns its place. Spot gold of $4,166 against Brent at $109.15 puts the ratio at 38. It has a job: it tells you which leg is expensive relative to the other, and it switches the sizing emphasis.

The 2026 path illustrates the switch working. Through Q1 the ratio sat well above 40 and average margins were at a record $3,076/oz, so the right hedge was revenue — producers wanted floors, not cost caps. By October the ratio was 38 with Brent up 79% year over year, and the marginal dollar of hedging belonged on the fuel bill.

5. The forward sale is not a hedge

A producer that sold 180,000 ounces forward at $4,500 on 20 January 2026 was right. Gold closed the following nine months at $4,166, below that level. The hedge also surrendered $1,095/oz against the $5,595 record printed the following week — $197 million on the covered volume. That is the price of the hedge, and it is not a mistake.

The mistake is what the hedge was against. A forward prices at spot minus the gold lease rate — 1% to 2% annualized through 2026, with one-month rates jumping from 2.6% to 3.3% in a single session in April. A forward struck at a $4,500 spot really locked $4,455. And a forward touches nothing else: it does not cap the fuel bill, it does not lock the AUD cost book, and it leaves the producer fully exposed to the 16% year-over-year AISC increase that arrived in the same quarter.

A forward sale is a zero-cost price bet. A margin hedge is a small, known, budgetable premium. The forward looks free exactly when the premium looks expensive, which is the wrong moment to discover that the cost line has moved.

6. The equity leg and the offsets

A producer’s revenue risk and its shareholders’ equity risk are different trades with different instruments, and the service should keep them separate.

The equity hedge runs on the listed index. GDX carries $26.25 billion of net assets, roughly 21.8 million shares of daily volume, and a usable options chain — 11,171 open interest at the 88.5 strike with implied volatility in the 43–45% range, against GLD’s 3.5 million contracts of call open interest. A junior or a thin mid-cap can therefore hedge its systematic equity beta directly on an index with listed options, and the operating risk that survives is grade, stripping, permitting and jurisdiction.

The leverage is why the equity leg is not a substitute for the revenue leg. Gold miners have run a 1.65 beta to gold over twenty years, a 46.5% five-year maximum drawdown against GLD’s 26.4%, and a 12.4% decline over the three months to 20 July against GLD’s 6.7%. A miner and a miner shareholder are levered to the same metal at different points in the chain, and the hedge for the revenue line does not reach the equity line at all.

Two offsets do real work and neither is a substitute for the three legs.

Byproduct. Silver sits at a 70:1 gold-silver ratio with roughly 72% of supply arriving as a byproduct of other mining. For a polymetallic producer that is a partial natural offset to gold revenue. It is not a hedge, because it carries the same macro exposure: it offsets gold revenue with more gold, and the two fall together in the demand-destruction state. Copper does the same for the polymetallic subset.

Royalty and streaming. Selling a royalty is a revenue hedge sold for cash upfront — no premium, no counterparty, no roll. It is the same trade run in reverse by Franco-Nevada or Wheaton, who price it with no operating cost and no cost inflation attached. A producer that wants to de-risk without paying a premium or giving up the melt-up should quote this alongside the derivative package, not after it.

7. Worked example

A representative mid-cap Australian producer. The numbers are illustrative; the market inputs are not.

Legs, priced off COMEX December gold options at 23.67% implied volatility: 180,000 ounces of gold (60% of production) at a $3,666 floor, cost $25.41/oz or $4.58M a year. A Brent $130/$160 call spread on $11.1M of six-month notional, $4.90/bbl or $1.0M a year. A rolling forward on $175M of AUD costs, about $0.44M a year. Package cost $6.0M, or 0.48% of revenue. It buys protection worth $89M of revenue at risk on the gold leg alone — a 15-to-1 ratio, and 19.5-to-1 before the fuel and currency legs.

Four twelve-month scenarios, with the FX forward settling inside the AISC line rather than as a separate payoff:

Royalties are modelled at 5.4% of revenue, so they fall with the price and cushion the downside. That is the embedded short call from section 2 working in the producer’s favour on the way down.

The trade is an asymmetry, not a flat reduction in variance. Unhedged, the demand-destruction case takes margin to $456M — 34% below base. Hedged, it takes margin to $516M, 25% below base, because the put spread pays $65.9M on 180,000 ounces at $3,300 gold. The supply-shock case gives up $4.5M of hedge cost on a $135M gain. $6.0M a year buys $60M in the state that hurts and costs $5M in the state that pays.

That is the actual trade, and it should be stated plainly. A 12%/22% put spread on 60% of production is a partial hedge. In the 2026 replay — gold at its $3,283 low with Brent at $135 — the put spread pays $383/oz on 180,000 ounces, or $69M, against a revenue loss of $263M. It takes a 38% margin loss to a 29% margin loss. It does not replace the revenue, and it is not meant to. It converts an unbounded loss into a bounded one at a known annual cost, which is what lets a board hold through a drawdown without a forced action.

Coverage is the dial that closes the gap. At 100% of production rather than 60%, the same spread costs $7.62M and pays $115M in the replay. At 40%, it costs $3.05M and pays $46M. The right setting depends on the debt schedule and the cost of capital, not on the view.

8. Sizing, triggers, and where it fails

The boundaries are not loopholes in the structure; they are the structure.

The structure does not make the gold price predictable and it does not make the cost curve fall. It converts an unbounded revenue exposure into a bounded margin at a known annual cost under half a percent of revenue, and it does that without surrendering the upside the metal printed in January. What it removes is the board’s exposure to a price drawdown arriving in the same quarter as a cost spike — which is not a scenario. It is what 2026 actually did.