Variance Finance: The Building Block of Portfolio Risk Math

Key Takeaways

• Variance finance uses squared deviations to make positive and negative swings contribute equally to the risk measure, and to penalise large outliers more than small ones.
• Portfolio variance depends on the covariance between holdings, not just individual variances, two 15% standard deviation funds can combine below 15% if they are uncorrelated.
• A common mistake is averaging component variances to estimate portfolio variance, which ignores the covariance term and overstates actual portfolio risk.
• Variance is the natural unit for portfolio optimisation because it is additive; standard deviation should be the final conversion only after all the math is done.

What It Is

Variance measures how spread out a set of numbers is around their average. In finance, the numbers are usually asset returns, and the variance tells you how much those returns typically bounce above and below their mean. A small variance means returns cluster tightly. A large variance means they fan out.

Because the deviations from the mean are squared before averaging, variance is always a non-negative number, and it is reported in squared units. If returns are in percent, variance is in "percent squared." That unit is awkward for human intuition, which is why investors usually convert variance into its square root, standard deviation, for reporting. Variance itself, however, is the form that actually composes cleanly in portfolio math.

The Intuition

Why square the deviations instead of just taking their absolute value? Two reasons.

First, squaring makes positive and negative deviations contribute in the same direction. Without squaring, positive and negative swings would cancel out when averaged, and the result would always be zero.

Second, squaring punishes large deviations more than small ones. A return that lands 10 percent away from the mean contributes 100 to the squared sum, while a return only 2 percent away contributes 4. That quadratic weighting lines up with how most investors feel about risk, where a single big loss hurts far more than many small ones.

The practical payoff: variance is additive in a way that absolute deviation is not. When you combine assets into a portfolio, the portfolio variance can be written as a clean function of the individual variances and the covariances between the assets. That property is why Harry Markowitz built modern portfolio theory on variance rather than any other risk measure.

How It Works

For a sample of N returns, the sample variance is:

mean = (R1 + R2 + ... + RN) / N

variance = sum((Ri - mean)^2) / (N - 1)

The divisor is N minus 1 for a sample, a correction known as Bessel's correction, which keeps the estimate unbiased. For a full population, you divide by N.

Standard deviation is simply the square root of variance:

std dev = sqrt(variance)

For a two-asset portfolio with weights w1 and w2, the portfolio variance is:

var_p = w1^2 * var1 + w2^2 * var2 + 2 * w1 * w2 * cov(1,2)

Where cov(1,2) is the covariance between the two assets' returns. The covariance term is what makes diversification work: if two assets are not perfectly correlated, the portfolio variance comes out lower than a weighted average of the individual variances.

Worked Example

Take the same five annual returns used in the standard deviation article: 10 percent, 4 percent, minus 2 percent, 12 percent, and 6 percent.

Step 1, the mean:

(10 + 4 + (-2) + 12 + 6) / 5 = 30 / 5 = 6%

Step 2, each return minus the mean:

10 - 6 =  4
 4 - 6 = -2
-2 - 6 = -8
12 - 6 =  6
 6 - 6 =  0

Step 3, square each deviation:

16, 4, 64, 36, 0

Step 4, sum and divide by N minus 1:

variance = (16 + 4 + 64 + 36 + 0) / (5 - 1) = 120 / 4 = 30

The variance is 30 "percent squared." To get back to a unit humans can read, take the square root:

std dev = sqrt(30) ~= 5.48%

Now a portfolio illustration. Suppose Asset A has variance 256 (16 percent standard deviation), Asset B has variance 625 (25 percent standard deviation), and the covariance between them is 120. You put 80 percent in A and 20 percent in B:

var_p = 0.8^2 * 256 + 0.2^2 * 625 + 2 * 0.8 * 0.2 * 120
      = 0.64 * 256 + 0.04 * 625 + 38.4
      = 163.84 + 25.00 + 38.4
      = 227.24

Portfolio standard deviation is sqrt(227.24) = 15.07 percent. Notice the result is below the naive weighted average of 16 and 25 percent, because the covariance term captured the diversification benefit.

Frequently Asked Questions

Q: What is variance in finance in simple terms? Variance measures how spread out a set of returns is around its average, using squared deviations. It always comes out positive, is in squared units, and takes the square root to become the standard deviation most investors quote.

Q: How does variance affect investment decisions? Variance is the core ingredient in mean-variance portfolio optimisation. Changing one position's weight affects the whole portfolio's variance through the covariance matrix, not just through its own individual variance.

Q: What is a real-world example of variance in finance? A fund with five annual returns of 10%, 4%, -2%, 12%, and 6% has a mean of 6% and a variance of 30 (percent squared). Taking the square root gives a standard deviation of about 5.5%, which is the number you would actually compare to other funds.

Q: How can investors use variance to improve portfolio construction? Adding a low-variance asset that is negatively correlated with an existing holding reduces total portfolio variance more than adding a low-variance asset that is highly correlated. The covariance term, not the individual variances, drives the diversification benefit.

Q: How is variance different from standard deviation? Variance and standard deviation measure the same thing, but variance is in squared units (percent squared) while standard deviation is in the same units as the returns (percent). Standard deviation is easier to interpret; variance is more convenient for the underlying math.

Common Mistakes

1. Averaging variances when you should use standard deviations, or vice versa. Variance is additive under certain portfolio math. Standard deviation is not. Mixing the two produces garbage. Always operate in variance space, then convert to standard deviation at the end.

2. Forgetting the Bessel correction. Dividing by N instead of N minus 1 gives a biased estimate for a sample. For small datasets this matters materially. Most spreadsheet functions distinguish between VAR.S (sample) and VAR.P (population). Pick the right one.

3. Ignoring covariance when combining assets. Portfolio variance is not the weighted average of component variances. The covariance term can either reduce risk (uncorrelated or negatively correlated assets) or inflate it (highly correlated assets). Skipping it is the number-one beginner error in portfolio math.

4. Treating variance itself as a risk number in conversations. "My portfolio has a variance of 225" is mathematically fine but practically unhelpful. Convert to standard deviation (15 percent) so the figure is in the same units as returns.

5. Assuming historical variance predicts future variance. Volatility clusters and regimes shift. A calm-period variance can understate the true long-run figure by a wide margin. Most practitioners blend recent and long-run variance, or use exponential weighting, rather than a single window.

Sources

1. Corporate Finance Institute. "Variance." https://corporatefinanceinstitute.com/resources/data-science/variance/
2. Corporate Finance Institute. "Variance Formula." https://corporatefinanceinstitute.com/resources/data-science/variance-formula/
3. Corporate Finance Institute. "Portfolio Variance." https://corporatefinanceinstitute.com/resources/data-science/portfolio-variance/
4. Robinhood Learn. "What is Variance?" https://robinhood.com/us/en/learn/articles/41NvPULxnUjc0GSyotjuNT/what-is-variance/
5. AnalystPrep. "Portfolio Standard Deviation." https://analystprep.com/cfa-level-1-exam/portfolio-management/portfolio-standard-deviation/

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.