TP ICAP’s unsettled matched-principal notional rose 85% to £50.2bn

TP ICAP reported £50.211 billion of notional unsettled matched-principal purchases and £50.233 billion of corresponding sales at 31 December 2025, about 85% above a year earlier. These are gross contract amounts disclosed in a note, not £50 billion of assets and liabilities recognised on the face of the balance sheet. That accounting boundary changes how the apparent scale should be read.[1]

In a matched-principal trade, the intermediary enters a purchase and a corresponding sale. TP ICAP says these transactions are generally matched, normally settle within a few days and use delivery-versus-payment arrangements. The disclosed notional is a snapshot of contracts awaiting settlement; it cannot be treated as a directional market bet, expected loss or deposit balance.[1]

The year-end balance nearly doubled

The annual report records unsettled sales of £50.233 billion against £27.137 billion in 2024 and purchases of £50.211 billion against £27.155 billion. Calculating each change on its own prior-year base gives 85.1% and 84.9%. The £22 million difference between the 2025 notionals is not a measure of net economic exposure. Contracts, currencies, valuation and settlement arrangements matter beyond the difference between two totals.[1]

A separate cash-flow discussion says roughly half of the year-on-year movement associated with net settlement balances reversed immediately after year-end. That statement concerns the working-capital movement. It does not say that half of the £50.2 billion gross notional disappeared or that half of the increase in that notional reversed. The disclosures cannot be substituted for one another.[1]

How matched principal works

Under the model, TP ICAP enters one contract to buy from one counterparty and another to sell to another counterparty. It is principal to both legs even where the economic terms are matched. This structure can simplify settlement, allow multilateral netting and let counterparties face a recognised intermediary rather than each other directly. TP ICAP described the same logic when announcing that its Fusion Digital Assets venue would move to matched principal: clients could trade before prefunding, settle off-venue and use TP ICAP as the settlement counterparty.[2]

The model does not eliminate risk. Between execution and settlement, one counterparty can fail while the other remains entitled to performance. TP ICAP may then have to replace the failed leg at the prevailing market price. The annual report says the pre-settlement period is usually two or three business days and identifies replacement-cost exposure as the key risk. Delivery-versus-payment reduces principal risk by linking the movement of securities and cash, but operational failures, timing mismatches and counterparty default remain possible.[1]

Note 23 distinguishes the contract notionals from valuation and accounting balances. It reports fair-value gains of £177 million and losses of £176 million on the outstanding matched-principal transactions, alongside recognised matched-principal assets of £38 million and liabilities of £17 million. These are reporting-date measures, not annual trading gains and losses. None is interchangeable with the £50.2 billion notional.[1]

Three measures answer three questions

Notional measures the contractual scale of the unsettled purchases and sales. Fair value measures the value of the outstanding contractual positions at the reporting date. The recognised asset and liability figures reflect the accounting presentation described in the note. The first can be very large relative to the others without any contradiction: they measure different things.[1]

Dividing the purchase notional by annual profit or revenue would compare a point-in-time contract quantity with an annual income flow. Such a multiple is not leverage, a capital ratio or a stress loss. Likewise, applying an arbitrary number of basis points to the entire notional would not establish plausible exposure without a model of defaults, replacement prices, netting, collateral and recovery.[1]

The closely matched purchase and sale totals support the description of paired intermediation. They do not prove that every counterparty will perform or every settlement will complete on time. Both observations can be true: gross operational activity is large, and the amount exposed to a particular failure is not simply the full purchase notional.[1]

The number sits behind a profitable interdealer broker

TP ICAP reported 2025 revenue of £2.353 billion and profit before tax of £230 million. Wholesale-market intermediation can involve high contract values and a short settlement cycle while earning a comparatively small fee on each transaction. The annual income statement and the outstanding-contract note therefore describe different dimensions of the business.[1]

The 85% growth in outstanding notionals can reflect activity, product mix, customer composition or year-end timing. The cited note does not break that increase into those causes. The separate working-capital reversal provides no numerical answer to the gross-notional question. Average and peak figures would be needed to establish whether 31 December represented an unusually busy settlement date.[1]

What would complete the risk picture

A fuller disclosure would show average daily unsettled purchases and sales, peak balances, ageing beyond the normal settlement window, counterparty concentration, collateral held, failed-settlement rates and stressed replacement-cost exposure. It would also distinguish products settled through central counterparties from bilateral flows and show how enforceable netting agreements reduce exposures in different jurisdictions.[1]

None of those missing metrics justifies assuming a weakness. The annual report contains risk-management and fair-value disclosures, and the matched nature of the transactions is explicit. The point is that the largest visible number is a gross operational quantity, while the most decision-useful risk measures sit elsewhere or are not separately quantified.[1]

Gross settlement balances can concentrate even when market risk is matched

Matching removes much of the intended price direction, but it does not guarantee that both legal obligations perform simultaneously. If one counterparty defaults, disputes a trade or cannot deliver an asset, TP ICAP may still owe the other side. The potential loss is the cost of replacing the missing leg plus fees, funding and operational consequences, reduced by collateral, netting and recoveries.[1]

Concentration matters because a book that is balanced by product can still rely heavily on a small number of counterparties, settlement agents or currencies. The annual-report headline does not disclose how the £50.2 billion was distributed. A diversified set of small, highly rated counterparties presents a different risk from a few very large settlement obligations even at the same gross notional.[1]

Delivery-versus-payment reduces principal risk, not every failure mode

Delivery-versus-payment is designed so that cash and securities transfer together. It limits the danger of paying cash without receiving the asset, or delivering an asset without receiving cash. It does not remove cyber incidents, message errors, sanctions blocks, custodian outages, incorrect standing instructions or a counterparty's inability to fund settlement. Those operational events can delay or break an otherwise matched chain.[1]

The two-to-three-day pre-settlement window is also a period in which market prices can move. If a failed purchase must be replaced after a sharp price increase, or a failed sale after a fall, the replacement cost can become material even though the original legs were matched. Stress testing should therefore combine default probabilities with plausible market moves and the time needed to replace the trade.[1]

Why year-end and average exposure may tell different stories

Year-end timing can affect the population of unsettled trades. The working-capital reversal confirms timing effects in a different net measure, but cannot quantify the normal range of gross contracts. Comparing daily average and maximum notionals with the year-end figure would address that question directly.[1]

Year-end window effects can arise without manipulation: settlement holidays, large client transactions and the timing of security delivery all affect a snapshot. The analytical response is not to dismiss the figure but to ask whether the controls and capital resources are designed for the peak, not merely the average.[1]

Capital intensity is not visible from notional alone

Regulatory exposure measures may recognise netting, collateral and replacement cost rather than gross notional. The £50.2 billion therefore cannot be divided by equity to produce a meaningful leverage ratio without the applicable prudential treatment. Different products and entities within the group can also face different capital regimes.[1]

The useful risk comparison would pair a defined replacement-cost exposure with the resources available to absorb it. The public gross notional alone cannot supply either the probability or size of loss. Counterparty limits, collateral and settlement ageing are consequently more informative than the notional-to-profit multiple.[1]

Fair value is a stock, not an annual flow

The £177 million fair-value gain and £176 million fair-value loss figures relate to outstanding transactions at year-end. Calling them annual flows would misstate the reporting period and accounting meaning. Their near equality is consistent with matching, but subtracting them does not calculate annual profit from the activity.[1]

A trade can have a large contractual principal and a much smaller current replacement value. A matched purchase and sale can also create offsetting values while leaving a timing or counterparty obligation. That is why the note’s notional, fair-value and recognised-balance columns need to be read together. The systems still have to confirm and settle the contractual legs accurately.[1]

For customers and counterparties, the relevant assurance is not that purchases and sales look equal in aggregate, but that each matched pair is identified, confirmed and settled under enforceable agreements. A group-level balance can hide unmatched exceptions inside two similar totals. Reconciliation ageing and exception counts would therefore provide a more direct control measure than the gross year-end notional alone.[1]

That transparency would also improve peer comparisons. Interdealer brokers can report similar revenue while carrying very different gross settlement footprints depending on whether they act as agent, matched principal or risk principal. Business-model labels are therefore insufficient without balance-sheet and settlement data.[1]

For counterparties, the reporting-date reconciliation offers a useful boundary: £50.2 billion describes the purchase-contract notional, £177 million the fair-value gain on outstanding matched-principal transactions, and £38 million the recognised asset balance. Selecting one number and labelling it the whole exposure would erase the distinctions in TP ICAP’s own disclosure.[1]

The established finding is growth in the outstanding settlement footprint, not evidence of an equivalent increase in loss risk. Purchases rose 84.9% and sales 85.1%. The annual report supplies the accounting mechanism and matching explanation; it does not supply enough detail to attribute the increase to a particular cause or convert it into a forecast of losses.[1]

Report an error or submit an update