Red-Blooded Risk: The Secret History of Wall Street

Black Holes

The third tool is entirely quantitative, searching for holes in the profit and loss distribution. Stress testing and scenario analysis concentrate on dramatic external events that could harm the firm. But most large losses occur instead from unexpected combinations of events that are not individually unlikely. There are a number of mathematical techniques to search for these. I won't go into technical details, but one simple approach is to look at the price movements in all past days in your database, to find the ones that would cause large losses in your current portfolio if they had been scaled up to the movement size of the average day. For example, you might find a quiet day in the past when most market prices moved only about a tenth as much as their average moves. Take the price moves from that day—up and down—and multiply each one by 10. Then apply those scaled price moves to the firm's current positions. Say that gives you a loss of six times VaR. This could be a hole in the profit and loss distribution. You don't know that the scaled day is a plausible event, but you know that the relative price movements that occurred on the day are possible, and you know the total size of price moves on the day is not unusual. That's enough suspicion to investigate, with a presumption of guilt. Unless you can find a strong reason to reject the scaled day as implausible, you list it as a hole. Risk management is not guessing what will happen; it's preparing for anything that ...