Yeah-- what Puffin said.
Change when the calculus surrounding "least-worst" seems to change. It's really that simple. Well, and at the same time, that complex, I suppose, because of all of the various factors (some of them idiosyncratic) that go into the calculation, there.
We found it was wise to keep our eyes open to signals that things were slipping into a critically bad zone-- or, conversely, to pay sufficient attention to options previously dismissed, so that we were aware of shifts/changes in THOSE things, too.
That way you can always run a thumbnail calculation and have a sense of when it is wise to do a more thorough analysis.