A digital-asset business can be solvent on Friday and unable to make a required payment on Sunday. The assets exist, but they may be in the wrong currency, network, account or operating window. A daily cash forecast that aggregates everything into one closing balance can miss that problem entirely.
The forecasting task is to identify when each obligation becomes payable and when the corresponding asset becomes usable for that obligation. Continuous market activity raises the frequency of that question. It does not mean every funding route operates continuously, and it does not mean all banking rails close at the same time.
Start with the payment clock
A forecast should record the contractual deadline and required settlement asset for each material outflow. A bank transfer in dollars, a stablecoin margin call and native-network transaction fees are different obligations. An excess of one does not satisfy the other until a conversion and transfer route has completed.
Known outflows include scheduled redemptions, loan repayments, supplier payments and withdrawals already accepted. Contingent outflows include margin calls, client withdrawal demand and liquidity commitments triggered by market moves. The latter need scenarios and capacity assumptions rather than a false point estimate.
Our article on programmable treasury payment exceptions explains why an instruction is only the beginning of an operating process. The cash forecast must reflect the same distinction between a requested action, an accepted transaction and confirmed usable proceeds.
Banking hours are a route-specific input
The Federal Reserve's published Fedwire Funds schedule identifies funds-transfer business days and different message cutoffs. Treasury should use the current schedule together with its bank's own customer deadlines. A bank may require instructions earlier than the underlying service's final cutoff.
The FedNow operating-hours guidance describes continuous payment processing, including weekends and holidays, while distinguishing support activities available during standard business hours. The existence of an instant rail does not establish that a particular bank account, issuer or payment type can use it on the required terms.
Record availability for each actual route. Include account eligibility, transfer limits, currency, operating hours, maintenance, provider approvals and destination crediting. A generic field labelled bank open leaves too many assumptions hidden. The issuer's minting window and the receiving custodian's crediting process may be the binding constraints.
A hypothetical weekend gap
Consider a hypothetical trading firm with $10 million of unencumbered assets on Friday afternoon: $6 million in a bank account and $4 million of a stablecoin on network A. It expects a $3 million stablecoin obligation on network B on Sunday. Its approved bridge or conversion route can reliably provide only $1 million before that deadline under the assumed limits.
The aggregate forecast shows ample liquidity. The route-specific forecast shows a $2 million shortfall in the asset and location required. The bank balance cannot fill it unless a supported weekend funding path reaches network B in time. Treating bank dollars and onchain dollars as one balance would conceal the gap.
Now assume another $2 million of stablecoins is expected from a client on Saturday. That inflow can close the forecast only if it arrives, is cleared for use and can reach network B. A scenario delaying the client payment should still expose the original shortfall. Expected cash and committed contingency capacity are not interchangeable.
This example illustrates timing and location risk, not a recommended funding amount. The right buffer depends on the business's obligations, confidence in inflows and tolerance for missed payments. Those are institution-specific decisions supported by the forecast.
Use time buckets that can reveal the gap
A weekly view is useful for planning, but a continuous operation needs a shorter view around critical windows. Hourly buckets can expose a funding shortage that disappears by the end of the day. The forecast horizon should cover the longest period during which a required replenishment route may be unavailable.
Weekend and holiday windows need their own assumptions. A three-day bank holiday can coincide with uninterrupted client activity, while market volatility can raise collateral demand at the same time. Averaging those days into ordinary weekday flows understates the concentration of demand.
Time zones need explicit treatment. Store operational deadlines in a common reference and display local cutoffs with the relevant daylight-saving rule. A recurring instruction scheduled at the same local hour can shift relative to another jurisdiction's market. The forecast should change automatically with verified calendars rather than depend on a remembered one-hour adjustment.
Count availability after restrictions
Forecastable resources include free cash and other assets that can be converted through identified routes before the deadline. Exclude pledged balances, pending withdrawal locks and amounts reserved for already committed settlements. A wallet balance can be technically transferable but unavailable under policy or agreement.
The availability of an authorised signer also matters. If the weekend route requires two approvals and one role has no cover, it is not operational capacity. Our article on institutional wallet transaction policies explains the control side. Forecasting should ingest the resulting constraints rather than assume signatures are available on demand.
Funding facilities require equally precise treatment. An undrawn limit is useful only if drawdown conditions can be met in time, the lender can deliver the required asset and the facility remains available in the scenario. A facility whose documentation permits suspension in the relevant circumstances should not be modelled as unconditional cash.
Reconcile forecasts to usable receipts
Compare each forecast inflow with the time it became available for its intended use. Transaction inclusion, required finality and custodian crediting can produce different timestamps. Measuring only the first makes a route appear faster than the institution actually experiences.
Forecast errors should be classified by cause: incorrect amount, late counterparty, provider cutoff, failed approval or conversion capacity. That allows the model to improve. Repeatedly adding an unexplained buffer hides the route that is failing and may tie up cash without fixing the underlying constraint.
Contingency actions should have a lead time. If liquidity on network B must be positioned before a Friday deadline, an alert on Sunday is only an incident notification. A useful trigger identifies the approaching funding gap while treasury still has an approved path to address it.
The forecast should identify the next constrained payment
Continuous operations do not require every asset to sit idle in every location. They require a credible connection between obligations and usable resources across the hours that matter. The tradeoff is between pre-positioning more liquidity and accepting reliance on conversion, counterparties and operating windows.
A good forecast makes that tradeoff visible before the window closes. It can name the payment, the asset needed, the expected source and the latest moment at which a fallback must start. That is more useful than a large consolidated cash number that cannot fund the next obligation.