Volatility

Annualised volatility is sampling-frequency invariant

← Back to Home

The Key Idea

Historical volatility is the annualised standard deviation of log returns. A fundamental property is that the result is the same regardless of the sampling interval — daily, weekly, or monthly — as long as the data covers the same period and you use the correct annualisation factor.

The annualisation factor scales one period's standard deviation up to a full year:

σ_annual = σ_period × √(periods per year)

Daily: σ_annual = σ_daily × √252
Weekly: σ_annual = σ_weekly × √52
Monthly: σ_annual = σ_monthly × √12

This works because variance (σ²) scales linearly with time, so n independent periods have variance n × σ², and standard deviation scales with √n.

Gold Futures (GC=F) — Oct 2008 to Jul 2012 Full-period annualised volatility

From daily returns
21.4%
σ_daily × √252
From weekly returns
20.31%
σ_weekly × √52
From monthly returns
20.15%
σ_monthly × √12

The three estimates differ only due to compounding and the finite sample size — they converge to the same value as the sample grows.

Rolling Annualised Volatility Daily 40-day vs Weekly 5-week vs Monthly 6-month windows

Weekly and monthly series are noisier because each window has fewer observations than the daily series, but all three track the same underlying volatility level.