<p><a href=“Bloomberg Politics - Bloomberg”>Bloomberg Politics - Bloomberg;
<p>"On a freezing day in March 2007, Nassim Taleb walked into a conference room at Morgan Stanley’s Manhattan offices on 47th Street and Broadway to address a group of the firm’s risk managers. His message: Your models don’t work. </p>
<p>Using a whiteboard to scribble out his calculations, Taleb, now 48, began one of his rants, this time against stress tests – Wall Street lingo for examining how a market rout will play out. Stress tests are inherently risky because they ignore rare but potentially devastating events, Taleb said. </p>
<p>``Past shortfall doesn’t predict future shortfall,‘’ the options trader turned best-selling author recalls telling the assembled group of about 40. The risk managers, part of a tribe of mathematical model makers known in the finance world as quants, stared back at him blankly, and a debate ensued, according to people who were there.
Only six months later, Morgan Stanley experienced its own rout. The world's second-biggest mergers adviser announced in December that it had written down its subprime-related holdings by $9.4 billion after the firm's traders misjudged how fast and far prices of the debt would fall. Their risk management had failed."
"Past shortfall doesn't predict future shortfall"
"In one pilot experiment, they posed the following question to participants: ``You're on vacation in a foreign country and are considering flying the national airline to see a special island you have always wondered about. Safety statistics in this country show that if you flew this airline once a year, there would be one crash every 1,000 years on average. If you don't take the trip, it is extremely unlikely you'll revisit this part of the world again. Would you take the flight?''
Everyone answered yes, assuming that one crash every 1,000 years was a minimal risk.
Finding the Extremes
Another group was given the same problem except they were told that an average of 1 in 1,000 flights on this airline crashes. Although it's the same risk mathematically, 30 percent refused to fly when presented with this wording.
``This one-in-every-X-years framing is something you hear concerning market crashes in financial reports on TV,'' says Goldstein, 38, who holds a Ph.D. in psychology.
Extremes are more likely in finance than in the real world, Taleb says. At a conference for risk managers in London last June, he used the following illustration: ``Say I sample from the world population and find two people cumulatively 14 feet tall. What's the most likely allocation for Gaussian? One and 13? No, it's seven and seven.'' </p>
<p>In wealth, it’s the opposite. <code>If we sample from the world population and get two people whose net worth totals 14 million pounds, what’s the most likely combination?‘’ he asked.</code>Seven and seven? No, it’s 5,000 pounds and 14 million pounds minus 5,000 pounds.‘’</p>