build-cash-forecast-and-liquidity-plan

Skill: build-cash-forecast-and-liquidity-plan

Invoked by: cash-and-risk-operations-specialist (primary, for the position + forecast build) and treasury-strategy-lead (for the buffer/facility policy the forecast grounds).

When to invoke: "build our 13-week cash forecast"; "what's today's cash position?"; "how much minimum cash / liquidity buffer should we hold?"; "direct or indirect forecasting?"; "committed vs uncommitted facilities / how much revolver headroom?"; any "how much cash and where" question.

Output: the daily cash position + the rolling 13-week (direct) forecast + a variance loop + the stress-tested minimum-cash/buffer target + the committed vs uncommitted facility mix & revolver headroom + the 1-2 conditions that resize it.

Procedure

1. Position cash first — from where the money actually is. Build today's cash position per currency and entity: opening bank balances + confirmed receipts − confirmed disbursements → closing / available liquidity (net of holds, minimums, and un-cleared items). This is the anchor; a forecast that doesn't start from the real position is untethered.
2. Pick the forecast method by horizon. Direct (receipts-and-disbursements) — actual expected cash in/out, driver-based — for the operating horizon (the 13-week and shorter); it's what treasury runs day-to-day. Indirect (net income + working-capital changes + non-cash add-backs) — for the longer, statement-linked view where you forecast from the P&L/balance sheet. Don't use indirect for the 13-week (too coarse) or direct for the annual (too granular to sustain).
3. Build the 13-week from drivers, not a straight line. Lay out 13 weekly columns; populate receipts (collections driven by DSO / the AR aging, plus known one-offs) and disbursements (AP payment-run timing driven by DPO, payroll, tax, debt service, capex, dividends). Each line is a driver, not a plug — so a variance is diagnosable.
4. Attach a variance loop — every week. Forecast → actuals → variance by line → re-forecast. Tune the drivers from the miss (collections slower than DSO assumed? a payment run slipped a week?). A 13-week without a back-test is a wish, not a forecast.
5. Size the minimum-cash / buffer against a stress, not the average. Traverse the buffer-sizing branch in ../../knowledge/treasury-management-decision-tree.md: take the forecast's trough (the worst intra-period low), then stress it — a receipts shock (a big customer slips), a facility pulled, a covenant tightening, a seasonality low — and set the buffer so the trough-under-stress stays above zero (and above any covenant/minimum-operating-cash floor).
6. Choose the facility mix — committed over cheap. Cover the gap between the buffer and on-hand cash with committed facilities (a contractually available revolver worth the commitment fee) rather than uncommitted lines (withdrawable exactly when you need them). Keep revolver headroom you can actually draw when stressed, and check the draw conditions/covenants don't lock it at the wrong moment.
7. State the resize conditions — the 1-2 facts that change the buffer or facility plan (e.g., "if a top-3 customer's terms extend, the receipts trough deepens and the buffer must rise"; "if the revolver's covenant tightens, its headroom is not really available and the buffer must absorb it").

Worked example

User: "We're a seasonal consumer-goods company with a Q4 build. Build our 13-week and tell us how much cash to hold and whether our $50M revolver is enough."

• Position: consolidate the operating accounts to today's available balance per entity — say $18M on hand.
• Method: direct / R&D for the 13-week — receipts from the AR aging (DSO ~52 days), disbursements from the AP run calendar + the pre-season inventory build (a big DPO-timed outflow), payroll, and the interest payment in week 9.
• 13-week: the inventory build drives a trough in weeks 6-8 where cumulative cash dips to $6M before Q4 collections arrive.
• Variance loop: back-test weekly — if collections run 4 days slower than the DSO assumption, the trough deepens ~$3M; tune the receipts driver.
• Buffer (stressed): trough $6M, stress it for a slow-collections + a supplier-prepay demand → the buffer target is ~$12M minimum operating cash. On-hand at the trough ($6M) is below it → draw is needed.
• Facility: the $50M committed revolver covers the gap with headroom; confirm the leverage covenant won't be breached by the seasonal inventory swing (which would block the draw exactly when needed).
• Resize condition: if the pre-season build grows 30% or the revolver covenant tightens, the trough deepens and the committed headroom must be re-tested.

Guardrails

• Position cash first; a forecast that doesn't start from the real per-currency/entity position is untethered.
• Direct for the 13-week, indirect for the horizon — don't mismatch method to horizon.
• Every forecast carries a variance-to-actual loop; drivers are tuned from the miss, not left as plugs.
• Size the buffer on the stressed trough, not the average month; include any covenant / minimum-operating-cash floor.
• Committed beats uncommitted — an uncommitted line can be pulled exactly when you need it; verify draw conditions/covenants don't lock a committed line at the wrong moment.
• Sizing the buffer is a policy call (the treasury-strategy-lead owns it); the build/variance loop is execution (the operations specialist) — keep the seam clean.
• This is not the FP&A budget / P&L plan — that's finance; treasury forecasts the cash, not the earnings.
• Volatile specifics (facility terms, covenant definitions, bank-portal balance feeds) carry a retrieval date and are re-verified before a board/bank commitment. See ../../knowledge/treasury-management-patterns-2026.md.