"No, what I meant was that excluding regions which don't fit is acceptable. Naturally it resets the clock."

Exactly my point. And one has to be careful to treat ALL the regions that "didn't fit" the same way, not just the one gumming up the latest correlations the most.