Glossary

Useful as a top-line gauge of vault activity, but Registered is what matters for an actual delivery squeeze.

Flat Combined + falling Registered = metal reclassified from Registered to Eligible (de-warranting); bars stay in the vault. Possible reads:

• Tightening — owners pulling metal off the deliverable pile, unwilling to part with it at current prices.
• Mundane — warrant/storage cost management, someone who took delivery de-warranting to hold long-term, ownership consolidation.

Caveat: the same headline numbers can also arise if Registered is delivered out and fresh Eligible flows in — so confirm against open interest, delivery notices, and lease rates before calling it.

We compute it over 7d / 30d / 90d windows: (Combined today − Combined then) / days. Negative = net outflow.

Treat any 'days-until-empty' figure as a crude linear extrapolation, not a forecast.

Note: combined inventory is sticky. Delivery itself transfers warrants (title), not metal — the bars stay in the depository — so it doesn't directly drain combined stocks. Delivery-related stress shows up faster in Registered inventory and the Registered/Eligible split than in this number.

Warranting it (eligible → registered) is mostly paperwork, often 24–48h, but it's entirely the owner's choice. Eligible is usually the larger category, though the split can invert (gold, early 2026).

From First Notice Day through the last delivery day, shorts who hold their positions issue delivery notices (making delivery) and longs get assigned and stop them (taking delivery). Anyone who doesn't want to make or take delivery must offset their position instead — and that's only possible while the contract still trades, up to Last Trading Day (for longs wanting to be safe, before FND).

For silver: (front-month OI × 5,000 oz) / registered_oz. For gold: (front-month OI × 100 oz) / registered_oz. Above 1.0 means there's more claimed metal in front-month paper contracts than there is deliverable metal sitting at COMEX.

Note: this is the inverse of the conventional 'coverage ratio' (registered ÷ OI). The two carry the same signal — a coverage ratio below 1 corresponds to a paper-to-physical above 1 — but the directions read oppositely. We show both side-by-side so you can sanity-check against whichever convention you're used to.

The metric is approximate — most contracts settle in cash and never demand delivery — but extremes signal physical stress.

Warranting — done by the approved depository at the owner's request — flips a bar from Eligible to Registered with no physical movement; the metal stays privately owned, not pledged to the exchange. Owners can cancel warrants to move bars back to Eligible, shrinking deliverable supply. Only registered metal is immediately deliverable, though eligible can be warranted and then delivered.

SHFE publishes only this on-warrant figure — there is no separately disclosed total-deliverable or 'eligible' bucket.

Because no total-inventory figure is published, a fall in on-warrant tonnage is ambiguous from SHFE data alone: you can't tell whether bars physically left the approved warehouses or simply had their warrants cancelled while staying put.

The name is a false friend with COMEX, where 'Eligible' means the opposite (metal stored in an approved vault but not committed to delivery); here 'eligible' is used in its plain sense of eligible for delivery — the committed tier that COMEX calls Registered and SHFE calls on-warrant.

Despite the name, the LBMA owns the benchmarks but doesn't operate the auctions; they're run by independent administrators — ICE Benchmark Administration (IBA) for gold and silver, and the London Metal Exchange (LME) for platinum and palladium.

In each auction, accredited participants submit buy and sell volumes over successive rounds until the imbalance falls within a set threshold, and the final clearing price (in US dollars) becomes the published benchmark.

Silver fixes once daily at noon; gold twice daily at 10:30 (AM) and 15:00 (PM); platinum and palladium twice daily at 09:45 (AM) and 14:00 (PM) — all London time (GMT in winter, BST in summer).

The metal physically present spans ETF backing, central-bank reserves, allocated client accounts, and the bars underpinning unallocated balances, but the LBMA publishes a single combined total per metal — not a breakdown by owner or category.

Figures are reported monthly in metric tonnes (with troy-ounce, bar-count, and USD-value equivalents), measured at each month-end.

The CFTC's 'money manager' category: registered Commodity Trading Advisors (CTAs), registered Commodity Pool Operators (CPOs), and unregistered funds — with hedge funds landing here regardless of registration status ('registered' means CFTC/NFA registration, not SEC). They don't produce, process, or use the metal, so they sit on the speculative side of the old commercial/non-commercial divide and serve as the standard trend-following sentiment gauge.

Because they mostly chase trends, the contrarian read applies at the extremes — where positioning is crowded and exhausted — not to the direction of the position in general, and it's a heuristic, not a rule.

Mind the lag: each report is a Tuesday snapshot published the following Friday afternoon. Net position alone misses intensity, so pair it with net-as-a-share-of-open-interest (scale) and the largest-4 / largest-8 concentration figures (how few hands hold it).

Positions held by traders too small to meet the CFTC's reporting levels. Unlike the four reportable categories, these traders are never classified individually; the figure is derived as a residual — total open interest minus every reportable position — and shown only as aggregate long and short.

It's predominantly small and retail participants, though the CFTC doesn't actually identify who's in it. As a signal it's the weakest bucket: the popular read is to fade it at extremes (treating retail as contrarian), but it's slow-moving and noisy, so it's better treated as context than as a trigger.

Every trader above the CFTC's position-reporting thresholds that the Commission doesn't slot into the other three categories ends up here — typically proprietary trading firms, family offices, large individual traders, and some institutional or sovereign accounts.

Because it's defined purely by exclusion, it's a catch-all rather than a coherent group, and the least narratively useful of the four. Read it as a residual, not a signal.

Entities that predominantly produce, process, or use the physical metal and use futures to hedge business risk (swap dealers are excluded — they're broken out separately). In metals they're typically net short: producers and refiners sell futures forward to lock in prices on expected output, which outweighs the smaller offsetting buying from fabricators and industrial users.

A shrinking net short can hint that hedgers are lifting hedges — sometimes read as anticipating higher prices — but it can equally reflect changes in production, inventory, or financing, so treat it as a soft signal, not a directional call. Unlike the other categories, P/M reports only long and short, with no spreading column.

Entities dealing primarily in commodity swaps that use futures to offset the risk from that client book. Their net futures position reflects the net of the client exposure they've absorbed, not a house view, so a large net short in gold or silver is structural hedging rather than a bearish call.

Direction varies with the client book and isn't fixed; the common 'swap-dealer net tracks ETF flows' shorthand is unreliable for precious metals, since the big ETFs are physically backed and don't hold futures.

The gap between Swap Dealer net and Managed Money net is a useful rough read on hedging/structural flow versus speculative flow — context, not a clean institutional-vs-speculative signal.

All contracts past the active front month. Liquidity thins quickly beyond the front and its nearest one or two months, so spreads widen the further out you go. Commercials use back months to hedge production or consumption that runs past front-month expiry; spread traders use them to trade the shape of the curve via front-vs-back calendar spreads.

Near-dated futures priced higher than longer-dated ones. In consumable commodities this reflects immediate physical scarcity or strong spot demand.

For monetary metals like gold and silver it's unusual — huge above-ground stocks keep the curve normally in contango — so backwardation there usually signals a squeeze in the borrowing/lease market (the cost to borrow physical metal rising above the financing cost of holding it) rather than literal consumption shortage.

When it does appear it's noteworthy and has often accompanied delivery squeezes, refiner and coin shortages, and short squeezes — a flag to investigate, not a mechanical buy/sell signal.

Longer-dated futures priced above near-dated ones. It's the typical shape for storable goods because holding the physical costs money — storage, insurance, and financing — and the curve's upward slope roughly approximates that cost of carry.

For gold and silver the carry is dominated by financing (interest rates), with storage and insurance small relative to the metal's value. Unusually deep contango in a market that's normally tight can point to excess supply or a glut.

The active contract closest to expiry that hasn't yet entered its delivery window. These markets don't trade every calendar month equally: COMEX gold's liquid months are the even months (Feb, Apr, Jun, Aug, Oct, Dec), and COMEX silver's are March, May, July, September, and December. The front month is simply the nearest active contract month — the closest-to-expiry contract on that cycle that's still trading ahead of delivery — not just next calendar month. Traders roll out of it before First Notice Day (roughly a week ahead) to avoid being caught for delivery, and the next active month then becomes the front.

Note: some price feeds build a "continuous front month" by rolling to whichever contract is most active (highest open interest/volume) rather than the nearest, so a chart may flip to the next contract a few days before the calendar would.

Cycle months (source): CME Silver Futures & Options fact card · CME Gold Futures & Options fact card

Live open interest & volume by expiry: Gold · Silver

See also: First Notice Day (FND) · Last Notice Day (LND) · First Delivery Day (FDD) · Last Delivery Day (LDD)

The relationship between prices of contracts expiring at different dates; plotting front-month, second-month, third-month and so on against expiry traces the futures curve.

Two main shapes: contango (back > front), the normal state, where deferred prices carry storage and financing costs; and backwardation (front > back), signalling near-term tightness, with the market paying up for metal now.

Curves aren't always cleanly one or the other — they can be humped or mixed (in contango at the front, backwardation further out, or vice versa), so read the whole curve rather than just the nearest spread.

Most contracts are closed before expiry via offsetting trades — only a small fraction (typically a few percent) goes to physical delivery. Those that do follow exchange-defined procedures: the short tenders a warehouse warrant for bars meeting contract specs (size, purity, approved brand) and the long pays the invoice price (based on the settlement price). In practice the bars usually stay in the vault — it's the warrant that transfers. COMEX, SHFE, and MCX each have their own approved depositories and delivery windows; on COMEX, the daily Issues & Stops report shows which firms delivered to whom.

The number of contracts that remain open — not yet closed, expired, or delivered (each contract counted once, across its long and short side). Unlike volume, which counts every contract traded during the day (including ones opened and closed the same session), OI measures what's still standing at the close.

Rule-of-thumb reads: rising OI with rising price suggests new longs; rising OI with falling price suggests new shorts; falling OI suggests positions being covered or liquidated.

OI × contract size gives the notional paper claim on metal, which divided by Registered stock gives the Paper-to-Physical Coverage Ratio — though most OI is offset well before delivery, so treat the ratio as a structural/sentiment gauge, not a literal delivery demand.

Each business day the exchange computes a settlement price for every active contract — typically a volume-weighted average over a defined closing range. All open positions are marked-to-market against it, with gains and losses credited or debited daily (variation margin), and it drives margin requirements and options moneyness at expiry. It's the figure quoted as the contract's official daily close. (Note: it is not the same as the LBMA Fix, which is set by its own auction — see that entry.)

Convert SHFE's quote — CNY/kg for silver, CNY/gram for gold — to USD/oz using the live USD/CNY rate and 31.1035 g/oz, then compute the % gap versus COMEX.

The VAT treatment differs by metal, and getting it right matters:

• Gold — effectively VAT-neutral. Standard/investment gold traded on the exchange carries a long-standing exemption (with a full Special VAT Invoice offset on withdrawal), so no VAT adjustment is applied to the gold premium.
• Silver — a real 13% VAT applies, but it's deferred until metal is physically withdrawn for domestic use, so whether to strip it from the benchmark is a methodology choice. One camp (e.g. CPM Group) treats the SHFE silver quote as VAT-inclusive and removes the tax before comparing; another treats the traded benchmark as already a clean pre-VAT reference and makes no adjustment. We treat it as clean VAT reference.

A persistent positive premium signals Chinese demand outpacing imports (also shaped by import quotas, PBOC licensing, and capital controls); a persistent negative premium signals export pressure or weak domestic demand.

Note (Nov 2025 reform): China changed its gold VAT rules — exchange trading and investment-grade gold stay effectively exempt, but non-investment (jewelry) withdrawals lost the full offset. This doesn't affect the futures-premium math (which uses exchange prices), but matters if you extend the comparison to physical/jewelry channels.

COMEX is the primary international futures venue for silver, gold, copper, and other metals. Owned by CME Group since 2008. Silver contract = 5,000 oz, gold = 100 oz, both physically deliverable from a roster of CME-approved depositories. COMEX prices are the global reference for institutional metals trading; the Daily Metal Stocks Report and Daily Delivery Notices publish here.

MCX trades the SILVER and GOLD (plus mini-variants) futures contracts in India, with delivery against bullion stored in MCX-approved vaults operated by Brinks, Sequel, and Malca-Amit. Silver is quoted in INR per kilogram and gold in INR per 10 grams (the standard Indian convention). MCX also publishes daily warehouse-vault-wise stock position PDFs.

SHFE trades silver (Ag) and gold (Au) futures plus base metals (copper, aluminium, zinc, etc.). Silver quoted in CNY per kilogram; gold in CNY per gram. Day session 09:00–15:00 + night session 21:00–02:30 (next day), Beijing time (CST, UTC+8). The published "on warrant" inventory is the SHFE equivalent of COMEX Registered.

On VAT: Chinese exchanges quote tax-inclusive prices, but the effective burden differs by metal — investment/standard gold is effectively VAT-exempt (no adjustment needed), while silver carries a 13% VAT that's deferred until physical withdrawal. Whether to strip the 13% from silver for an international comparison is a methodology choice — see SHFE − COMEX Premium.