# Trailing stop (long-only) based on annualized volatilty

I am wondering if there is a rule-of-thumb for setting a trailing stop (fixed percent) if you know the annualized volatity. I calculate volatility from daily close price using the simple returns, $$r_t = (P_t - P_{t-1})/P_{t-1}$$, then get the s.d. $$\sigma$$ of the returns $$r_t$$, and annualize via $$\sigma \sqrt{252}$$. So if I rebalance e.g. monthly (every 20 trading days), is $$\sigma \sqrt{20}$$ a resonable estimate of the stop loss?

Thinking about a 20-day volatility, however, it would represent the expected total variation of jumpyness over the 20-day time series, so the trailing stop would need to be a fraction(multiple) of the total. Hence, my gut feeling for the existence of a general rule-of-thumb. Would "2-sigma from the mean" be an appropriate assumption, where $$2\sigma \sqrt{20}$$ is used?

UPDATE

Based on looking at charts of SPY, MRNA, AYI, RBLX, AYI, etc, with a trailing stop indicator with fixed percent, it is looking like $$\sigma\sqrt{20}$$ is almost perfect.

• Yes, what you say is reasonable. Oct 15 at 12:31