Standard Deviation and What Volatility Measures
In this chapter
Intermediate
A monthly return variance of 16 percentage points squared corresponds to a standard deviation of four percentage points. Standard deviation measures spread around the average in the same units as the returns. That makes it easier to read than variance when you are comparing how widely two funds have moved.
Take the square root to get back to return units
Variance combines squared distances from a mean. Taking its nonnegative square root reverses the squaring of the units. If variance is in decimal return units squared, the root is a decimal return. If variance is in percentage points squared (pp²), the root is in percentage points (pp).
\[ \sigma=\sqrt{\sigma^2}\qquad s=\sqrt{s^2} \]
\(\sigma^2\) and \(\sigma\) are population variance and standard deviation. \(s^2\) and \(s\) are sample variance and standard deviation. Take the root of the variance suited to the data and the question. The standard deviation of a series of identical observations is zero.
Worked case. Read two funds’ monthly spread
The following figures describe two complete, equally spaced sets of hypothetical monthly nominal USD total returns, including investment income and before fees and taxes. Both funds averaged 2.00% over the chosen months.
| Fund | Descriptive monthly variance (pp²) | Monthly standard deviation (pp) | Mean minus and plus one standard deviation |
|---|---|---|---|
| P | 16.00 | 4.00 | −2.00% to +6.00% |
| Q | 36.00 | 6.00 | −4.00% to +8.00% |
Fund Q had one and a half times Fund P’s spread on this scale. For a return quoted as 2%, adding four percentage points gives 6%. The ranges help you picture the size of the spread. A return can land beyond either endpoint. How often that happens depends on the shape of the distribution. The familiar rule that about 68% of outcomes lie within one standard deviation of the mean assumes a normal distribution.
If the worksheet gives Fund P’s variance as 0.0016, the square root is 0.04, which is the same monthly spread. Check the observation interval before comparing. A four-point monthly standard deviation and a four-point yearly standard deviation describe movement over different lengths of time. Turning monthly spread into annual spread needs assumptions about how returns behave across time.
A big gain can raise measured volatility
In this lesson, volatility is standard deviation of returns over a stated period. Standard deviation counts unusually high and unusually low returns because both can be far from the mean.
Worked case. Replace one gain and keep the losses
The first set below describes all five months in an illustrative window. In the second set, a single unusually good month replaces the final observation. Treat each set as complete and use the same return basis as above.
| Window | Month 1 | Month 2 | Month 3 | Month 4 | Month 5 | Mean |
|---|---|---|---|---|---|---|
| Original | −4% | −2% | 0% | +2% | +4% | 0.00% |
| With the jump | −4% | −2% | 0% | +2% | +14% | 2.00% |
In the original window the squared distances sum to \(16+4+0+4+16=40\) pp². Divide by five and take the square root of 8. The spread is 2.83 pp.
The 14% gain raises the mean to 2%. Rebuild all five deviations around the new mean. They are −6, −4, −2, 0 and +12 pp. The squares sum to \(36+16+4+0+144=200\) pp². The variance is 40 pp² and the spread is 6.32 pp. Both windows still contain exactly two losses, −4% and −2%. Standard deviation jumped because the outcomes spread out, including on the upside. When you care about shortfalls below a stated target, downside deviation focuses on those observations.
A steady loss can also have zero standard deviation. A fund that returns −1% in each month has no distance from its own average, although the balance declines month after month. If your question is how often the fund lost money, count returns below zero in the window. If you need to know how badly it missed a required return, choose that threshold and measure the shortfalls. The loss count and the shortfalls both require the individual return observations. You can also have little measured spread in a short run that misses the conditions you care about.
Practice questions
Choose one answer per question. Round only the final result where needed.
1. Take the root in the right units
Monthly variance is 0.0025 in decimal return units squared. What is monthly standard deviation?
2. Compare two variances
Fund A’s variance is 9 pp² and Fund B’s is 81 pp² on the same basis. How many times as large is B’s standard deviation?
3. Recalculate after a big gain
A complete three-month set changes from 0%, 0%, +3% to 0%, 0%, +6%. What happens to standard deviation?
4. Read a zero
A fund returns −1% each month in a complete three-month window. What was its monthly standard deviation?
5. Interpret a one-standard-deviation band
A fund averages 4% per month and its monthly standard deviation is 3 percentage points. What follows for the band from 1% to 7%?
6. Choose a measure for a loss question
You want to know how many months the return fell below −2%. What would answer that directly?