Common Errors In Portfolio Risk Calculations

Why Portfolio-Risk Errors Are Easy to Miss

A portfolio-risk calculation can look plausible and still be wrong. A result may have the right general scale while omitting a covariance term, treating correlation as if it were covariance, or combining weights measured against different totals. These errors matter because portfolio risk depends not only on each asset’s standalone volatility but also on how asset returns move together.

The central discipline is to keep the inputs and the calculation aligned: use weights that refer to the same portfolio, express returns and standard deviations consistently, and use covariances in the variance formula. The earlier material on portfolio diversification establishes the relevant relationships; here the emphasis is on diagnosing frequent calculation failures.

Definition: For asset returns $R_1,\ldots,R_n$ and fixed portfolio weights $w_1,\ldots,w_n$, the one-period portfolio return is $R_p=\sum_{i=1}^n w_iR_i$. Portfolio variance is $\operatorname{Var}(R_p)$, and portfolio standard deviation is $\sigma_p=\sqrt{\operatorname{Var}(R_p)}$. For a fully invested portfolio, the weights sum to 1. Short positions are permitted when a weight is negative, provided all weights still refer to the same portfolio value.

The Formula Behind the Error Checks

Assume that all asset returns refer to the same holding period, that weights are fixed at the start of that period, and that each return is measured in the same units. The variance formula follows by expanding the variance of the weighted sum. Covariance is defined by $\operatorname{Cov}(X,Y)=E[(X-E[X])(Y-E[Y])]$; it measures joint movement in the variables’ units.

\begin{aligned} \operatorname{Var}(R_p) &=\operatorname{Var}\left(\sum_{i=1}^n w_iR_i\right)\\ &=\sum_{i=1}^n w_i^2\operatorname{Var}(R_i) +2\sum_{i<j}w_iw_j\operatorname{Cov}(R_i,R_j)\\ &=\sum_{i=1}^n w_i^2\sigma_i^2 +2\sum_{i<j}w_iw_j\operatorname{Cov}(R_i,R_j). \end{aligned}

To see why the cross terms appear, center each return by subtracting its mean and expand the square: $(\sum_i w_i(R_i-E[R_i]))^2$. The square contains each individual squared term $w_i^2(R_i-E[R_i])^2$ and, for every pair $i<j$, two identical cross-products. Taking expectations gives the variance terms and twice each covariance term. This is why omitting covariance is not a harmless simplification unless all relevant covariances are zero.

When the input is a correlation rather than a covariance, first convert it. For two assets, $\operatorname{Cov}(R_i,R_j)=\rho_{ij}\sigma_i\sigma_j$, where $\rho_{ij}$ is correlation and $\sigma_i,\sigma_j$ are standard deviations. Correlation is dimensionless and lies between $-1$ and $1$; covariance is in squared return units and is not generally bounded by 1.

Worked Examples: Catching the Common Errors

Worked Example: Averaging Standard Deviations Instead of Calculating Portfolio Risk

A hypothetical investor places 60% of a portfolio in Fund A and 40% in Fund B. Their annual return standard deviations are 10% and 20%, and their return correlation is 0.25. Calculate the portfolio standard deviation. Identify the error in using the weighted average of the two standard deviations.

Method. Convert the percentage inputs to decimals, calculate covariance, and substitute into the two-asset variance formula. The weights are $w_A=0.60$ and $w_B=0.40$, and they sum to 1.

First calculate covariance: $\operatorname{Cov}(R_A,R_B)=0.25(0.10)(0.20)=0.005$. The diagonal terms in the variance are $0.60^2(0.10^2)=0.36(0.01)=0.0036$ and $0.40^2(0.20^2)=0.16(0.04)=0.0064$. The covariance contribution is $2(0.60)(0.40)(0.005)=0.0024$.

\operatorname{Var}(R_p)=0.0036+0.0064+0.0024=0.0124, \qquad \sigma_p=\sqrt{0.0124}=0.111355\ldots

Rounded to two decimal places as a percentage, the portfolio standard deviation is 11.14%. The tempting weighted average, $0.60(10\%)+0.40(20\%)=14\%$, is not the portfolio standard deviation formula. It ignores the way the two returns move together. As a check, the calculated risk is below the 14% weighted average of standalone risks, which is consistent with the imperfect positive correlation in this example.

Common mistake avoided: averaging standard deviations gives 14%, not the correct 11.14%. Another tempting error is to drop covariance; that would give $\sqrt{0.0036+0.0064}=10\%$, also incorrect because the correlation is not zero.

Worked Example: Converting Correlations and Keeping Percentages Consistent

A hypothetical three-asset portfolio has weights $0.20, 0.50, 0.30$, standard deviations $10\%,12\%,20\%$, and correlations $\rho_{12}=0.30$, $\rho_{13}=0$, and $\rho_{23}=-0.20$. Calculate all three pairwise covariances and the portfolio standard deviation.

Method. Use decimal standard deviations in the covariance conversion, then include each covariance once in the pairwise sum, multiplied by 2. The weights sum to $0.20+0.50+0.30=1$.

The standard deviations in decimal form are $0.10,0.12,0.20$. Thus:

$\operatorname{Cov}(R_1,R_2)=0.30(0.10)(0.12)=0.0036$.
$\operatorname{Cov}(R_1,R_3)=0(0.10)(0.20)=0$.
$\operatorname{Cov}(R_2,R_3)=-0.20(0.12)(0.20)=-0.0048$.

The weighted variance terms are $0.20^2(0.10^2)=0.0004$, $0.50^2(0.12^2)=0.0036$, and $0.30^2(0.20^2)=0.0036$. The three covariance contributions are $2(0.20)(0.50)(0.0036)=0.00072$, $2(0.20)(0.30)(0)=0$, and $2(0.50)(0.30)(-0.0048)=-0.00144$.

\begin{aligned} \operatorname{Var}(R_p) &=0.0004+0.0036+0.0036+0.00072+0-0.00144\\ &=0.00688,\\ \sigma_p&=\sqrt{0.00688}=0.0829458\ldots \end{aligned}

The portfolio standard deviation is 8.29% when rounded to two decimal places as a percentage. A check is that the variance is positive, and the negative covariance between assets 2 and 3 reduces the total variance. The square root is taken only after all variance contributions have been added.

Common mistake avoided: entering correlations directly as covariances would use $0.30$ and $-0.20$ in place of $0.0036$ and $-0.0048$, creating a serious unit and scale error. The opposite mismatch also occurs: writing 10% as 10 rather than 0.10 while other inputs are decimals. Use one representation consistently.

Worked Example: Using Weights Based on the Correct Portfolio Total

A hypothetical investor has a total portfolio worth $100,000: $40,000 in Asset A, $30,000 in Asset B, $20,000 in a cash account, and $10,000 in another risk-free holding. For this calculation, combine the two risk-free holdings into one cash position. Asset A has annual standard deviation 10%, Asset B has annual standard deviation 20%, and their correlation is 0.50. Assume the cash return is constant over the period. Calculate the risk of the total portfolio and explain why normalizing only by the risky assets’ total gives the wrong answer for this question.

Method. Divide each holding by the stated total portfolio value. The weights are $0.40$ in A, $0.30$ in B, and $0.30$ in cash; they sum to 1. A constant cash return has zero variance and zero covariance with the asset returns.

The covariance of A and B is $0.50(0.10)(0.20)=0.010$. The total portfolio variance is therefore $0.40^2(0.10^2)+0.30^2(0.20^2)+2(0.40)(0.30)(0.010)$. The individual terms are $0.0016$, $0.0036$, and $0.0024$, respectively.

\operatorname{Var}(R_p)=0.0016+0.0036+0.0024=0.0076, \qquad \sigma_p=\sqrt{0.0076}=0.0871779\ldots

The total portfolio standard deviation is 8.72%, rounded to two decimal places as a percentage. If the question instead asked for the risk of the risky assets considered as a separate portfolio, the denominator would be $70,000, giving weights $40/70$ and $30/70$. That is a different portfolio: its variance is $0.0076/0.70^2=0.0155102\ldots$, and its standard deviation is about 12.45%. It is not the risk of the stated $100,000 portfolio.

Common mistake avoided: using weights $40/70$ and $30/70$ silently removes the cash allocation and changes the portfolio being measured. Always make the denominator explicit before calculating weights.

A Practical Error-Diagnosis Sequence

A reliable calculation can be checked in a fixed order. This sequence does not replace the formula; it helps locate the source of a wrong result when the final number seems plausible.

Define the portfolio.
Confirm the total value, the holdings included, and whether the target is the total portfolio or a subportfolio. Calculate each weight using that same denominator.

Standardize units.
Write percentage returns and standard deviations as decimals before multiplying, or use percentage units consistently throughout. Do not mix the two conventions in one expression.

Identify the dependence input.
If the data give correlation, convert it to covariance using both standard deviations. If the data already give covariance, do not multiply by the standard deviations again.

Include every pair.
For $n$ assets, include all $n$ diagonal variance terms and all $n(n-1)/2$ distinct covariance pairs, each multiplied by 2 in the expanded formula.

Take the square root last.
Sum signed variance and covariance contributions to obtain variance; only then take its nonnegative square root to obtain standard deviation.

The covariance terms may be positive, zero, or negative. A negative covariance contribution reduces portfolio variance, but it does not by itself guarantee a low-risk portfolio; the diagonal terms and all other pairs still matter. If calculated variance is negative, it cannot be a valid variance. Recheck arithmetic and the inputs, including whether a proposed covariance set is mathematically consistent.

Common Mistakes and Why They Matter

Averaging standard deviations. A weighted average of standalone standard deviations is generally not portfolio standard deviation. The portfolio formula is based on variance and covariance.
Omitting cross terms. Leaving out covariance assumes that every pair has zero covariance. That assumption is not justified unless it is part of the problem data.
Using correlation as covariance. Correlation has no return units; covariance depends on the magnitudes of both standard deviations.
Mixing percentages and decimals. The value 10% is 0.10 as a decimal. Squaring or multiplying percentage figures as though they were decimals can change the answer by large factors.
Using inconsistent weights. Weights based on the risky-asset subtotal describe a different portfolio from weights based on total portfolio value, which may include cash or other holdings.
Reporting variance as standard deviation. Variance is in squared return units; standard deviation is its square root and is reported in return units.

The same checks apply whether the inputs are population quantities or estimates from historical returns. Estimated covariances must correspond to the same assets and return periods as the weights and standard deviations. A sound final answer should state whether it is a variance or a standard deviation and whether a percentage is rounded.

Key takeaway: Portfolio risk is determined by the variance of a weighted sum, not by a weighted average of standalone risks. Use a single set of portfolio weights, convert correlations to covariances when needed, include every pairwise term, and take the square root only after calculating variance.

Check Your Understanding

For each problem, keep the return units and portfolio denominator consistent. Round portfolio standard deviations to two decimal places as percentages.

A portfolio invests 25% in Asset A and 75% in Asset B. Their standard deviations are 8% and 12%, and their correlation is $-0.25$. Calculate the covariance and portfolio standard deviation. (Answer: covariance $-0.0024$; portfolio standard deviation $8.72\%$.)
A two-asset portfolio has weights 0.40 and 0.60, standard deviations 15% and 10%, and correlation 0.20. Calculate its variance and standard deviation. (Answer: variance $0.00864$; standard deviation $9.30\%$.)
Two assets have standard deviations of 5% and 16% and correlation 0.40. Find their covariance, using decimal return units. (Answer: $0.0032$.)
A portfolio has weights 0.50, 0.30, and 0.20; standard deviations 10%, 20%, and 15%; and correlations $\rho_{12}=0$, $\rho_{13}=-0.20$, and $\rho_{23}=0.50$. Calculate all pairwise covariances and portfolio standard deviation. (Answer: covariances $0$, $-0.003$, and $0.015$, respectively; variance $0.0082$; standard deviation $9.06\%$.)
An investor’s total portfolio is $80,000, including $32,000 in Asset A, $24,000 in Asset B, and $24,000 in cash. Asset standard deviations are 10% and 20%, their correlation is 0.50, and cash has constant return. Calculate the total-portfolio standard deviation, then explain why weights based on the $56,000 risky-asset value would answer a different question. (Answer: total-portfolio standard deviation $8.72\%$; normalizing the two risky holdings by $56,000 measures the risky-asset subportfolio, not the $80,000 total portfolio.)

What You'll Learn