Quick answer: an FX trade exchanges one currency for another, but the two payments may travel through different payment systems, currencies and time zones. If one side pays away the currency it sold before receiving the currency it bought, it can lose the full principal if the counterparty fails in between. Payment-versus-payment (PvP) changes the settlement rule so that the final transfer of one currency occurs if and only if the final transfer of the other occurs. Netting reduces the amount that needs to be paid; PvP changes the conditional logic of final settlement.

The dangerous interval is the time after “I have paid” but before “I have received.”

Why this belongs in mathematics and algorithms

Foreign-exchange settlement is a problem of synchronisation, state, conditional execution, timing windows, netting, liquidity and finality. The exchange rate determines what is owed. The settlement algorithm determines whether the two obligations complete safely.

This is different from asking whether the trade was profitable. A trade can have been priced perfectly and still create a settlement loss if one currency is delivered without receiving the other.

1. The basic two-leg problem

Suppose Bank A agrees to sell US$10 million to Bank B in exchange for ¥1.5 billion on settlement day.

Bank A owes US$10 million.
Bank B owes ¥1.5 billion.

Economically, these are two sides of one FX trade. Operationally, they can be two separate payment instructions moving through different currency systems.

If Bank A’s US-dollar payment becomes final first and Bank B fails before its yen payment becomes final, Bank A may have delivered the full US$10 million principal while receiving nothing. That is the core settlement exposure.

2. Herstatt risk: why the name exists

FX principal settlement risk is often called Herstatt risk, after the 1974 failure of Bankhaus Herstatt in Germany. Its closure occurred after counterparties had made payments in Deutsche marks but before corresponding US-dollar payments were completed in New York. The event made visible a structural problem created by non-synchronised settlement across currencies and time zones.

The name is historical; the mathematics is general. Whenever two linked obligations can become final at different times, there is an interval during which one party may have performed while the other has not.

3. Principal risk is not the only FX settlement risk

Risk	Question
Principal risk	Could one party pay the currency sold and fail to receive the currency bought?
Replacement-cost risk	If the counterparty fails before settlement, what does it cost to replace the trade at current market rates?
Liquidity risk	Can required currencies be funded at the right time, even if the institution is solvent?
Operational risk	Can systems, instructions, confirmations and cut-offs execute correctly?
Legal risk	Are netting, settlement finality and default rules enforceable?

Basel supervisory guidance treats these as distinct but connected risks. A system can eliminate principal risk through PvP yet still need to manage liquidity, operations and legal finality.

4. Payment-versus-payment changes the logical condition

Without PvP, a simplified process can look like:

Send currency A → wait → receive currency B.

PvP changes that to:

Final(currency A) if and only if Final(currency B).

The two transfers become conditionally linked. If the required counter-payment is not available under the system’s rules, the settlement pair does not complete as a one-sided final transfer.

This is an algorithmic control, not a prediction about counterparty quality. It reduces principal exposure by changing the settlement mechanism itself.

5. Netting and PvP solve different problems

Suppose two banks have hundreds of FX trades with each other during the day. Gross settlement might require very large two-way payments. Netting compresses offsetting obligations.

If Bank A owes Bank B US$100 million across one set of trades and Bank B owes Bank A US$94 million across offsetting obligations that are eligible for netting, the net dollar funding requirement can fall toward US$6 million rather than both parties funding their gross amounts.

But bilateral or multilateral netting by itself does not necessarily eliminate settlement risk on the remaining net amount. PvP addresses the condition under which final transfers occur. This distinction is central to the updated global settlement-risk architecture.

6. CLS: synchronisation at global scale

CLSSettlement is a major global PvP infrastructure for FX. It links the two currency payments so that they settle simultaneously under its settlement process. CLS currently states that its main service settles more than US$8 trillion of payments each day across 18 currencies, with multilateral netting reducing funding requirements by more than 96% on average.

See the current CLSSettlement description.

The key point is not the brand name. It is the architecture: matched eligible instructions, common settlement rules, multilateral compression of funding obligations, linked final settlement and strong legal/operational controls.

7. What the latest global data show

The BIS published new settlement-method measurements from its 2025 Triennial Survey in June 2026. It reported that just over US$5 trillion, or 36% of average daily FX settlement during April 2025, used PvP. Another US$7.6 trillion, or 54%, used methods such as pre-settlement netting that reduce but do not eliminate settlement risk. More than US$1.4 trillion, or 10%, still settled gross bilaterally and remained fully exposed to settlement risk.

See the BIS Quarterly Review study Uncovering FX settlement risk: new measures from the 2025 BIS Triennial Survey.

This evidence matters because it prevents a naturalised assumption: “the Herstatt problem was solved decades ago.” It was greatly reduced in important parts of the market, but it has not disappeared.

8. A simplified PvP state machine

Trade execution. Two parties agree on currencies, amounts, rate and settlement date.
Trade capture. Both sides record the transaction.
Matching. Critical economic terms must agree.
Eligibility check. The trade, counterparties and currencies must be supported by the settlement route.
Netting/compression. Eligible obligations are combined to reduce funding needs.
Funding calculation. Each participant learns the currency amounts it must make available.
Funding receipt. The infrastructure confirms required pay-ins under its rules.
PvP settlement. Linked currency legs settle under the condition that both sides complete.
Finality. The settled payments become final and irrevocable according to the governing legal framework.
Reconciliation. Internal books, payment records and settlement results are matched.

9. Timing still matters even with safe settlement logic

A bank may owe several currencies during a funding window. It can be economically solvent yet fail to have the right currency available at the right moment. This is liquidity risk.

Imagine an institution expects incoming euros later in the day but must fund an earlier euro settlement obligation now. The future inflow does not pay the present deadline. The institution may need intraday liquidity, credit lines or foreign-exchange funding to bridge the timing mismatch.

This is the same mathematical lesson found in bank liquidity stress testing: totals are insufficient when timing constraints matter.

10. Why gross bilateral settlement is structurally different

In a gross bilateral arrangement without PvP, the full principal of one leg can become exposed. The exposure may last from the point at which the outgoing payment can no longer be cancelled until the incoming payment is finally received with certainty.

The size of settlement risk therefore depends on both:

amount at risk, and
duration of the irrevocable-but-not-yet-received window.

Good settlement design tries to shrink both.

11. Why settlement instructions must be correct before they become final

Automation does not rescue bad input. A wrong beneficiary account, incorrect currency amount, duplicated instruction or unmatched trade can become a highly efficient error if the system processes it automatically.

This is why trade matching, cut-off management, exception handling and reconciliation remain part of the settlement algorithm. The companion article How Banks Reconcile Transactions explains the post-movement record problem in more detail.

12. Creative-work lens: operational drama is usually less visible than market drama

Financial films often focus on traders, prices and spectacular losses. Works such as Rogue Trader are useful reminders that operational controls, booking, confirmation and independent records can matter as much as the visible market position. But the film is not evidence for FX settlement architecture. The useful transfer is narrower: a financial system can fail because records, controls and execution states do not line up with economic reality.

13. Failure modes

Assuming netting equals PvP. Funding is compressed but remaining principal still settles one side at a time.
Coverage blindness. A bank assumes every currency or counterparty can use the same PvP route.
Cut-off failure. An instruction misses a deadline and falls to a less protected route.
Liquidity mismatch. The right total funding exists but not in the required currency at the required time.
Operational mismatch. Trade terms disagree across systems.
Legal-finality uncertainty. Participants assume settlement is irreversible when legal treatment is unclear.
Gross bilateral complacency. Principal exposure is treated as a minor operational detail rather than a full-value credit exposure.
Concentration. Large flows rely on one route, one nostro provider or one operational dependency without adequate fallback.

14. Diagnostics and falsifiers

What percentage of daily settlement is true PvP?
What percentage is netted but still exposed on the residual principal?
How long is each currency’s irrevocable-to-final-receipt window?
Which currencies cannot use the preferred PvP infrastructure?
What happens if one major funding bank or nostro account is unavailable?
Can settlement complete if a participant funds one currency late?
How many trades fall out of automated matching into manual repair?
Are all finality and netting assumptions legally documented?

Suppose someone claims, “We net our FX trades, therefore settlement risk is eliminated.” A falsifier is any remaining net payment that can be released irrevocably before the counter-currency is finally received. If that state exists, principal settlement risk remains.

15. Verification and update triggers

reconcile settlement records against trade records and cash accounts;
measure gross, netted and PvP-settled volumes separately;
test contingency routes for unavailable currencies or infrastructure;
monitor missed cut-offs and settlement fails;
review intraday liquidity needs by currency;
update controls when new currencies or products become eligible for PvP;
reassess legal finality when jurisdictions or rules change;
use new BIS survey evidence to test whether old assumptions about market-wide coverage remain true.

Connections across the finance-and-banking algorithms lane

Payment systems: graphs, queues, routing and real-time settlement — the wider payment infrastructure beneath currency movements.
Transaction reconciliation — how internal records are made to agree after movement.
Liquidity stress testing — why timing and currency availability matter.
Counterparty credit risk — related credit exposure before and around settlement.

Research anchors

The deeper lesson

FX settlement risk shows that two obligations can be economically inseparable but operationally disconnected. PvP repairs that by making finality conditional across the pair. Netting reduces how much must move. Liquidity management makes sure the required currencies arrive on time. Reconciliation proves that the system’s records match what actually settled. Good finance algorithms do not merely calculate value; they control the state transitions by which value becomes final.

Educational note: This article explains public payment-system and banking concepts. It is not trading advice, financial advice, legal advice or institution-specific settlement guidance.

Small Group Tutorials

How Foreign-Exchange Settlement Algorithms Reduce Principal Risk: Herstatt Risk, Netting, Payment-versus-Payment and CLS