Methodology · reviewed
Ups and downs: volatility, biggest fall, Sharpe and Sortino
These measures describe how much a fund’s price has jumped around in the past. They need a long price record, so we show them only for funds whose closes we have recorded over the whole period.
The formula
volatility = st.dev(weekly returns) × √52 · Sharpe = mean ÷ st.dev × √52 · max drawdown = lowest (price ÷ previous peak − 1)
Sortino replaces the standard deviation with the downside deviation: √(mean of min(weekly return, 0)²), averaged over every week — weeks with a gain count as zero shortfall (Sortino & Price). The target and the risk-free rate are taken as zero.
Worked example
Weekly returns with a standard deviation of 2% give an annual volatility of 2% × √52 ≈ 14%.
How it is done
Windows are chosen by calendar date (the last 1, 3 or 5 years), the price record must reach back to within about six weeks of the window’s start, and a window is only computed for a fund that had launched by then — a merged or relisted line’s older prices are not its own.
Every record is first reduced to one close a week (the last close of each week), so older weekly history and newer daily closes are measured the same way. Figures are annualised with 52 weeks a year.
Weekly bars are dated to the Friday they close, not the day they are filed under, so no week’s close is lost or counted twice when the clocks change.
Every figure is in pounds: a fund quoted in dollars or euros is converted close by close at the reference exchange rate for that day, so the figures include the currency moves a UK holder bears.
A price that jumps more than threefold in one step is treated as a change of unit (pence read as pounds, say) and only the record after it is used.
A single close more than 15% away from both neighbours, while the neighbours agree, is treated as a data error and left out. Leveraged products are exempt, because their prices genuinely do this.
Limits to know about
- Past ups and downs do not tell you what future ones will be.
- Sharpe here uses a zero risk-free rate, so it is higher than a version measured against cash (SONIA) would be.
- Measures from price records leave out income paid out by distributing funds.
- Weekly closes miss moves that reverse within a week, so the biggest fall can read slightly smaller than on daily data.
Reviewed . If a figure on the site does not follow this method, tell us. This is information, not advice.