How Do You Decide Whether a New Trading Technique Is Worth Testing?

New techniques can be interesting, but experimenting with too many ideas at once can make it difficult to determine what actually works. What minimum evidence or clear hypothesis would you want before committing time to testing a new technique?