Metal stocks: separate warrant cancellation and load-out
The metal underlying a cancelled warrant can remain in a warehouse. A physical withdrawal can also move metal to another storage location rather than a factory. Research should reconstruct warrant-state changes and physical movements separately before interpreting an exchange stock headline as industrial consumption.
The deliverable is a metal-and-location stock ledger with live warrants, cancelled warrants, deliveries in and deliveries out. It preserves report dates, delays and coverage. All quantities below are fictional and expressed in thousands of tonnes; this article does not provide a metal-price forecast or interpret current sanctions eligibility.
Locate the reporting and physical distinction
LME’s Warehouse and stock reports page describes separate stock, movement, queue and off-warrant reports, including different reporting frequencies and delays. Locate Stocks breakdown report, Warehouse and queue data and Off-warrant stock reporting. The reports do not all describe the same population on the same observation date.
LME Notice 24/171, dated April 13, 2024, paragraph 16 on printed page 4, explicitly distinguishes warrant cancellation from underlying metal being delivered out. This article uses that limited physical distinction only. It does not extend the notice’s 2024 sanctions arrangements into a statement about present trading rights or every jurisdiction.
For a real ledger, obtain metal, warehouse or location, observation date, publication date and the provider’s field definitions. Confirm whether the numbers are tonnes, lots or warrant counts. A warrant count cannot be added to a tonne balance without matching quantities.
Rebuild the opening and cancellation case
Assume reported opening stock consists of eighty live and twenty cancelled, totalling one hundred. During the day, fifteen move from live to cancelled status. There is no physical movement at that step.
Live becomes sixty-five, cancelled thirty-five and total remains one hundred. Cancelled share rises from 20% to 35%, but the warehouse still holds the same reported quantity. Subtracting fifteen from total stock would confuse a state transfer with metal leaving the reporting population.
| Step | Live | Cancelled | Reported total |
|---|---|---|---|
| Opening | 80 | 20 | 100 |
| Cancel fifteen | 65 | 35 | 100 |
The example defines total as live plus cancelled stock within its stated reporting boundary. Do not rename it a global stock total or assume another report’s use of on-warrant means exactly the same thing. Preserve each source’s categories before combining series.
Add actual deliveries and re-registration
Assume eight physically leave from the cancelled population and five physically arrive and enter live status. After the withdrawal, live remains sixty-five and cancelled falls to twenty-seven, totalling ninety-two. After the arrival, live becomes seventy and cancelled remains twenty-seven, totalling ninety-seven.
Now assume five of the remaining cancelled quantity are re-registered into live status. Live becomes seventy-five, cancelled twenty-two and total remains ninety-seven. This is another state transfer, not five thousand tonnes of new metal arriving.
| Event | Live change | Cancelled change | Total change |
|---|---|---|---|
| Cancellation | −15 | +15 | 0 |
| Physical withdrawal | 0 | −8 | −8 |
| Arrival placed live | +5 | 0 | +5 |
| Re-registration | +5 | −5 | 0 |
| Closing balances | 75 | 22 | 97 |
The physical identity is 100 + 5 − 8 = 97. The state identities are 80 − 15 + 5 + 5 = 75 and 20 + 15 − 8 − 5 = 22. These independent checks expose double counting. Subtracting both cancellation fifteen and withdrawal eight would give the wrong physical stock decline.
The five incoming quantity’s state is an explicit assumption. If the actual arrival does not enter the same live category, use the actual report treatment. A physical receipt and a new warrant record can occur on different clocks; do not force them into one event when the evidence does not establish that timing.
Test a different boundary: location transfer
Suppose eight leave Location A but the same eight arrive at Location B within the broader reporting network. A’s stock falls, while the combined two-location physical total is unchanged if the periods and coverage match. This is a transfer, not eight of demonstrated end-use consumption.
If B reports later than A, the current published snapshots can show an apparent decline before the receipt appears. Store observation and publication dates separately. Do not treat the temporary timing gap as proof that metal vanished or that one report is false.
Even a confirmed departure from the exchange reporting population does not show the destination was a consumer. Metal can move to non-exchange storage or another jurisdiction. Industrial consumption requires separate evidence about actual use, not merely a load-out record.
Preserve queue and off-warrant coverage
Cancellation can indicate an intention or arrangement relevant to withdrawal, but does not establish its timing. Obtain location-specific queue and delivery information before assuming the whole cancelled balance will leave tomorrow. A monthly waiting-time observation is not automatically the schedule for a particular warrant.
Off-warrant stock has its own report definition and coverage. Do not add a monthly off-warrant figure to a daily main stock figure until overlap, location and dates have been checked. The exchange’s approved-warehouse reports also do not establish every private warehouse, producer stock or consumer inventory worldwide.
If coverage is incomplete, state that exchange-reported stocks changed within the defined population. Do not upgrade that observation to global physical scarcity, a guaranteed price move or a measure of final demand. Those require additional data and an explicit market model.
Read proportions with both numerator and denominator
After all main-case events, cancelled share is 22 ÷ 97, approximately 22.68%. That differs from the earlier 35% after cancellation because physical movements and re-registration changed the components. A daily percentage alone cannot tell which mechanism occurred.
Keep absolute live and cancelled tonnage beside the percentage. A share can rise because the denominator falls, even with no new cancellation. Conversely, re-registration can lower cancelled share without any withdrawal. Match the underlying event bridge rather than infer demand from the direction of the ratio.
Use the stock verification sheet
| Field | Obtain | Check |
|---|---|---|
| Population | Metal, warehouse and field definitions | Tonnes, lots and warrants not mixed |
| State | Opening/closing live and cancelled | Transfers preserve the defined total |
| Physical movement | Deliveries in/out | No cancellation counted again as withdrawal |
| Location | Origin, destination and report coverage | Internal transfer distinct from consumption |
| Timing | Observation and publication dates | Different delays reconciled |
| Other stocks | Off-warrant and external coverage | No unsupported overlap or global total |
When balances disagree, inspect re-registration, corrections, metal/location scope and data timing before assigning a residual to hidden consumption. Keep the residual and the missing event record visible. A closed total does not prove every individual state was reported correctly.
If movement detail is unavailable, report the observed stock change and leave the decomposition unknown. A large cancelled balance cannot replace missing physical outflow data. Update after revised reports, re-registration, observed deliveries or a coverage change. The conclusion should state which metal changed status, which physically crossed the reporting boundary and what remains unknown about its destination or use.
Turn the bridge into an event ledger
Keep one event identifier for each movement, with separate columns for physical quantity and warrant-status quantity. In the cancellation step, the physical-change column is zero while the status columns move fifteen from live to cancelled. During a load-out of eight thousand tonnes, the physical column is minus eight and the relevant status row declines by eight. A chart should not stack these two columns as independent withdrawals.
The case can be checked after every step. Opening live eighty and cancelled twenty total one hundred. Cancellation yields sixty-five and thirty-five. Assuming the delivery of eight thousand tonnes leaves the cancelled category and the receipt of five thousand tonnes is warranted live, the balances become seventy live and twenty-seven cancelled, total ninety-seven. Re-registering five of that cancelled stock gives seventy-five live and twenty-two cancelled, still ninety-seven. No new metal is created by that last entry.
If only daily opening and closing totals are available, these individual events are not uniquely recoverable. A net decline of three could conceal many different gross receipts and deliveries. Label the movements inferred only from a difference as net change, and keep independently reported gross flows separate. The hypothetical decomposition is useful because its inputs are explicitly supplied, not because every published balance reveals them.
Record a local boundary for each event as well. Metal transferred between two covered warehouses can leave one location and enter another without changing the combined system total. Metal in transit can create a timing difference between those local observations. Until the receiving record is obtained, show the unmatched leg rather than call it consumption or force a same-day receipt.
These checks also clarify the limits of a tight cancellation ratio. It indicates the composition of the selected reported stocks, not the amount already used by industry or the number of days every warehouse will need to dispatch metal. Destination, shipping and end-use evidence answer those later questions.
FAQ
Does cancelling fifteen mean fifteen thousand tonnes physically left?
No. The main case leaves total stock unchanged at that step.
Does confirmed load-out establish industrial consumption?
No. The destination and actual use need separate evidence.