The Calm Window Problem: How Recent Volatility Understates What Can Happen

[Shadow Trader]

Many risk tools estimate volatility from a recent window, perhaps the last twenty or sixty days. When that window is quiet, the result is a small number, and position sizing rules built on it allow larger exposure. The system therefore recommends the most risk at the time when the sample may be least informative about rare events.

The flaw is that a short sample rarely contains rare events. A stretch without a shock says little about the probability of one, yet the calculation treats the absence as evidence of safety. Large dislocations then arrive against positions that were scaled up on the strength of calm.

Volatility targeting and ATR-based stops share this weakness, since both shrink in calm periods. Stops placed at a multiple of recent range end up tight, and size grows to compensate, so a single oversized move does disproportionate harm.

Some practitioners respond by blending a longer lookback with the short one, setting a floor under the volatility input, or running stress scenarios using historical shocks. None of these remove tail risk, but each reduces the tendency of calm conditions to inflate exposure.