Commodity houses don’t sell products; they arbitrate time, quality, and location under tight credit and currency constraints. A single copper lot can be financed by an LC at shipment, hedged on the LME, priced on a rolling quotational period, and finally settled after the refinery’s assay—while treasury juggles USD margin calls, CNH payables, and EUR fees. When pre-trade credit, trade documentation, price discovery, and cash operations are treated as one system, P&L stops leaking into “other adjustments” and working capital becomes predictable instead of theatrical.
Credit before cargo: pre-trade limits and collateral you can actually monetize
Trading starts with credit scaffolding—if limits and collateral rules live in emails, finance will spend months explaining avoidable losses.
- Counterparty limits: set exposure caps per buyer/seller by tenor and by instrument (open account, collection, LC, UPAS). Split by country and sanction risk; keep “eligible bank list” for confirmation/discounting.
- Borrowing base: inventory and receivables finance only works if field warehouse receipts, metal warrants, or tank receipts are controlled by an independent collateral manager. Eligibility criteria (age, location, grade, insurance, lien status) should be computable, not negotiated at month-end.
- Initial & variation margin: if you hedge on LME/COMEX/ICE, margin calls arrive in USD (or exchange base currency) on exchange calendars, not on your A/P calendar. Treasury needs a ring-fenced liquidity buffer plus auto-sweep rules so desks can’t starve operations to feed margin.
Treat all three as deal prerequisites. If a trader can fix a cargo without an approved limit and margin headroom, the cash team is already underwater.
Letters of credit, collections, and UPAS: picking the right trade instrument
Instrument choice dictates cash timing and fee stack.
- Sight LC (UCP): bank assures payment on compliant docs; exporter monetizes on presentation. Good for new counterparties or tougher jurisdictions.
- Usance LC: buyer gets tenor; exporter can discount (with or without recourse). Watch the discount curve vs the forward curve on your hedge.
- UPAS LC (Usance Payable at Sight): buyer enjoys deferred payment; seller is paid at sight by financing bank—useful when you want shipping liquidity without loading buyer credit risk onto your balance sheet.
- Documentary collection (URC): cheaper but weaker; D/P (documents against payment) holds title until cash lands; D/A (documents against acceptance) creates time-draft risk you must measure against the buyer’s track record.
- Open account with credit insurance or standby LC when counterparties and jurisdictions justify the risk/price trade-off.
Whatever you pick, document sets must be clean: bill of lading type (to order/straight), weight notes, certificate of origin, inspection & analysis certificates, insurance, and any sanctions/attestations your bank demands. One missing stamp turns a 3-day LC into a 30-day problem.
Assay, weights, and the quotational period: why “final price” isn’t final until it is
Metals and concentrates are not priced like sugar.
- Provisional vs final: invoice provisionally on shipment using QP conventions (e.g., “M+1 average LME cash”) and a provisional assay; final invoice adjusts for the refinery’s umpire assay and the realized QP.
- Deductions & penalties: TCRC for concentrates, moisture, impurities (As, Sb, Bi, Pb) cut payables; premiums for shape/brand (ingots, cathodes, warrants) add back. Price sheets should be parameter files, not PDFs.
- Assay governance: nominate labs and umpire procedures in the contract. Your ledger must store the assay version that drove each pricing event; disputes then become arithmetic, not rhetoric.
- Hedge alignment: hedge volume and maturities to expected QP; if the physical flips to a different QP, treasury must roll or layer positions deliberately—no “catch-up” trades without paper trails.
The reconciliation object for a single lot should show: provisional price math, hedge IDs, final assay deltas, QP deltas, and net FX effect. If it doesn’t, “price variance” becomes a laundry basket for real money.
FX exposure: USD paper, local cash, and basis you forgot to book
Most base metals clear in USD, but suppliers, fees, and taxes are rarely all USD.
- Structural USD: exchange margin, futures P&L, freight, and many port/inspection fees want dollars.
- Local legs: royalties, VAT/GST, trucking, and handling often land in local currency you can’t net against exchange cashflows.
- Policy: natural hedge first (collect/spend in same currency); lock deal-level rates when you fix a provisional invoice; lock payment-instruction rates on the cash day; report FX in basis points of lot P&L—split realized vs translation, hedge vs unhedged.
- Instruments: short-dated forwards and NDFs for non-deliverable currencies; options (collars) for cargoes with skinny contribution and long QPs.
Don’t roll 100% coverage on one day; ladder coverage against shipment schedules and QP profiles.
Rails and calendars: SWIFT for value, local rails for cost
Big tickets and multi-bank deals still ride SWIFT, but local rails belong in your stack.
- Outbound: SWIFT for bank-to-bank under LCs and collections; local ACH/instant rails for domestic fees, storage, trucking, and taxes. Publish value-date rules per corridor; “we paid yesterday” doesn’t satisfy a counterparty whose funds arrive T+3 with charges deducted.
- Inbound: assign virtual accounts/IBANs per counterparty and, for concentrates, per contract; remittances auto-segment and cash app stops reading PDFs at 2 a.m.
- Cut-offs & holidays: build a trade calendar that overlays exchange (LME/COMEX) cut-offs, bank cut-offs, and port working hours. A Thursday margin call plus a Friday local holiday equals an expensive lesson if buffers aren’t in place.
Measure cost per successful payment (fees + FX + ops minutes) and arrival variance (promised vs actual value date). Those two metrics expose under-the-line leakage fast.
Freight and insurance: aligning cash with Incoterms instead of emails
Incoterms decide which side pays which leg; your invoice logic must encode that.
- FOB/FCA: buyer lifts at origin; freight is buyer’s problem—but you still see port costs until the ship’s rail.
- CFR/CIF: you carry ocean cost (and insurance under CIF); separate freight/insurance lines with currency and rate metadata so their FX doesn’t pollute commodity P&L.
- DAP/DDP: you own the destination pain; make sure duty/VAT and local charges are priced and cash-forecasted before sailing.
Demurrage/detention is not “misc.” Rate it from terminal timestamps and keep contracts (free time, tariffs) in a rules file the ledger understands.
Sanctions, origin, and attestations: cash can’t outrun paperwork
Banks care about origin, route, and counterparties.
- Screen vessels, shippers, consignees, and banks at fixture and pre-presentation.
- Keep proof of origin and brand/warrant lists current; some brands are restricted or require additional attestations.
- For gold and certain minor metals, extra diligence on “responsible sourcing” can be a documentary precondition to banking the deal—even if not a marketing talking point.
Routing logic should block restricted corridors automatically; finance should not be the last line of defense via email.
Data model: if one lot spans ten events, store ten events
A lot is a chain of monetary states. Make it computable.
- Money representation: integers in minor units plus currency code; every monetary event (prov/final invoice, hedge P&L, fees, margin call, payment) carries rate, source, timestamp, and a link to the pricing parameters or contract clause used.
- Assay & weight: versioned objects with lab IDs, methods, and umpire outcomes; pricing pulls a specific version by key.
- Hedge map: paper IDs and fills tied to physical lots; realized/unrealized P&L snaps at the time of each physical pricing event.
- Variance classes: QP delta, assay delta, weight delta, FX, fees, demurrage/detention, financing cost. Exceptions are categorized with owners and SLAs.
When the board asks why a cargo missed its desk budget by 42 bps, you show the waterfall by class—not a slogan.
Paying and getting paid: netting that doesn’t hide audit trails
Netting reduces wire count but can wreck explainability if you’re careless.
- Bilateral netting with a counterparty: fine, but line-level mapping must preserve which invoices, hedges, and fees were extinguished.
- Multilateral netting via a central account: only if you can still prove tax, FX, and origin by line in your books and the counterparty’s statement.
- Interest & discounting: usance LC discount, early-pay options on open account—post them as explicit finance lines, not as blended price changes.
Auditors don’t mind netting; they mind missing provenance.
What to monitor if you actually want stable contribution
- Margin utilization vs policy and days at risk with buffers below threshold.
- Hedge coverage & alignment to QP; slippage when physical QP or volumes change.
- Provisional vs final price variance by product and counterparty.
- FX cost in bps of lot P&L; covered vs uncovered share.
- Payment arrival variance and cost per successful payment.
- Auto-match rate (volume and value) on AR/AP with ISO 20022 statements and virtual accounts.
- Document exceptions: LC discrepancies per presentation and their fee impact.
Dashboards that surface these dials weekly do more for P&L than any after-the-fact post-mortem.
Where a payment intermediary actually helps
Trading houses that span Africa, LATAM, and Asia spend real time wiring local fees, paying inspection companies, refunding LC discrepancies, and moving cash to/from field warehouses. A specialist such as Collect&Pay can provide multi-currency accounts, virtual IBANs per counterparty or contract, and local rails for port and logistics bills—plus line-level fee/FX breakdowns. Corridor breadth and failure-recovery speed matter more than a few bps on headline fees.