Sortino ratio
How do you judge a return against only the downside moves, ignoring the upside volatility that a position can tolerate?
In plain words
The Sortino ratio divides a period's return by downside volatility, which counts only the moves below a target or below zero. Unlike ordinary volatility measures, it treats upward swings as harmless, so a large positive move does not raise the penalty.
Why it matters
Volatility measures used for risk treat a sharp rally and a sharp drop the same way, even though only losses damage a long position. Measuring risk per unit of downside makes a lopsided week look different from a week with equal swings in both directions, and it can be estimated from only a few observations of falling prices.
In a real week: ETH, week of 5 March 2018
In the week of 5 March 2018, ETH returned -16.82% against annualised realised volatility of 121.3%. Its worst day was -7.8% on 2018-03-07, the same date as the largest one-hour move of -6.27% at 16:00 UTC, while its best day was +5.92% on 2018-03-11. The low of 640.0 on 2018-03-09 at 06:00 UTC followed the high of 871.75 on 2018-03-05 at 12:00 UTC, and the close sat in the middle third of the week range. With only 3 weeks of history, any Sortino estimate from this data rested on few falling-price observations.
A common mistake
Assuming the Sortino ratio rewards upside. It only reduces the penalty for rising markets; a week with a very negative return still produces a very negative ratio.