How Do You Determine Whether a Trading Strategy Benefits From Separating Entry and Exit Testing?

Changing entries and exits simultaneously can make it difficult to understand which adjustment caused a performance change. How would you organise testing to isolate the contribution of each component? What baseline should be established before experimenting with alternative exit methods?