What a bad month costs
The loss that one month in twenty was worse than.
In the demo portfolio
-9.6%
How it is computed here
Overlapping 21-session windows of this portfolio's own history, rather than the daily figure multiplied by the square root of 21.
Worked example
- Every window of 21 sessions, about a month, overlapping: 1,540 of them.
- One month in twenty lost at least 9.6%; the average of those worst months is -13.7%.
Demo portfolio: four invented companies, generated prices, measured by the real engine. Past returns do not predict future ones.
Where it misleads
That square-root rule assumes days are independent and normally distributed. Bad days come in runs — which is exactly what turns a 3% fall into a 30% one — so scaling up the daily figure understates the month precisely when it matters.