Person A has $1.5mil in assets (let's say, stocks/bonds). He is ~5 years from retirement. He has decided that he wants to enter retirement with no less $1mil, say. So, he can afford to lose 1/3 of total value. He chooses some asset allocation X, which he decides is an appropriate balance of risk and reward for his situation; under this allocation, he believes it is sufficiently unlikely to dip below $1mil. He recognizes it is still possible, but he adjudges the potential gain to be worth the risk. (You can imagine this to be whatever stock/bond mix you'd like - 50/50, 40/60, 30/70 - whatever you think is best. But let's suppose it has at least

*some*amount of stock.)

Person B is exactly the same, except he has $1.1mil. He goes through the same thought process and chooses some asset allocation Y.

I'm going to assume that Y is more conservative than X. Both Person A and B can afford to lose some money, but B can only afford to lose about 9% of this worth, whereas A can afford to lose 33%. Person B needs to play it safer. If you accept that premise, then allocation Y has a lower stock percentage than allocation X. If you can't accept this premise, don't bother reading on.

Now suppose that Person A

**becomes**Person B due to a large drop in both stocks and bonds, in the course of just a month or two. If we accept the premise that Person B's optimal allocation had less stocks than Person A, then this person's new optimal allocation is Y, not X. His original allocation of X is no longer appropriate - it takes on too much risk. If he needs to sell some stock to get there, so be it.

That's it.