Why Portfolio Variance Matters

Worked Solution

How to Think About It: If you hold multiple assets, the total risk of your portfolio is not just the sum of each asset's risk. It depends critically on how the assets move together. That is the entire point of portfolio variance -- it is a single number that encodes both individual asset risks and their co-movement. The reason this matters practically: a portfolio of ten volatile-but-uncorrelated assets can be far less risky than a single "safe" asset. Portfolio variance is the formula that makes this precise.

Key Insight: The cross terms (covariances) in the portfolio variance formula are the mathematical source of diversification. When correlations are below 1, these cross terms shrink the total variance below what you would get from a naive weighted sum. When $\rho = -1$, you can hedge away all risk entirely.

The Formulas:

*Two-asset case:*

$\sigma_p^2 = w_1^2 \sigma_1^2 + w_2^2 \sigma_2^2 + 2 w_1 w_2 \sigma_1 \sigma_2 \rho_{12}$

The first two terms are the variance contributions from each asset in isolation. The third term -- the covariance term -- is where diversification lives. If $\rho_{12} < 1$, this term is smaller than it would be under perfect correlation, and total portfolio variance drops.

*General $n$-asset case (matrix form):*

$\sigma_p^2 = \mathbf{w}^T \Sigma \mathbf{w} = \sum_{i=1}^{n} \sum_{j=1}^{n} w_i w_j \sigma_{ij}$

where $\mathbf{w}$ is the vector of portfolio weights and $\Sigma$ is the covariance matrix with $\sigma_{ij} = \rho_{ij} \sigma_i \sigma_j$.

Extreme Cases (two-asset, equal weights $w_1 = w_2 = 0.5$ for intuition):

• $\rho = 1$ (perfect correlation): $\sigma_p^2 = (w_1 \sigma_1 + w_2 \sigma_2)^2$. The portfolio standard deviation is just the weighted average of individual standard deviations. No diversification benefit at all -- the assets are effectively the same risk factor.

• $\rho = 0$ (uncorrelated): $\sigma_p^2 = w_1^2 \sigma_1^2 + w_2^2 \sigma_2^2$. The cross term vanishes. With equal weights and equal volatilities $\sigma$, you get $\sigma_p = \sigma / \sqrt{2}$, a $\sim 30\%$ reduction in risk. This is why naive diversification (just spreading your money around) already helps.

• $\rho = -1$ (perfect negative correlation): $\sigma_p^2 = (w_1 \sigma_1 - w_2 \sigma_2)^2$. You can achieve $\sigma_p = 0$ by choosing weights:

$w_1 = \frac{\sigma_2}{\sigma_1 + \sigma_2}, \quad w_2 = \frac{\sigma_1}{\sigma_1 + \sigma_2}$

This is a perfect hedge. In practice, $\rho = -1$ almost never occurs between real assets, but this is the theoretical ideal that diversification is chasing.

Practical Considerations:

• In practice, estimated covariance matrices are noisy, especially with many assets. This is why regularization (shrinkage, factor models) matters so much in real portfolio optimization.
• Portfolio variance assumes returns are fully characterized by their first two moments. For assets with heavy tails or skew (which is common in finance), variance alone can miss important risks. That said, it remains the standard starting point because mean-variance optimization is tractable and well-understood.
• The number of parameters in $\Sigma$ grows as $n(n+1)/2$, so with 500 assets you need to estimate over 125,000 covariance terms. Factor models reduce this by assuming returns are driven by a small number of common factors.

Answer: Portfolio variance is important because it captures total portfolio risk including the diversification benefit from imperfect correlations between assets. It is computed as $\sigma_p^2 = \mathbf{w}^T \Sigma \mathbf{w}$, where the off-diagonal covariance terms are what drive diversification. When $\rho < 1$, portfolio risk is less than the weighted average of individual risks, and in the extreme case of $\rho = -1$, risk can be eliminated entirely with the right weight combination.