Liquidity Stress Test

Liquidity Stress Test

When to use

Use whenever cash position needs to be tested against a genuine downside, not just tracked under the current plan — before a board or lender conversation that will ask "what happens if revenue drops," or as a standing discipline for any business carrying real liquidity risk. Especially use to replace a stress test that stops at identifying the risk without sizing what specific actions would actually close the gap.

What it does

Applies a stated revenue-drop shock to the monthly net burn via a bundled calculator, projects the cash balance across the stress horizon, finds the trough month and the shortfall if cash would go negative, and ranks a list of candidate mitigations by exactly how much cash or runway each one restores — turning "we should probably cut costs" into a ranked, sized list of specific actions.

Method

1. State the base monthly net burn under the current plan, before any shock is applied.
2. Define the shock explicitly: a specific revenue drop percentage, not a vague "if things get worse." A defensible shock size usually comes from a real historical precedent or a specific plausible risk (a major customer churning, a macro downturn comparable to a past cycle).
3. Account for collection delay as part of the shock, not just the revenue drop itself — a revenue shortfall often arrives with a lag in when cash actually stops coming in, and modeling only the revenue drop without the collection timing understates the near-term cash impact.
4. Run the bundled calculator (scripts/stress_test.py) to get the month-by-month stressed cash balance, the trough month, and the shortfall amount if the balance would go negative.
5. Build the mitigation list before the stress test, not after seeing a scary result. Real candidate actions: a credit line that could be drawn, specific opex cuts with their monthly savings, planned hires that could be delayed — each with an honest estimate of its cash or monthly-savings impact.
6. Read the ranked mitigation output as the actual output of the exercise, not the trough number alone. The trough tells you there's a problem; the ranked list tells you what to actually do about it and how much each option is worth.
7. Pre-commit to which mitigations trigger at which point, rather than deciding under pressure once the stress scenario starts to look real — name the specific cash level or month that would trigger drawing the credit line versus cutting discretionary spend.

Inputs

• Current cash and base monthly net burn
• The stress shock: revenue drop percentage and collection delay
• Stress horizon in months
• Candidate mitigations with their cash impact (one-time) or monthly savings (ongoing)
• Config saved as JSON matching the format documented at the top of scripts/stress_test.py

Output format

Month-by-month stressed cash balance; trough cash and month; explicit shortfall flag if the balance goes negative; ranked mitigation list showing exactly how much cash or runway each option restores.

Example

A 30% revenue-drop shock, with a one-month collection delay factored in, shows the business would run out of cash by month 12, a $105,000 shortfall at the trough. Rather than stopping there, the ranked mitigation list shows drawing an available credit line restores $1M, dwarfing what an opex cut alone would provide in the same window — the specific, sized comparison the trough number alone never would have surfaced.

Common pitfalls

• Running a stress test that only identifies risk without ranking what mitigations would actually close the resulting gap.
• Applying the revenue shock without accounting for the collection delay that typically comes with it, understating the near-term impact.
• Deciding on mitigations only after the crisis starts, instead of pre-committing to specific triggers ahead of time.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Natan-Mohart
• Source: Natan-Mohart/24-finance-skills-for-claude
• License: MIT

Install and usage instructions live in the source repository linked above.