Can You Trade Gold on Weekends? Yes, but Prices Disagree

Gold now trades through the whole weekend in five kinds of venue, and none of them is the market that sets its price. That market is the main gold futures contract on COMEX. It stops at 5pm New York time on Friday and reopens at 6pm on Sunday. London shuts for the weekend too. So can you trade gold on weekends? Yes, through a broker's weekend market, a gold token on a crypto exchange or a crypto perpetual. Since 24 July 2026 you can also trade CME's 1-Ounce Gold futures straight through the weekend, and since 10 September, Kalshi's gold perpetual. What none of them can do before Sunday evening is check its number against the benchmark.

That gap stopped being a technicality on Saturday 28 February. The United States and Israel struck Iran while the benchmark was dark and London was shut, and gold kept moving anyway. Tokenized gold held near $5,400 onchain all weekend, while the official price sat at Friday's close of $5,296. "A $104 premium the market couldn't see until Monday," Tom Wan of Entropy Advisors called it. If you held gold that Saturday, you had a decision. Act on the weekend number, or wait for the market that closed on Friday?

Magnus would tell you to wait. Thirty years on trading floors in three countries taught him to trust the places where metal actually changes hands. To him, a Saturday quote is somebody's estimate of Monday, however confidently the screen shows it. Halvor would tell you to act. He is the systems engineer who bought Bitcoin the year his savings paid a tenth of a percent. For him, a price is whatever people will trade at, on whatever day they happen to be awake. The two have argued gold against Bitcoin before. This time the fight is smaller and stranger: a price that exists only while the market it refers to is asleep.

Ava is at the table to settle it, the way she reads any market: as pressure and geometry, with more trust in a sample size than in a speech. Once both cases are on record, she will rule one question at a time. Lucia came to watch, with her mind made up. She reads the crowd's emotional weather from the tape, and a Sunday spike has a smell she believes she knows. "A weekend move is fear with nobody on the other side of it. It fades when the real market opens," she says.

Where does the price come from when you trade gold on weekends?

Magnus agrees with Lucia, but only halfway. A weekend move can fade, he allows. His objection is not the fade but the word "price" itself. "Take any Saturday quote and ask who produced it," he says. The broker's weekend market is the easy case, because the number is the broker's own. "The weekend instruments prices are determined by IG based on various factors," IG says on its product page. Positions there "will be rolled into a weekday contract at the close of trading," it adds. A gold token is at least a market of its own, though a thin one, and each exchange that lists it keeps its own order book. What it fetches on a Saturday tells you little about what a gold token actually entitles you to. Crypto perpetuals sit one step further from the metal, since each one prices off its own venue's index.

Then he reaches the one that ought to worry Halvor. CME's 1-Ounce Gold futures really do trade all weekend. Under CME's own launch terms, though, the contract is "financially-settled based on the daily settlement price of the global benchmark Gold futures contract," so it pays out in cash. The weekend book is real, and every position in it still resolves against the market that was closed while it traded. Kalshi's gold perpetual, the newest of the five, pays daily funding against a price from Pyth Network. That price, in Kalshi's description, draws "on quotes from market makers, venues, and institutions." On a Saturday, that means whichever of them are still publishing.

"CME opened a weekend market, but it did not open the benchmark," Magnus says.

Halvor lets him finish, then goes after the word "benchmark" itself. Every price is somebody's, he says. The COMEX and London prices are simply the ones open during the hours their institutions keep. What changed this summer is that the exchange running the benchmark decided a weekend market was worth building. Between the 24 July launch and 11 August, CME reported, more than 53,000 1-Ounce Gold contracts traded in the new weekend sessions. That is three weekends and about $219 million in notional value. "The largest liquidity pool for weekend trading in Gold futures," CME called it. "The news also doesn't stop on the weekends, but markets do," Kalshi's launch note said in one line. Halvor's version is blunter. "The odd market is not the one that stays open but the one that shuts two days a week," he says.

Side by side, the five weekend prices raise one question: on a Saturday, what does each of them answer to, and what happens to it when the benchmark reopens at 6pm on Sunday?

Five venues give five different answers. Only one of them, CME's small contract, is tied by its own terms to the benchmark's settlement. That is Magnus's point at full strength. It is also, awkwardly for him, a point for Halvor. The exchange that owns the benchmark now runs a weekend book beside it, and it did so because enough people wanted to trade gold while the benchmark slept.

A price is a kind of permission: the right to say what something is worth and have the number stick. On weekdays that permission sits with two institutions. On weekends it is rented to whoever stays open, and each tenant writes its own lease.

So the concessions come quickly: Magnus grants that a weekend gold price exists and moves, and Halvor grants that it is not one price. Its deepest regulated version even settles against the benchmark he keeps treating as one venue among many. Lucia, though, is stuck on the 53,000. A spike with nobody on the other side does not print fifty-three thousand contracts. Every one of those trades had a buyer and a seller, and both chose to be there on a weekend.

"Somebody was there, and I never counted them," she says, mostly to herself.

Did the Sunday price know something on 28 February?

Halvor takes the wobble as an invitation. He goes straight back to 28 February, when the open venues had the news and the benchmark did not, so that when it reopened on Sunday evening it had catching up to do. CryptoSlate's analysis of the reopen put a number on it. Hyperliquid's weekend gold price "sat closer to the first benchmark print than Binance's tape by approximately 22 to 31 basis points," it found. To Halvor, that is the whole case: the market that was awake got closer to the answer.

"Everyone knows the important news waits until Monday," he says, and lets the date do the rest.

Magnus disputes not the war but the definite article. Which weekend price got closer? "A median premium of roughly 75 to 78 basis points above Binance's equivalent contracts" is how the same analysis priced Hyperliquid's gold and silver perpetuals. That is two crypto venues, the same metals and the same headlines, sitting three-quarters of a percent apart. "If gold has two prices on a Saturday, then on Saturday it has an argument," he says. "One weekend isn't a law, and broader sampling complicates the narrative," the analysis adds. The broader sample is a Blockworks study of equity perpetuals. There, weekend prices came closer to Monday's reopen "only about 50.7% of the time," with a median improvement of roughly 0.4 basis points.

Ava writes both numbers down and taps the gap between them. "A war weekend and a sample of ordinary ones," she says. "They answer different questions, and neither of you gets to borrow the other's." Then she lets them keep going.

Lucia is quieter now, because the war weekend is the case she came to dismiss. She had expected the February premium to evaporate at the open. In her experience, a Sunday premium loses its nerve once the deep market comes back. This one held, and the benchmark reopened toward the weekend price. A fear trade that the benchmark then confirms is not one she knows how to wave away.

Why does Kalshi keep gold's funding running on weekends but pause silver's?

Lucia's next question comes out of the same weekend. Silver out-traded gold on Hyperliquid, more than $227 million in a day against about $173 million. So why does Kalshi's silver perpetual stop its funding on weekends, while its gold perpetual never does? Nobody at the table can answer, because Kalshi has not published a reason. What it has published is the schedule: funding on its precious-metals perpetuals settles once a day. It never pauses for gold; silver's pauses on weekends and holidays.

CME never has to make that choice, because its weekend gold is a dated future with no funding at all; the contract carries its price to expiry and settles against the benchmark. Kalshi's gold perpetual never expires. That is why it needs a funding payment to hold its price near the thing it tracks, and the payment runs every day of the week. Its silver twin switches the machinery off on weekends.

Magnus finds gold's design the harder one to defend. Funding exists to hold a perpetual to its reference. On a Saturday, gold's reference is Pyth's blend of whoever is still quoting, with weights no one outside can see. Kalshi self-certified the contract with the CFTC and calls it "fully regulated by the CFTC." Magnus takes that at its word. The regulation covers the contract and the exchange that lists it, while the publishers pricing gold on a Sunday sit outside both. "Funding paid every morning, weekends included, against a price the benchmark can't check until Sunday night," he says. "It is the most modern thing I have heard all year." He means it as a diagnosis.

Halvor thinks silver is the design with the problem. Pausing the funding does not remove the weekend, but hands it over on Monday as a single gap. That gap is the lump a perpetual exists to smooth. He has seen the same choice on a stock-index perpetual that only charges funding during the cash session. There, moves outside the session never enter the funding and arrive instead as a gap at the open. "The price moved on Saturday whether or not anyone was invoiced for it," he says.

Then he does the arithmetic that cuts against him. The February premium was $104 on a $5,296 close, just under 2%. Kalshi's gold perpetual launched with a reported maximum leverage of 15.2 times, a ceiling the venue can change. At that ceiling, a 2% weekend move is about 30% of the margin posted. It lands before the benchmark has printed a single trade, and Halvor does not dress it up.

Leverage turns a weekend price from information into a margin call.

A weekend gold price is only as good as the news behind it

Ava waits until both cases are complete, then starts with whether a weekend gold market exists at all, because Lucia's opening claim depended on it not existing. It does, and the point goes to Halvor. More than 53,000 contracts, about $219 million in notional value, changed hands in CME's first three weekends. Each had a willing buyer and seller. Whatever a Sunday gold price is, it is not an empty room.

Existing is not the same as knowing something. Her second question is whether the price carries information when news breaks, and on 28 February it did. That one is Halvor's too. One weekend venue landed 22 to 31 basis points closer to the benchmark's first print than the other. The premium the official price "couldn't see until Monday" was there on Saturday for anyone looking.

The trouble starts when those two wins get stretched into a third: is there one weekend gold price? No, and on this Magnus is simply right. Two crypto venues sat 75 to 78 basis points apart over the same weekend. At least five kinds of venue set weekend prices by their own rules, and Pyth's weekend publisher weights are not public. Even the contract CME calls the largest weekend pool in gold futures settles against the benchmark.

The ordinary weekend goes to Magnus too, with a scope attached. In the broad sample the analysis cites, weekend prices came closer to Monday's reopen only about half the time. The median edge was around 0.4 basis points. That is close to a coin toss. The sample was equity perpetuals rather than gold, and a gold version needs weekends the regulated books have only just started recording.

That leaves the question the builders answered differently. Should a regulated contract pay funding against a weekend price at all? Ava calls it unsettled, and she means it. CME sells a dated future with no funding. Kalshi charges funding every day on gold and pauses it on silver, with no published reason for the split. When the two venues that built regulated weekend gold markets made opposite choices, a verdict here would be not found but invented.

Put together, the ruling never splits the difference. Halvor wins the existence of the weekend market and its worth on a news weekend, Magnus wins the single price and the quiet weekend, and the funding question stays open until someone collects the data.

A Sunday gold price is information in proportion to the news behind it: early on a loud weekend, close to noise on a quiet one.

Lucia's update is smaller than a conversion. It is also larger than she planned. "I came in saying a weekend move is fear with nobody on the other side," she says. "Fifty-three thousand contracts tell me somebody was on the other side. Twenty-two to thirty-one basis points tell me that on 28 February the fear was not behind the market but ahead of it. I still read Sunday through the fear." What changes, she says, is the first thing she checks: whether anything actually happened, or whether it is a quiet weekend dressed up as a loud one.

What would it take to trust a Sunday price?

Magnus answers first, and one war will not move him. A run of ordinary gold weekends in which the Sunday price beats Friday's close at predicting the reopen would. The regulated weekend books at CME and Kalshi now make that sample possible to collect, one Sunday at a time. Halvor's condition is the mirror image. Show him a thin Sunday print that liquidates leveraged positions and then reverses when the benchmark reopens. That would convince him a weekend price can hurt people in exactly the way Magnus fears.

Until one of those arrives, the weekend belongs to whoever stays open, and Monday belongs to whoever sets the benchmark. The venue that stays awake hears the news first. The one that sleeps still gets the last word, because everything else is eventually measured against it.

Next time gold jumps on a Sunday, ask whose price it is, and what it will answer to at 6pm.

Gold in the Kodex simulator trades as XAUT, a gold token, so on a weekend you are looking straight at one of those five Sunday sources. Hold a simulated position through a quiet weekend and a loud one, write down each Sunday-night price, and see which one Monday agrees with. It takes a free account, and a wrong Sunday guess costs nothing but a line in your notebook.

Start free →