As others have said, the idea is that MI adapts to changes (might not always work), but a sudden change can lead to a period of time when MI methods stop working. Jim (mungofitch) has explained this before for, say a momentum screen at a V type market bottom. Typically the best performers at such a bottom is the junk that dropped most to the bottom but your momentum screen immediately after the bottom will be looking for what dropped least. This will essentially 'break' your MI screen until the bottom has mostly disappeared from your momentum look back period. Similarly, if there was a sudden economic impact from PRC, it might affect the performance of some MI screens for a period but the MI theory would suggest that the impact should be short-lived and should pass just like any other economic shock.

StevnFool