What happens when a gold futures contract hands over to the next one

Why there is no single gold futures price

Gold futures are listed as a ladder of separate contracts, each tied to a delivery month. They are different instruments with different prices, different open interest and different liquidity, even though they all reference the same metal. Only a subset of listed months attracts real volume, and within that subset activity is overwhelmingly concentrated in whichever contract traders have collectively agreed is currently the one to trade.

Two labels get confused here and the difference matters. The front month is simply the nearest contract to expiry. The active month is the one carrying the most volume and open interest. For most of a contract cycle they are the same. Approaching expiry they separate, because volume leaves the nearest contract before it dies while that contract is still technically the front one. Charting the front month blindly means your chart changes instrument at a different moment from when the market actually moved on, and you spend several sessions watching a thinning, unrepresentative book.

The deadline that forces the decision

Nobody rolls because of the expiry date itself. They roll because of a date before it. The relevant deadline is the first day on which a holder of a long position can be issued with a delivery notice, which means being told that metal is coming and payment is due. Any trader who does not want to take delivery of bars has to be out of the expiring contract before that point, and the overwhelming majority of participants fall into that category.

This creates a predictable migration. In the window before that deadline, speculative and hedging positions move from the expiring contract to the next liquid one. Liquidity follows them, not the other way round, so the market tips over from one contract to the next across a handful of sessions rather than at a single instant. There is no bell. Volume in the old month declines, volume in the new month climbs, and at some point during that handover the new contract becomes the real market and the old one becomes a backwater with wide spreads.

How the roll is executed as one trade

A trader rolling a position does not sell the old contract and then buy the new one as two separate decisions, because the minutes in between would leave them exposed to the price moving. Instead they trade a calendar spread: a single order that simultaneously closes the near leg and opens the far leg, quoted as the price difference between the two months rather than as two outright prices.

This matters for anyone watching the tape. During the roll window, a large share of activity is spread trading, where the participant has no directional opinion whatsoever and is indifferent to whether gold rises or falls. They care only about the difference. That is why volume and open interest figures look strange in those sessions, and why apparent surges in participation during a handover are usually nothing more than positions existing briefly in two places at once. Reading conviction into roll week volume is a reliable way to get the market's mood completely wrong.

Three ways a continuous chart gets stitched

Because the old and new contracts trade at different prices, joining them creates a seam. There are three standard approaches and they produce visibly different charts.

• Unadjusted. The series simply switches instrument on the roll date, leaving a vertical step wherever the two prices differed. Every price shown is a price that genuinely traded, and the chart contains gaps that no trader ever experienced.
• Difference adjusted. All historical bars are shifted by the gap at each roll so the seam vanishes. The shape of every move is preserved exactly, but older prices are no longer the prices that traded, and in a long history they can drift a long way from reality.
• Ratio adjusted. History is scaled rather than shifted, which keeps percentage moves honest at the cost of distorting absolute levels even more.

None of these is wrong. Each is right for a different question, and the trap is using one for a question it cannot answer. Horizontal levels drawn on an adjusted chart sit at prices that never existed. Returns measured on an unadjusted chart include fake gaps that no position ever suffered.

Why the futures chart and the spot chart disagree

Traders who move between the two instruments often assume one of the feeds is faulty. Neither is. A futures price contains the cost of carrying metal to its delivery date, so it normally sits above the spot price, and that premium shrinks as expiry approaches. A spot price contains no carry at all because there is nothing to carry it to. The two therefore track each other closely in shape while sitting at different levels, and the gap between them changes over the life of each contract.

On top of that, futures trade in exchange sessions with a formal settlement price, while spot trades continuously through a network of dealers with no official close. So the daily bar boundaries differ, the highs and lows differ slightly, and a level that looks clean on one chart can look like a minor overshoot on the other. If the basic structure of the spot instrument is still fuzzy, how the spot gold pair works is the foundation to lay first.

What the roll does to a backtest

This is where the roll stops being trivia and starts costing money. A strategy tested on an unadjusted continuous series will encounter price steps at every roll. If those steps happen to sit near your entry or stop logic, the test will report trades that could never have occurred, in both directions. A long history of such tests accumulates dozens of these artefacts and the equity curve becomes partly fictional.

A difference adjusted series removes the steps and introduces a subtler problem. Because every historical bar has been shifted, any rule referencing an absolute price level, a round number, a long term high, is now referencing something that never traded. Worse, in a persistent contango the adjustment process systematically alters the slope of the historical series, which quietly flatters or punishes trend following rules depending on the direction of the adjustment.

Any test on futures history needs to state which series was used and why. That belongs on the same list as the other quiet distortions in the backtest honesty checklist, because it is the kind of error that produces a confident number from a broken input.

Roll week behaviour on the tape

A few things reliably change during the handover and they are worth anticipating rather than interpreting.

The expiring contract gets thinner. Spreads widen, the resting book shallows, and stops placed in that contract fill worse than they would have a week earlier. Meanwhile the new contract is still building depth, so for a short period neither venue has the liquidity you are used to. Spread trading inflates volume figures without reflecting any directional appetite. And because many institutional participants have scheduled roll procedures, a predictable, mechanical flow passes through the market that has nothing to do with anybody's view on gold.

The honest caveat is that none of this is tradeable on its own. There is no edge in the roll for a retail trader, and attempting to anticipate mechanical flow without knowing its size or timing is guesswork wearing a suit. The value of knowing the calendar is purely defensive: you stop misreading volume, and you avoid holding size in a contract that is quietly emptying out.

Practical handling for a chart trader

Keep it simple. Trade and mark levels on the instrument you actually have exposure to, and do not import a level drawn on an adjusted futures chart onto a live spot chart or the reverse. If you reference futures for volume and open interest, which is reasonable since spot publishes neither honestly, read the active month rather than the front month, and know roughly when the handover is due so that a volume spike does not get mistaken for conviction.

For most retail traders the cleanest arrangement is to execute on the instrument in the account, use the live chart for structure, and treat futures data purely as context. One last point on discontinuities: the step at a roll is an artefact of stitching, whereas the breaks discussed in weekend gaps on gold are real price jumps that real positions lived through. Confusing a bookkeeping seam with a genuine gap leads to drawing levels on nothing at all.