Downside Deviation and How to Measure Return Shortfalls
In this chapter
Intermediate
A fund returns +1% this month and your goal was +2%. For a downside-deviation calculation, that positive month still fell short. Downside deviation measures the size of returns below a chosen target. Months that meet or exceed the target contribute zero. Large misses matter more because the gaps are squared.
You choose the question when you choose the target. A 0% target asks about losses in the currency of the return. A 2% monthly target asks about missed growth, including months that gained less than 2%.
Turn each return into a shortfall
Write \(r_t\) for the return in period \(t\), and \(\tau\) for the target over a period of the same length. The minimum acceptable return (MAR) is the target used to decide which observations fell short. \(d_t\) below is the signed shortfall, zero or negative. The magnitude of a shortfall is \(|d_t|\).
\[ d_t=\min(0,r_t-\tau) \]
For \(n>0\) observed periods, square the signed shortfalls, average across all \(n\) observations, and take the square root.
\[ \mathrm{DD}_{\tau}=\sqrt{\frac{1}{n}\sum_{t=1}^{n}d_t^2} \]
\(\mathrm{DD}_\tau\) is in the same units as the returns. The inner average includes zeros for periods that meet the target. A calculation divided by the number of misses measures the root mean square size conditional on a miss. That figure answers a different question and generally comes out higher when some periods meet the target. The \(n-1\) divisor often used in sample standard deviation belongs to that measure’s estimation around a sample mean. Here you measure against a specified target with divisor \(n\).
Worked case 1. Count only losses with a zero target
You have four hypothetical monthly nominal USD total returns with income reinvested, before fees and taxes, and no external flows. Use the last close of each month. The target is 0% per month.
| Month ended | Portfolio return | Signed shortfall (pp) | Squared shortfall (pp²) |
|---|---|---|---|
| 31 January | −4.00% | −4.00 | 16.00 |
| 28 February | +1.00% | 0.00 | 0.00 |
| 31 March | 0.00% | 0.00 | 0.00 |
| 30 April | +5.00% | 0.00 | 0.00 |
The squared sum is 16 pp². Divide by four and take the square root to get 2.00 pp per month. Dividing by one would give 4.00 pp, the size of the sole loss. Including the three zeros lets the calculation reflect that only one of four months missed zero. You can report the one-in-four miss count separately. Another return path could produce the same downside deviation with a different number of losses.
Worked case 2. Raise the target on the same four months
Now ask how much those returns fell short of a +2.00% monthly target. Keep the dates, returns and all other assumptions fixed.
| Monthly return | Return less the 2% target (pp) | Signed shortfall (pp) | Squared shortfall (pp²) |
|---|---|---|---|
| −4.00% | −6.00 | −6.00 | 36.00 |
| +1.00% | −1.00 | −1.00 | 1.00 |
| 0.00% | −2.00 | −2.00 | 4.00 |
| +5.00% | +3.00 | 0.00 | 0.00 |
The sum of squares rises to 41 pp². The downside deviation is \(\sqrt{41/4}\), or 3.20 pp per month. February and March now contribute. January’s shortfall also deepens by 2 pp. The fund earned exactly the same returns in both cases. The higher result describes a more demanding goal.
Keep the horizon in the question
If the target is 12% effective annually and you want a monthly equivalent, use the monthly rate that compounds to 12%, about 0.9489%. Using 12% as a monthly target would classify many more observations as misses. If the question is about missing the target over full years, compound the returns into annual observations and compare them with the annual target. Applying a shortfall rule to compounded annual returns can give a different answer from scaling monthly downside deviation.
If no observation misses the target, the historical result under the full-sample convention is zero. A future shortfall remains possible. Disclose the observation count, target, measurement frequency and fee basis when making comparisons. The Sortino ratio uses this downside measure in its denominator and must use the same target when finding return above it.
Practice questions
Choose one answer per question. Use the lesson’s conventions unless the question specifies a change. Round only the final result.
1. Count a positive miss
The monthly target is +1.50% and the fund returns +0.50%. What is the signed shortfall?
2. Use every observation in the divisor
Four returns are −6%, −2%, 0% and +4%. The target is 0%. What is downside deviation?
3. Move the target
Keep every observed return fixed and raise the target. What can happen to downside deviation under the full-sample convention?
4. Report no misses
All monthly returns exceed the stated monthly target. What is the measured full-sample downside deviation?
5. Match a yearly objective
A target is 21% effective annually. You use quarterly returns. Which constant quarterly target compounds to it?
6. Distinguish two types of dispersion
One fund has several unusually high gains but very small misses below its target. Which statement is possible?