Matt Levine, Columnist

Robot Retail Investors

Automatic dip-buying, a Satoshi candidate, dispersion and counting widgets.

People do not like to lose money on stocks. Sometimes a stock goes down, and people who own it sell it to avoid losing more money, so it goes down more. This behavior is to some extent predictable, so there are circumstances in which you might think “if this stock goes below $50, it will probably keep going down to $45.” You might think that as a pure matter of investor psychology: “If the stock goes below $50, a lot of people who bought it at $50 will have losses and panic and hit the ‘sell’ button, driving it down more.”

You might have a slightly different theory. Because people do not like to lose money on stocks, and are not monitoring their stocks constantly, they might use stop-loss orders, where they tell their broker “if this stock goes below $50, sell it immediately.” Intuitively, people who do this are more likely to use round numbers like $50 than, like, “if this stock goes below $51.37, sell it,” so there is some clustering of stop-loss orders; there are more of them at some numbers than at others. And so if you are trying to predict market behavior, you might think “if this stock goes below $50, a lot of people will automatically sell it, and it will go down to $45.” None of this is investing or technical-analysis advice, but a certain amount of “stop hunting” probably goes on in financial markets: If you know or suspect that a lot of people have stop-loss orders to sell if a stock falls to $50, you might try to short the stock until it hits $50, expecting to trigger the stops and buy it back at $45.