What Is Your Method for Testing a Strategy Across Different Market Regimes?

A setup may perform well in a trend but poorly in a range. How do you separate results by market regime, and what evidence would show that a strategy is robust rather than suited to one period?