Quick answer: intraday liquidity management asks whether a bank has enough usable money and collateral at the right moments inside the business day to settle payments when they are expected. The daily closing balance is not enough. Banks forecast incoming and outgoing payment flows, monitor real-time balances, identify time-critical obligations, manage collateral and intraday credit, and use payment queues or liquidity-saving mechanisms where the infrastructure permits them. The core mathematical problem is time-ordered feasibility: can every required payment be released before its deadline without the bank exhausting available intraday resources?

A bank can have enough cash by 5 p.m. and still fail a payment due at 10:07 a.m.

Why this belongs in mathematics

Intraday liquidity combines time series, queueing, optimisation, graph flows, collateral constraints, stochastic forecasts and stress testing. It is a sharper version of the general liquidity lesson: totals are not enough when sequence matters.

The Basel Committee defines intraday liquidity as funds accessible during the business day, usually so banks can make payments in real time. Its monitoring framework exists because poor intraday management can leave a bank unable to meet payment obligations on time and can transmit liquidity dislocations across institutions and systems. See Basel SRP50 and Monitoring tools for intraday liquidity management.

1. Start with a running balance, not an end-of-day total

Let B(t) be the bank’s usable settlement balance at time t. A simplified intraday state equation is:

B(t+1) = B(t) + inflows(t) + intraday credit(t) − outflows(t) − collateral/funding constraints(t).

If B(t) would become negative beyond the amount of permitted intraday credit, some outgoing payments cannot be released immediately. The bank must obtain liquidity, receive incoming payments, pledge more collateral, reorder eligible payments or use a liquidity-saving mechanism.

This is a recurrence relation with deadlines. Two days with identical total inflows and outflows can require very different liquidity if the timing order changes.

2. A tiny timing example

Suppose a bank begins with S$20 million of settlement liquidity.

Time	Flow	Running balance
09:00	Opening liquidity	S$20m
09:30	Outgoing payment −S$30m	−S$10m before credit/queue action
11:00	Incoming payment +S$25m	S$15m if earlier payment was funded
14:00	Outgoing payment −S$10m	S$5m

Across the full day, inflows and opening resources are enough. At 09:30 they are not. The bank needs at least S$10 million of extra intraday liquidity, a payment delay that is contractually and systemically acceptable, or an offsetting incoming flow that arrives sooner.

3. Forecasting payment flows

An intraday forecast can use scheduled obligations, historical payment patterns, customer instructions, securities settlements, FX settlements, CCP margin calls, correspondent-bank activity and expected incoming payments.

A useful forecast is not only a daily number. It can be a vector across time buckets:

F = [net flow 09:00–10:00, net flow 10:00–11:00, …].

The model can then estimate the minimum cumulative balance and the largest positive and negative positions during the day. Basel’s monitoring tools explicitly track daily maximum usage, available intraday liquidity, payment timing and situations where a bank is a direct participant or correspondent for others.

4. Forecast error matters more near a deadline

Suppose the model forecasts an incoming S$50 million payment at 10:00, but it arrives at 13:00. The daily amount forecast was correct. The timing forecast was wrong by three hours—and that can be enough to create a liquidity shortfall.

A good validation system should therefore measure more than total absolute error. It should track:

error in payment amount;
error in payment time;
error near critical settlement deadlines;
forecast bias by counterparty or payment type;
how forecast errors change peak liquidity usage;
whether supposedly reliable incoming flows remain reliable in stress.

5. RTGS solves settlement credit risk by creating a liquidity problem

Real-time gross settlement (RTGS) systems settle payments individually and with finality rather than waiting to net many obligations at the end of a cycle. This sharply reduces settlement credit risk, but participants need more liquidity because offsetting payments may arrive at different times.

The Bank of England explains that CHAPS payments settle individually on a gross basis through RTGS and that this removes settlement risk at the cost of greater liquidity need. See Payment and settlement at the Bank of England.

This is a general systems trade-off: one safety mechanism can move pressure into a different resource constraint.

6. Queues turn payment release into a scheduling problem

If every payment were released the instant it arrived, a participant might consume large intraday liquidity even when an offsetting payment is about to arrive. Some systems therefore permit payments to enter queues subject to priorities, deadlines and liquidity rules.

The optimisation problem is not simply “delay everything.” Delaying payments can create customer harm, market uncertainty or downstream liquidity stress. The system must trade off:

liquidity saved;
settlement delay;
payment priority;
systemic importance;
deadline risk;
fairness between participants.

BIS cross-country research finds that banks actively coordinate payment timing and adapt their behaviour to the incentives and queue features of payment systems. See Intraday liquidity around the world.

7. Liquidity-saving mechanisms find offsetting payment sets

Consider three queued payments:

Bank A owes Bank B S$10m.
Bank B owes Bank C S$9m.
Bank C owes Bank A S$8m.

Gross settlement would move S$27 million across three payments. A liquidity-saving mechanism can identify that the circular obligations offset heavily and release a compatible set with much less net liquidity committed at the same moment.

The Bank of England’s CHAPS Liquidity Saving Mechanism temporarily queues eligible payments and periodically matches groups of broadly offsetting payments for simultaneous settlement. See the CHAPS LSM User Guide.

8. The optimisation itself can become difficult

With a small queue, searching for offsetting payment groups is easy. With thousands of payments, participant priorities, deadlines and available liquidity, the number of possible subsets becomes enormous.

Recent BIS research has explored auction-based liquidity-saving mechanisms that let participants reveal preferences through bids and side-payments rather than relying only on fixed queue rules. Simulations reported liquidity savings relative to conventional RTGS processing while highlighting the trade-off between liquidity efficiency and settlement delay. See Auction-based liquidity saving mechanisms and Project Titus.

9. Daylight overdrafts: central-bank liquidity with credit risk attached

In the United States, eligible institutions can use Federal Reserve intraday credit, commonly called daylight overdrafts, when outgoing activity temporarily exceeds the balance in their Reserve Bank account. This supports payment-system liquidity but transfers risk to the Reserve Banks if an institution were to fail before the overdraft is extinguished.

The Federal Reserve’s Payment System Risk policy therefore combines intraday credit with collateralisation incentives, net debit caps, monitoring and fees for specified uncollateralised usage. See Federal Reserve Payment System Risk policy documents.

This is a useful boundary: central-bank intraday credit can reduce participant liquidity risk while creating controlled central-bank credit exposure. Risk is transformed, not annihilated.

10. Collateral is part of the intraday algorithm

A bank’s available intraday liquidity can include balances, incoming payments, committed facilities and collateral that can be mobilised for central-bank or secured credit. The quantity of securities on the balance sheet is not enough; the bank needs to know which assets are unencumbered, eligible, correctly positioned and operationally movable before the deadline.

This connects to How Banks Optimise Collateral for Repo and Margin. A theoretically valuable bond is not intraday liquidity if it cannot reach the facility in time.

11. Stress scenarios break incoming-payment assumptions

Normal-day forecasts often rely on recurring incoming payments. Under stress, counterparties may delay payments, customers may draw credit lines, FX settlement may require extra funding, CCPs may issue intraday margin calls, and collateral haircuts may rise.

A robust intraday stress test therefore asks what happens under combinations such as:

largest expected counterparty inflow is delayed;
one payment system becomes unavailable;
margin calls arrive earlier or larger than normal;
customer payment instructions spike;
collateral mobilisation is delayed;
the bank must support correspondent clients at the same time its own liquidity is under pressure.

The important stress variable is not only how much liquidity disappears. It is when it disappears relative to obligations.

12. Gridlock is a system property

If several banks each wait to receive before sending, the system can develop payment gridlock even when aggregate liquidity would be sufficient if payments were coordinated. Each participant’s individually cautious strategy can make everyone else’s incoming liquidity arrive later.

This is why intraday liquidity is partly a game-theory problem. Payment-system design changes incentives. Participants learn those incentives and adapt their timing behaviour. A rule intended to save liquidity can therefore have a different effect once banks respond strategically.

13. Creative-work lens: Apollo 13 and resources that must arrive before the clock runs out

Apollo 13 is not a banking source, but it offers a useful mental model for intraday liquidity: aggregate resources are meaningless if the right resource cannot be routed to the right subsystem before a hard deadline. Engineers cannot solve a present carbon-dioxide problem with oxygen that becomes available after the crew has already failed. A bank cannot settle a 10 a.m. payment with an inflow arriving at noon.

The creative work helps make deadline-constrained resource allocation intuitive. Banking evidence must still come from payment records, collateral, system rules and stress tests.

14. The algorithmic pipeline

Inventory settlement accounts and payment systems.
Forecast time-bucketed inflows and outflows.
Identify time-critical and priority payments.
Calculate opening balances and available intraday credit.
Map usable collateral and mobilisation times.
Estimate the running balance through the day.
Identify the projected peak negative position.
Optimise eligible payment timing without breaching deadlines.
Use LSM or queue functions where available.
Monitor real-time deviations from forecast.
Escalate when buffers, caps or deadlines approach.
Stress delayed inflows and operational outages.
Reconcile actual payment and collateral usage after close.
Update forecasting models using timing errors, not merely daily totals.

15. Failure modes

End-of-day blindness. Daily cash is sufficient but morning obligations fail.
Reliable-inflow assumption. Expected receipts are treated as guaranteed liquidity.
Queue gaming. Too many payments are delayed to conserve liquidity, pushing stress to others.
Collateral timing blindness. Eligible securities cannot be mobilised before the deadline.
Correspondent concentration. A participant underestimates client payment demands during stress.
System dependency. One RTGS or messaging outage removes a critical route.
Forecast aggregation. A good daily forecast hides poor hourly timing.
Strategic feedback. Other banks change their payment behaviour in response to the same stress.

16. Diagnostics and falsifiers

At what minute does the projected balance reach its minimum?
Which incoming payment contributes most to avoiding that minimum?
What if that inflow is two hours late?
How much liquidity is saved by the queue or LSM, and how much delay does it create?
Which payment cannot legally or operationally be delayed?
How much collateral is eligible but not pre-positioned?
Does the forecast systematically miss margin-call timing?
Can the bank extinguish intraday credit before the system closes under stress?

Suppose someone claims, “We finish every day with surplus cash, so intraday liquidity risk is low.” A falsifier is a day in which the bank repeatedly approaches or exceeds its available intraday resources before late incoming payments restore the balance. End-of-day surplus does not prove intraday feasibility.

17. Verification and update triggers

backtest forecast amount and timing by payment type;
reconcile peak intraday positions with payment-system records;
test queue priorities and LSM outcomes;
verify collateral mobilisation times through actual operational drills;
review daylight-credit or intraday-facility usage after large market events;
recalibrate after new payment rails or settlement-hour changes;
update stress assumptions after delayed or failed counterparties;
measure whether liquidity-saving behaviour increases downstream settlement delay.

Connections across the finance-and-banking algorithms lane

Payment systems: graphs, queues and real-time settlement — the infrastructure in which intraday liquidity is consumed.
Liquidity stress testing — the multi-day balance-sheet view complementing this within-day problem.
FX settlement — a major source of time-specific currency funding needs.
CCP margining — intraday margin calls can turn market movements into immediate liquidity demands.

Research anchors

The deeper lesson

Intraday liquidity is the mathematics of having enough at the moment of obligation. RTGS makes settlement safer by demanding liquidity sooner. Queues and LSMs trade timing for liquidity efficiency. Central-bank credit can bridge temporary gaps while creating controlled credit exposure. Forecasting converts expected flows into a path, but the path must return to real payment timestamps for correction. A strong model therefore knows both the amount and the clock.

Educational note: This article explains public payment-system and banking mathematics. It is not treasury advice, payment-operations instructions or institution-specific regulatory guidance.

Small Group Tutorials

How Banks Forecast Intraday Liquidity: Payment Queues, Liquidity-Saving Mechanisms, Daylight Overdrafts and Settlement Stress