risk
The Bell Curve Underrates Danger
The rule
The neat bell curve is the wrong map for markets. It treats violent crashes as nearly impossible, yet they keep arriving again and again.
Where it flips
Distrusting the bell curve can tip into believing no model is ever useful, which leaves you frozen or gambling. The fix is to keep using simple models for a rough guide, while always adding a fat-tail safety margin, rather than throwing measurement away.
Most standard risk tools assume returns spread out like heights in a class - bunched around an average, with wild extremes almost never happening. Mandelbrot's warning is that real markets have far fatter tails: big single-day drops that the bell curve says should happen once in a thousand years actually show up every few years. If your risk number quietly assumes the bell curve, it is telling you a comforting lie about how bad things can get.
A worked example
Rohan's fund sheet suggests a 5% single-day fall is a once-in-decades event. Yet the Nifty has had several such days in his own investing life. Sizing his loan-funded holding on that false comfort, he faces a margin call he was told could 'never' happen. [illustrative]
How to spot it
- ·A 'once in a thousand years' event that already happened twice
- ·Risk stated as a single neat percentage
- ·No room left for a crash bigger than the worst on record
Benoit Mandelbrot · The (Mis)behavior of Markets