Set out risk measures suited to return distributions that are not normal
The idea
Many alternative strategies produce returns that are not normally distributed: option-like payoffs, credit strategies and short-volatility positions generate negative skew and fat tails, where many small gains are punctuated by rare large losses. Standard deviation weights upside and downside equally and therefore misdescribes such a distribution, and the Sharpe ratio inherits the problem. Downside deviation, the Sortino ratio, maximum drawdown and value at risk each address part of it by focusing on the loss side, though none is a complete substitute for understanding the strategy's payoff structure.
Formulas
Sortino ratio
Sortino = (Rp − target) ÷ downside deviation
- Rp
- the portfolio's return
- target
- the minimum acceptable return
- downside deviation
- the standard deviation of the returns below that target only
For returns that are not symmetrical, where the Sharpe ratio penalises upside volatility as though it were risk. Alternative investments are the usual case: a strategy that is quiet most months and loses badly in a few looks better on Sharpe than it deserves.
A practice question
Why is standard deviation an incomplete risk measure for a negatively skewed strategy?
- Because it cannot be computed from monthly dataWhy not: Monthly data is the usual input for it.
- Because it treats upside and downside variation identically — correct
- Because it requires at least ten years of historyWhy not: No minimum applies, though a short history is a separate weakness.
An investor does not regard an unexpected gain as risk. A symmetric measure applied to an asymmetric distribution misses what matters.
A harder one
What does the Sortino ratio use in place of standard deviation?
- The maximum drawdownWhy not: A distinct loss measure, used in its own right rather than inside the Sortino ratio.
- The beta of the strategyWhy not: A market sensitivity measure, which belongs to the Treynor ratio.
- Downside deviation, computed only from returns below a target — correct
By penalising only shortfalls against a threshold, it describes an asymmetric payoff more faithfully than the Sharpe ratio does.
The hardest tier
A strategy has produced steady small monthly gains for five years with very low measured volatility. Why should this alone not be reassuring?
- Because that record precedes the rare large loss — correct
- Because five years is the maximum useful historyWhy not: Longer histories are better, and no length is a maximum.
- Because low volatility always indicates poor returnsWhy not: Low volatility with steady gains looks excellent, which is exactly what makes the pattern so persuasive.
A short-volatility or credit payoff produces exactly this history until the loss arrives. The record is consistent both with genuine skill and with selling insurance that has not yet paid a claim. Only understanding the payoff structure distinguishes them, which is why the strategy must be examined and not just its statistics.
In the app: 12 practice questions on this outcome, a narrated walkthrough with diagrams, spaced review that brings it back before you forget it.
Take the free diagnostic →