The computation of mean, variance and covariance statistics allow portfolio managers to compare the return-risk characteristics and potential portfolio impact of underlying securities. These metrics are quantitatively determined and rely on historic price or return data. While we can compute the historic profile, this does not necessarily mean the relationship between assets, or their return-risk profile will remain the same in the future.

**Mean**

The mean of a set of values or measurements is the sum of all the measurements divided by the sum of all the measurements in the set:

$$ \text{Mean} = \frac{\sum_{i=1}^{n} x_{i}}{n} $$

If we compute the mean of the population, we call it the parametric or population mean, denoted by μ (read “mu”). If we get the mean of the sample, we call it the sample mean and it is denoted by (read “x bar”).

**Population vs. Sample**

A population refers to the summation of all the elements of interest to the researcher.

- Examples: the no. of people in a country, the number of hedge funds in the U.S., or even the total no. of CFA candidates in a given year.

A sample is just a set of elements that represent the population as a whole. By analyzing sample data, we are able to make conclusions about the entire population.

- For example, if we sample the returns of 30 hedge funds spread across the U.S., we can use the results to make reasonable conclusions about the market as a whole (well over 10,000 hedge funds).

**Variance**

Variance is a measure of dispersion around the mean and is statistically defined as the average squared deviation from the mean. It is noted using the symbol σ².

$$ \sigma^2 = \frac{\sum_{i=1}^{N} (X_{i} – \mu)^2}{N} $$

Where μ is the population mean and N is population size.

The standard deviation, σ**,** is the square root of the variance and is commonly referred to as the volatility of the asset.Essentially it is a measure of how far on average the observations are from the mean. A population’s variance is given by:

The population standard deviation equals the square root of population variance. The sample variance is given by:

$$ S^2 = \frac{\sum_{i=1}^{N} (X_{i} – \bar{X})^2}{n-1} $$

Wher X-bar is the sample mean and n is the sample size.

Note that the sample standard deviation equals the square root of sample variance.

**Covariance**

Covariance is a measure of how closely two assets move together. In covariance, we focus on the relationship between the deviations of some two variables rather than the deviation from the mean of one variable.

If the means of random variables \(X\) and \(Y\) are known, then the covariance between the two random variables can be determined as follows:

$$ { \hat { \sigma } }_{ xy }=\frac { 1 }{ n } \sum _{ i=1 }^{ n }{ \left( { x }_{ i }-{ \mu }_{ x } \right) } \left( { y }_{ i }-{ \mu }_{ y } \right) $$

If we do not know the means, then the equation changes to:

$$ { \hat { \sigma } }_{ xy }=\frac { 1 }{ n-1 } \sum _{ i=1 }^{ n }{ \left( { x }_{ i }-{ \hat { \mu } }_{ x } \right) } \left( { y }_{ i }-{ \hat { \mu } }_{ y } \right) $$

### Correlation

Correlation is a concept that is closely related to covariance in the following way:

$$ { \rho }_{ xy }=\frac { { \sigma }_{ xy } }{ { \sigma }_{ x }{ \sigma }_{ y } } $$

Correlation ranges between +1 and -1 and is, therefore, much easier to interpret than covariance. Two variables are perfectly correlated if their correlation is equal to +1, uncorrelated if their correlation is equal to 0, and move in perfectly opposite directions if their correlation is equal to -1.

QuestionIn a two-asset portfolio, which combination of assets would result in the most diversified portfolio?

A. Correlation coefficient = 0.75

B. Correlation coefficient = -0.2

C. Correlation coefficient = 0

SolutionThe correct answer is B.

Within a two-asset portfolio, by combining negatively correlated assets, a diversified portfolio is produced and portfolio risk is lowered.

*Reading 52 LOS 52d:*

*Calculate and interpret the mean, variance, and covariance (or correlation) of asset returns based on historical data*