Let's say we have a set of investments, a target allocation to those investments, and commit to rebalancing at some particular frequency. In the worst-case scenario, if one of the classes catastrophically goes to zero, wouldn't that (if rebalanced mechanically) cause our entire portfolio to go to zero? Consider in the limit where class A drops, so we shift more funds in, then drops again, etc.
This is obviously a systems-collapse type scenario. But are there theoretically situations where one class could go to zero while another is still viable? Is there a lesson here that in extreme circumstances we should monitor and possibly abandon one class instead of rebalancing?