Deliberately pushing a system beyond normal operating conditions — extreme volatility, order surges, connectivity failures — to find its breaking point before the market finds it first. It overlaps with edge-case testing, but leans more on volume and load than on specific unusual scenarios.