The Lighter Ring

The idea

This note began with an article: David Uberti's "The Popular Jeweler That Got Caught in a Whirlwind When Gold Went Haywire", in The Wall Street Journal of October 4, 2026. It follows Mejuri, a Toronto jeweler whose pieces often sell for less than $300, through a market in which gold passed $3,000 an ounce, then $4,000, then $5,000. Uberti reports that the company leaned on 10-karat gold, on plating, on silver and even on stainless steel; that it raised prices, slowed the opening of new stores, and changed where its orders are filled and where its gold comes from; and that its plans now allow for gold at $8,000. He adds a figure that explains his headline. In the half century before 2025, gold futures moved by a hundred dollars or more in a single day three times. This year they have done it 42 times.

The article says nothing about hedging, and nothing in this note is a statement about Mejuri's finances, which are private. But it left me with a question the article had no reason to ask. A jeweler that buys its gold forward buys ounces. When it then changes what is in the ring, what becomes of the ounces?

So I drafted a model of it, with the help of an AI assistant: seven small models, in fact, each with invented numbers and exact arithmetic, on the gold in a piece, the hedge against it, the cash left after restocking, and the price a jeweler should charge. On October 5 I sent the draft, "Gold prices jewelry design and cash flow", to Vista's Desk and asked what was wrong with it. Over four rounds the Desk proposed joining the purchase of the gold and the settlement of the hedge in a single line of cash, which the draft adopted and which is now its center. It drew out something the numbers implied and the draft had left in algebra: that a redesign can change the sign of a jeweler's exposure to gold.

Most of this is known, and the people who said it first deserve the credit. That jewelers are turning to lighter and lower-karat pieces has been reported all year, by the World Gold Council and the trade press. That inventory gains flatter a jeweler's margin while restocking drains its cash was said plainly by Crisil about India's jewelers in April 2025 and by the Financial Times this April, and Britain's National Association of Jewellers now tells its members to price at replacement cost. A treasury magazine noted in 2024 that a buyer who changes its specifications needs fewer derivatives, and Ford in 2001 and Europe's airlines in 2020 showed what happens to a hedge when the need for the commodity collapses. Economists worked out long ago how a hedge should shrink when quantity moves against price, for farmers first and later for buyers. What we have not found anywhere is the link between them for jewelers: that redesign makes the gold behind each sale fall faster than the sales, so that a hedge bought for the old plan grows too large and turns the company's exposure to gold around; and that the right hedge then turns on who does the cutting, designers or customers, set against the industry's own response in 2025 and 2026.

Hiero's crown

In the third century before Christ, Hiero, king of Syracuse, vowed a golden crown to the gods and did what a careful buyer of gold still does. He fixed the price of the work, and he weighed out the gold in advance. Vitruvius, the Roman architect who preserved the story, writes that the king "contracted for its making at a fixed price, and weighed out a precise amount of gold to the contractor." The crown came back exquisite, and on the scale it weighed exactly what the gold had weighed. Then came the accusation: the smith had kept back some of the gold and made up the weight in silver. The king could not see how to prove it. He put the problem to Archimedes, who carried it with him to the bath, watched the water spill as he lowered himself into the tub, and ran home naked, shouting that he had found it. A crown alloyed with silver is bulkier than its weight in gold. It spills more water.

What Archimedes found, somewhere between the bath and his front door, was the difference between the weight of a piece and the weight of the gold in it. The jewelry trade has lived on that difference ever since, and today nobody hides it. The fraction is stamped inside the band. A karat is a twenty-fourth: 14-karat gold is fourteen parts gold in twenty-four, a little over 58%, and 10-karat is ten parts, a little under 42%. Hiero's goldsmith thinned the crown in secret and was caught by a bathtub. A jeweler today thins the ring in public, with a hallmark, because the gold has become too dear for the piece.

Here the old story turns over. In Syracuse the man deceived was the one who had weighed out the gold in advance. In 2026 nobody is deceived, and yet the party most likely to be caught out by a thinner ring is still the one who weighed out the gold in advance: the jeweler's own treasury, which bought gold forward for last year's ring.

Ounces, not pieces

I once spent a year in a warehouse. As I wrote in my book, The Scholars' Treasure (chapter 3, page 36), I spent it at Sony Music's warehouse in Carrollton, Georgia, where I "learned and eventually mastered the Catalyst WMS (warehouse management system)" and trained the people on the floor to use it. The system's whole job was to know where every pallet of CDs stood. Nobody ever had to ask how much plastic was in a disc. A jeweler's stockroom has to ask that question about every piece, because what its treasury buys is the metal inside.

The arithmetic is short. The fine gold a line of jewelry needs is

Q = N × w × k / 24

where N is the number of pieces, w the weight of gold alloy in each, and k the karat, so that k/24 is the share that is pure gold. Moving a piece from 14 karat to 10 at the same weight cuts its gold by two-sevenths, a little under 29%. Put the other way round, gold must rise 40% before the 10-karat piece holds as much gold, by value, as the 14-karat piece did. Mejuri's materials page gives the standards it works to: 41.7% pure gold for 10 karat and 58.5% for 14.

A hedge is bought in ounces against a forecast of Q, and its coverage is the hedge divided by what the business will in fact buy. My draft took a jeweler that had hedged 70% of its planned gold and then redesigned around the hedge:

From my draft's model (Appendix A1), rerun in exact fractions. Fine gold against a plan of 100; each piece keeps its alloy weight except in the two rows where it is a tenth lighter. Shaded rows are those where the hedge covers more than the jeweler now needs.

Three things stand out. At the same alloy weight, conversion alone never over-hedges a 70% hedge: move every piece to 10 karat, sell every piece, and the hedge still covers only 98% of the new need. It takes the redesign and the lost sales together, or a lighter piece: a 10-karat alloy is less dense than a 14-karat one, so the same design cast in 10 karat also weighs less, and that alone can tip a 70% hedge over. When customers buy four pieces in five, converting anything over about 44% of the line tips the jeweler over; if each piece also loses a tenth of its weight, under 10% of the line will do it. The third thing is the one the Desk drew out of my numbers when it reviewed them.

Before the redesign, the jeweler in my example was short 30 units of gold. It would have to buy them at whatever the market charged, so a rising price hurt. After the redesign it is long nearly 13 units, gold bought forward that the new line will never use, so a falling price hurts. Nobody traded anything. A design meeting changed which move in gold the company should fear.

Write the jeweler's cash after it restocks as one line and the point becomes exact:

C = p × q − L × q − K × Q + (H − Q) × (G − K)

Here p and q are the price and the number of pieces sold and restocked, L the cost of everything in a piece except its gold, Q the fine gold the restocked pieces need, G the price of gold when the hedge settles, and H the ounces bought forward at the price K. The first three terms are what the business would keep if its hedge matched its need exactly. The last is the leftover position, and the business's exposure to gold takes the sign of H minus Q.

From my draft's combined model (Appendix A7): 80 pieces sold at $220 and restocked, $60 of other cost a piece, a hedge of 70 units at $40 against a need of 57.1 after the switch to 10 karat. Invented figures throughout; a unit of gold here is a unit of account.

When gold falls from $40 to $20 the hedge loses $1,400, and it is tempting to call all of it the price of the mistake. Most of it is nothing of the kind. The 57 units the new line needs are cheaper by just as much, which offsets the loss on the part of the hedge that matches them. Only the loss on the 13 surplus units, about $257, comes from the mismatch, and against cash of about $10,500 it is small. Margin calls can make it feel larger. If the hedge is settled in cash as prices move, a fall to $20 before the jeweler buys or sells anything demands the whole $1,400 at once, returned only if gold comes back. That is a question of bridge financing, and it belongs to the whole hedge, of which the surplus is one part.

The parts of this story have all happened before, at larger scale. Ford took a precious-metals charge of $953 million before tax for 2001, about $1 billion as first announced in January 2002, after its engineers validated a catalytic converter that cut its planned palladium use by more than half; it had bought palladium forward to make sure of supply, and wrote the metal and the contracts down to market. When aircraft were grounded in 2020, IAG, the owner of British Airways, booked losses of €1.78 billion on fuel hedges that had no fuel left to cover, and Lufthansa, Air France-KLM and Cathay Pacific reported over-hedges of their own. One jeweler has had to unwind its gold hedges too, as we come to below. What is new in 2026 is the cause. The need for metal is falling by design, in a year when gold futures have moved by a hundred dollars or more in a single day on 42 occasions, Uberti reports, against three in the half century before 2025.

The hedge that answers the price

My draft held the jeweler's designs fixed while gold moved, and then changed them once, as if at a single meeting. The trade does not wait for meetings. The World Gold Council reports that lighter pieces gained share across many markets in the second quarter of 2026 and that lower-karat designs spread in parts of Asia; India's standards bureau added 9-karat gold to its hallmarking standard in July 2025, and India's Titan and Kalyan added 14- and 9-karat lines in their 2025-26 years; Signet's chief executive told investors in June that, in brands like Banter, its designs would keep taking "advantage of lower weight." When gold rises, the gold in the collection falls, as a rule. And a rule can be priced into the hedge before anyone calls the meeting.

Suppose the gold a jeweler buys shrinks by e percent for each 1% rise in the price of gold. Its gold bill, the price times the quantity, then rises by only about one minus e percent. A forward purchase of H ounces gains H dollars for each dollar the price rises. For moderate moves the bill and the hedge rise and fall together when

H* = Q × (1 − e)

where Q is the gold the jeweler expects to buy at today's price and e is how strongly that amount answers the price. With e at zero the familiar rule holds: hedge what you plan to buy. With e at one the bill does not move with the price at all, because every rise is met by an equal cut in grams, and any hedge is a position in gold.

The bill is not the whole business, and here the jeweler's case parts company with the textbook. Split e in two: the part that comes from designers taking grams out of each piece, e_g, and the part that comes from customers buying fewer pieces, e_q. A gram the designers remove saves its gold and costs nothing else, so the hedge should shrink with it. A piece a customer does not buy saves its gold too, but it takes its margin with it, and that lost margin grows with the price of gold. The hedge that keeps the jeweler's cash steady is then

H* = Q × (1 − e_g + μ × e_q)

where μ is what a piece earns above all its costs, as a multiple of the gold in it. If designers do all the cutting, the hedge is the forecast times one minus e. If customers do all of it, the hedge should be larger than the forecast, by the margin the lost pieces would have earned. Real jewelers sit between. My draft's pieces sell for $200 with $40 of gold and $60 of everything else, a margin two and a half times their gold; if designers did two-thirds of a 0.6 response and customers a third, the hedge that steadies its cash would be 110% of the plan.

The idea is older than any jeweler's treasury. Ronald McKinnon showed in 1967 that a farmer whose harvest tends to be small when prices are high should sell forward less than the crop he expects, and Jacques Rolfo worked the problem for a cocoa producer in 1980 and found the best hedge short of the whole expected crop. A jeweler is Rolfo's farmer turned around: its purchases shrink when the price rises, so part of its hedge already sits in the design studio. The buyer's version has been worked as well, by Yumi Oum, Shmuel Oren and Shijie Deng in 2006 for a power retailer whose purchases rise and fall with the price, and by Ankur Goel and Fehmi Tanrisever in 2017 for a buyer whose own sales move with its input's price. Gerald Gay, Jouahn Nam and Marian Turac showed in 2003 that as quantity risk grows, the best hedge holds fewer forward contracts, for fear of over-hedging, and, depending on how price and quantity move together, more options.

How large is e for jewelers? The industry gives a first reading, though a rough one. In the second quarter of 2026 gold averaged 37% more than a year earlier, and the world's jewelry buyers took 17% less of it by weight; across the whole of 2025 the price rose 44% and the weight fell 18%. Read as a response to the price alone, which is too simple, both give an e of about 0.6. Jewelry makers cut their fabrication by less, 12%, for an e of about 0.4, the difference being gold added to stocks along the trade. The academic estimates are of the same size: Batchelor and Gulley found price elasticities of 0.5 to 1.0 for jewelry demand in six rich countries.

The caution matters. These are a year's changes, read as if price were the only thing that moved, while buyers in China retreated and investors bought bars and coins in their place. And the industry's figures say how large the response was, and nothing about who made it. If it was mostly designers, a 70% hedge, the floor Pandora's treasury policy sets for silver and gold, is larger than the roughly 40% that offsets the gold bill, so on balance it gains when gold rises and loses when gold falls. If it was mostly customers buying fewer pieces that earn good margins, the same hedge is too small. A couple choosing wedding bands bends less than someone buying a fourth stacking ring. The split is the number to find, and it is what the people who drew the rings and set the hedges would know.

Two weights in the bag

The law of Moses tells a merchant, "You shall not have two differing weights in your bag, one heavy and one light." A careful jeweler keeps two, quite honestly. One is what the goods in the case cost when they were bought. The other is what it will cost to put the same goods back. In a year when gold doubles they are far apart, and a jeweler can look richer by the first while it grows poorer by the second.

My draft's second model shows how. A jeweler holds 100 pieces bought at $100 each, $40 of it gold. In a steady year it sells them at $200, buys them back at $100 and keeps $10,000. Then gold doubles. It sells 85 pieces at $220 and buys 85 back at the new cost of $140:

From my draft's model (Appendix A2), rerun in exact fractions. One season; the opening stock sells first; no credit, tax or overhead; the $60 of cost that is not gold is unchanged.

Gross profit rises 2% and the margin from 50% to about 55%, while the cash left after restocking falls 32%. Both are true. The gap of $3,400 is money that went back into the case, which now holds 85 pieces at $140 where it held 85 at $100. Economists named the difference long ago. Edgar Edwards and Philip Bell called it a holding gain in 1961 and kept it apart from current operating profit; BP still reports a replacement cost profit so that its gains on oil held while prices rose do not flatter its results; the national accounts strip it from corporate profits through the inventory valuation adjustment. Warren Buffett, writing to his shareholders in the inflation of 1981, gave inflation itself the name: "a gigantic corporate tapeworm," which takes its dollars for receivables, inventory and fixed assets whatever the business reports as profit. The tapeworm eats first.

Chow Tai Fook lived this in its year to March 2026. Its profit rose by half, to HK$9.0 billion. Its own bridge of the gross margin credits rising gold with 10.3 points of its 32.3% margin; the gold loans it uses as a hedge lost the equivalent of 6.6% of revenue, booked below gross profit. Meanwhile its inventory grew 15%, to HK$63.7 billion, mainly because each gram cost more, and the cash its operations generated fell from HK$10.3 billion to HK$1.3 billion, as it paid more for each gram it restocked and replaced gold it had handed back to its lenders; its own pro forma free cash flow, which counts gold loans as financing, fell from HK$20.8 billion to HK$7.8 billion. Its business differs from Mejuri's: much of what it sells is priced by weight at the day's gold price, and its hedge is a short one, gold borrowed against the stock it holds. But the two weights are the same. The one it reported was heavy. The one in its bank was light. Crisil, the Indian rating agency, said as much of India's jewelers in April 2025: margins up on inventory gains, borrowing up to pay for the restocking.

The two weights also give different advice about price. Take a jeweler that can hold its price at $200 and sell 100 pieces, or raise it to $220 and sell 80. Judged by what the pieces cost when they were bought, $100, holding the price earns $400 more. Judged by what they cost to replace, $140, raising it earns $400 more. The two rankings disagree whenever the higher price would sell between 75 and 83 pieces, and in that range customers have a vote, which they cast loudly. Uberti reports that one TikTok user, with a quarter of a million followers, called a price rise on a Mejuri 14-karat necklace "DIABOLICAL."

A redesign has its own threshold. My draft put the cost of a 10-karat version of a $220 piece at $117, against $140 in 14 karat, and charged $1,000 for the design work. It pays only if customers keep buying at least 89% of what they bought before: at 90% it earns a little more than the old piece, at 85% a little less. And a hedge, my draft's fifth model shows, changes none of these choices by itself. Its payoff is the same whatever the jeweler charges, so the best price is the same with it or without it, a result economists proved for competitive firms with a futures market in 1979 and 1980. The exception is cash. If the hedge pays out before the jeweler must pay for its next order, and nobody else would lend it the money, the hedge can let it buy more stock and charge less. Kenneth Froot, David Scharfstein and Jeremy Stein built a theory of why firms hedge at all on that exception.

Three jewelers, three hedges

No filing will say how many grams a ring lost this year. The listed jewelers do say how they hedge, and each shows a different face of the mechanism.

Pandora: a hedge sized to a plan that is changing. Pandora's metal is mostly silver, about 30% of its cost of goods in 2025, with gold another 8%. Its treasury hedges at least 70% of a rolling twelve months of planned purchases, updating the hedges monthly or whenever the production plan changes, and because metal takes five to ten months to pass from purchase into cost of sales, its 2026 accounts were in effect fully hedged. Silver in its cost of sales runs at about $32 an ounce through 2026, then climbs to about $44 in the first quarter of 2027 and $69 in the second; gold climbs from $2,913 to about $4,550 over the same stretch. In February the company announced its way out: platinum-plated pieces on a metal-alloy core, with at least half of the relevant silver range to move in 2027. In May it set about 80% of its current silver range, by revenue, for the end of 2028, and wrote that the shift "will, of course, structurally impact silver production plans and the hedging requirement." It had already acted on that: it bought no new hedges in the second half of 2025, partly because it expected to use less silver, and in early 2026 it let its cover fall below its own 70% floor while it waited for new production plans. In August it said around half of the range would move in 2027, after a pilot that began in July, later than first planned, and reported its 2027 silver costs 90% to 100% hedged at about $65 an ounce, an estimate that, its own footnote says, rests partly on projected product mix. Its silver and gold hedges, worth DKK 1.7 billion at the end of 2025, had become a liability of DKK 0.3 billion by June, as older contracts paid out and metal prices came off their January peaks. The plan is moving. If the conversion runs ahead of it, or sales fall short, part of the contracts, struck at about $69 to $70 an ounce for purchases from July 2026 to March 2027, would cover silver Pandora will not buy; the LBMA silver price was about $61 on October 2. If the conversion keeps slipping, Pandora will need more silver than it hedged, and at the October 2 price that extra silver would cost less than its hedged ounces.

Signet: a small hedge, and a precedent. Signet, the owner of Kay, Zales and Jared, began hedging gold again in the summer of 2025. Its forward purchases rose to about 38,000 ounces in November, then stood at 35,000 at the end of January, 26,000 in May and 29,000 in August: roughly $122 million of gold at the October 2 price, against sales of $6.8 billion a year. It is a small book. The forwards were an asset of $26 million in January and a liability of $11 million by August, as gold came off its peak. Its chief executive told investors in September that units show some resistance when gold costs are passed on, particularly at lower price points and gold weights; in North America its average price rose 7.8% in the second quarter while units fell 8.3%, mostly, he said, because of mix. Signet is also the one jeweler we found that has had to abandon hedge accounting on gold. In the spring of 2020 it de-designated and liquidated all of its gold hedges, 63,000 ounces at the start of February, because the pandemic left it unable to say how much gold it would buy; gold had risen, so the unwinding produced a gain of about $9 million.

Chow Tai Fook: the other side of the trade. Chow Tai Fook hedges the gold it already holds by borrowing gold, so its hedge loses when gold rises and its stock gains. It is the mirror image. Its hedge ratio by weight, a figure it does not define, fell from 69.1% in March 2024 to 55.4% a year later, held there through September, and then dropped to 39.0% by March 2026. The company gives no reason beyond adjusting the ratio to "operational requirements, prevailing market conditions" and its appetite for risk. A lower ratio leaves more of its stock exposed to the price, which paid while gold rose and works the other way when it falls; gold ended September 9% below its level at the March year end.

What the price assumes

Two of the three share prices can be read against the mechanism, one for each half of it.

At HK$10.86 on October 2, near the low end of its range over the past year, Chow Tai Fook was valued at about HK$107 billion, 12 times its profit for the year to March. Take out the net gain from gold, the roughly HK$9.7 billion that rising prices added to gross profit less the HK$6.3 billion its gold loans lost, after tax, and the multiple is about 17. At 17 times what it earned without the gold gain, the price needs those gains to recur or the jewelry business to grow. It does not tell us what the market expects for gold.

At DKK 791.60 the same day, Pandora was worth about DKK 75 billion including its debt, 10 times the operating profit it earned in 2025 at a margin of 23.9%. Its guidance for 2027 is a margin of at least 12%, or 14% before one-off conversion costs, set when it assumed silver at $82 an ounce; in August it said its hedge at about $65 adds about 2 points on its own. Its ambition once the platinum conversion is done is a margin above 21%. On 2025's revenue, margins of 12%, 14% and 21% would put the price at 19, 16 and 11 times operating profit. The price looks through 2027 to the ambition beyond it, which assumes the conversion works. That is not yet settled.

Where the model points in listed markets

None of these is a recommendation; each is a claim the market will test, with what would prove it wrong. The first two are the stronger ones; Pandora's and Signet's are watch items.

The industry keeps cutting grams. If the trade goes on answering the price as it did over the past year, with e near 0.6, the World Gold Council's count of jewelry bought in the third quarter of 2026, when gold averaged 23% more than a year earlier, should come to about 331 tonnes, against 376 tonnes a year before; an e anywhere from 0.4 to 0.8 gives 318 to 345. This measures the size of the response, not who made it. Falsified if the council's first estimate, expected at the end of October, is above 353 tonnes (an e below 0.3, where a 70% hedge stops being a position even if designers did all the cutting) or below 300.

Chow Tai Fook: the gold year turns. Rising gold supplied 7.9 points of revenue to its gross margin in the six months to September 2025 and 10.3 points over the year to March 2026, holding gains of the kind this note describes. With gold lower at the end of September 2026 than at the end of March, that gain should shrink. At about 17 times what it earned without the gold gain, the price needs it to recur or the business to grow. Falsified if, in its results for the six months to September 30, 2026, expected in late November, the gain it reports from the gold price is 7.9% of revenue or more.

Pandora: a watch on the silver hedge. With 2027 silver 90% to 100% hedged at about $65 an ounce and around half of the relevant silver range due to move to platinum plating in 2027, a conversion that runs ahead of the plan, or sales that fall short of it, would leave contracts struck at about $69 to $70 covering silver Pandora will not buy. So far the conversion has run later than first planned, which would leave Pandora short silver, not long. The trace of an over-hedge would be hedge ineffectiveness, or a hedge partly discontinued because the purchases it covered are no longer expected. Falsified for this cycle if Pandora's annual report for 2026 shows no discontinued metal hedges and commodity hedge ineffectiveness under DKK 26 million either way, its 2024 level.

Signet: a watch on the hedge book. Signet's risk factors list product redesign, with assortment changes and pricing, among its answers to costly metal; its chief executive expects lighter designs in brands like Banter; and its forward book has shrunk from 38,000 ounces in November to 29,000 in August. A smaller book would fit lighter pieces without proving them, since Signet hedges only a portion of its gold and only when price and volume warrant. At $100.58 the shares are valued, net of cash, at about 6 times the adjusted operating profit Signet expects this year; its gold hedge is too small to move that number. Falsified if the gold forwards in its report for the quarter ending October 31, 2026 total 35,000 ounces or more.

The data that we're missing is...

• Grams per piece, before and after. The fine gold in matched pieces, same design, size and construction, in 2024, 2025 and 2026, by brand and by price point. No filing gives it, and photographs cannot.
• Hedges against purchases, in the same unit. Signet gives ounces but not the share of its purchases they cover; Pandora gives shares but not ounces. Nobody gives both, or says how often the forecast behind the hedge is redrawn.
• How fast designs answer the price. The months between a rise in gold and a lighter or lower-karat piece reaching the shelf, and how much of a range moves in a season.
• What customers keep buying. The share of a 14-karat piece's sales that a 10-karat version holds, by price point. My draft's break-even is 89%.
• When the cash moves. Whether jewelers' forwards are margined in cash, and how settlement dates line up with purchases.
• Who carries the stock. How much of the rising cost of inventory sits with suppliers on consignment (Signet held $596 million of consigned goods in January), with gold loans and with lenders, and how much with the jeweler.

Primary research

Likely, we can get some of it from primary research, which is what Vista is all about. The people who sized these hedges, drew these rings and supplied this gold know already what the filings will take a year or two to show.

For a client engagement, Vista combines this report with that primary research.

The checkpoint, set down now

We will score this note on March 31, 2027, and publish the result on the scorecard whether it flatters us or not. Each test sides with the reading here:

• The industry keeps cutting grams. The World Gold Council's first estimate of jewelry consumption for the third quarter of 2026 lies between 300 and 353 tonnes. Our reading is about 331.
• The gold year turns. The gain Chow Tai Fook reports from the gold price for the six months to September 30, 2026 is below 7.9% of revenue, its figure for the same half of 2025.
• An over-hedge surfaces. Pandora reports a discontinued metal hedge, or commodity hedge ineffectiveness of DKK 26 million or more either way, in any report published by March 31, 2027.
• Signet's hedge stays lighter. Its gold forwards at the end of the quarter to October 31, 2026 are below 35,000 ounces.

Three or four passes and the reading here gains ground; two is too early to say; one or none and the mechanism stays where my draft first found it, in an appendix. Four yes-or-no tests cannot measure a hedge. They can show whether this one is visible from outside.

Archimedes needed a bath to see what was in the crown. A jeweler's treasury needs only the purchase plan, if someone remembers to send it. The gold in a ring is stamped inside the band. The gold in the hedge sits in a contract on another floor, drawn up in a different meeting and counted in ounces. Hiero weighed out his gold before the work began and was cheated by his smith. A jeweler that weighs out its gold before the redesign begins is cheated by no one. It has simply bought, in advance and in good faith, the gold for a ring it no longer makes.

Questions this note answers

Why are jewelers switching from 14k to 10k gold?

Because gold became too expensive for familiar price points. At the same weight, a 10-karat piece holds two-sevenths less gold than a 14-karat piece, a little under 29%, so a jeweler can hold a price while gold rises by up to 40%. Mejuri, Signet's brands, Titan and Kalyan have all added lower-karat or lighter pieces, and the World Gold Council reports lighter pieces gaining share across many markets in 2026.

How can a lighter or lower-karat design leave a jeweler over-hedged?

A jeweler hedges gold in ounces against a purchase plan. If it then cuts the gold in each piece and sells fewer pieces, it needs less gold than it bought forward. In the note's example, a jeweler that hedged 70% of its gold, moved its whole line to 10 karat and kept four sales in five ends up with a hedge of about 122% of its new need, and a falling gold price becomes the danger.

How much of its gold should a jeweler hedge when demand falls as prices rise?

It depends on who does the cutting. If designers take grams out of each piece as gold rises, the hedge that keeps the business steady is the forecast times one minus that response: about 40% of the plan if grams fall 0.6% for each 1% rise in gold. If customers buy fewer pieces, the hedge should not shrink, because each lost sale takes its margin with it. The industry's figures show the size of the response, about 0.6 over the past year, but not the split.

Why can a jeweler's profit rise while its cash falls?

Because it sells stock bought before gold rose and then pays the higher price to restock. Gross profit includes a holding gain; the cash left after restocking does not. Chow Tai Fook's profit rose by half in its year to March 2026 while its cash from operations fell from HK$10.3 billion to HK$1.3 billion.

How do Pandora, Signet and Chow Tai Fook hedge gold and silver?

Pandora hedges at least 70% of a rolling twelve months of planned silver and gold purchases and had 2027's silver 90% to 100% hedged at about $65 an ounce. Signet buys gold forward, about 29,000 ounces in August 2026. Chow Tai Fook hedges the gold it holds by borrowing gold, and cut that hedge from 55.4% of its gold stock by weight, a ratio it does not define, to 39.0% in the six months to March 2026.