What is stress test?

A stress test pushes a system, such as a bank, portfolio or supply chain, through severe but plausible conditions to check whether it survives. Instead of asking what usually happens, it asks what happens in a bad scenario. It matters because it finds weak points before a real crisis does.

How it works

A stress test starts by defining a scenario: for example, sales fall by 30%, interest rates jump by three points, or a key supplier stops delivering. The scenario is then applied to the model, and the outcomes are measured against thresholds, such as minimum cash, capital ratios or service levels. The results show how big the shock must be to cause failure, and which part gives way first.

There are two common approaches. A scenario test uses one defined story, while a reverse stress test starts from failure and works backward to find what combination of events would cause it. The trade-off is realism: scenarios that are too mild teach nothing, and ones that are impossible are ignored.

Example

Say a company has 200,000 in cash, monthly costs of 90,000 and monthly revenue of 100,000. In normal times, cash grows by 10,000 per month. The stress scenario assumes revenue drops 30% to 70,000 for six months. Monthly cash change becomes 70,000 − 90,000 = −20,000, so after six months cash is 200,000 − 120,000 = 80,000. It survives, but a 50% drop gives −40,000 per month and cash hits zero after five months. These are illustrative numbers.

How to model it in Machinations

Build the normal model with a cash Pool, a revenue Source and a cost Drain. Add Levers for the shock size and its duration, applied to the revenue rate through a state connection. Add an End condition for when cash falls below a minimum. Play at several Lever settings and note the shock size at which the End condition triggers, then Play many times at once to see how likely it is.