Why Portfolio Beta Is A Weighted Average

Why Portfolio Beta Can Be Calculated From Its Holdings

A portfolio’s return depends on the returns of its holdings. Its beta describes how that combined return covaries with the market return, relative to the market’s own variance. Because covariance is linear in its first argument, the beta of a portfolio can be calculated from the betas of its assets and their portfolio weights. The weights must remain signed: a short position has a negative weight.

This result is useful for assessing how a change in holdings changes systematic market exposure. It does not mean that portfolio beta measures total portfolio risk. In particular, different portfolios can have the same beta and different amounts of firm-specific risk or total return variability.

Definition: For an asset \(i\) with return \(R_i\) and a market return \(R_M\), its beta is \(\beta_i=\operatorname{Cov}(R_i,R_M)/\operatorname{Var}(R_M)\), provided \(\operatorname{Var}(R_M)>0\). For a portfolio with return \(R_P\), its beta is \(\beta_P=\operatorname{Cov}(R_P,R_M)/\operatorname{Var}(R_M)\). All returns in these definitions must refer to the same period and the same market return.

Suppose a portfolio has \(n\) assets. Let \(w_i\) be the amount invested in asset \(i\) divided by the portfolio’s net value at the beginning of the period. For a fully invested portfolio, \(\sum_{i=1}^{n}w_i=1\). A long holding has \(w_i>0\); a short holding has \(w_i<0\). Under the assumptions below, the portfolio return is the weighted sum of the asset returns.

Deriving the Portfolio-Beta Formula

The formula requires that portfolio returns are linear in the holdings. In particular, assume the weights are set at the start of the return period, returns are measured over the same period, and the portfolio has no unaccounted-for cash flows or nonlinear payoff features that break the weighted-return relation. The market return must also be the same for every constituent beta, and its variance must be positive.

Under these assumptions, the portfolio return is \(R_P=\sum_{i=1}^{n}w_iR_i\). Applying the covariance definition and its linearity in the first argument gives the result directly:

\begin{aligned} \beta_P &=\frac{\operatorname{Cov}(R_P,R_M)}{\operatorname{Var}(R_M)}\\ &=\frac{\operatorname{Cov}\left(\sum_{i=1}^{n}w_iR_i,R_M\right)}{\operatorname{Var}(R_M)}\\ &=\frac{\sum_{i=1}^{n}w_i\operatorname{Cov}(R_i,R_M)}{\operatorname{Var}(R_M)}\\ &=\sum_{i=1}^{n}w_i\frac{\operatorname{Cov}(R_i,R_M)}{\operatorname{Var}(R_M)}\\ &=\sum_{i=1}^{n}w_i\beta_i. \end{aligned}

The key step is that multiplying an asset return by a fixed weight multiplies its covariance with the market by that weight. Summing those weighted covariances gives the portfolio’s covariance with the market. Dividing by the common market variance then gives the weighted sum of the constituent betas. This derivation does not require the assets’ returns to be independent; covariance between assets affects portfolio variance, but it does not add a separate term to this beta formula.

Theorem: For a portfolio whose return is \(R_P=\sum_i w_iR_i\), with constituent betas calculated against the same market return over the same period, \(\beta_P=\sum_iw_i\beta_i\). If the weights are nonnegative and sum to one, this is a conventional weighted average, so portfolio beta lies between the smallest and largest constituent betas. If some weights are negative, the formula remains valid, but the result need not lie within that range.

For a long-only, fully invested portfolio, the weights are nonnegative and sum to one. Each weight can then be interpreted as a share of the portfolio’s value. With short positions, weights are still measured relative to net portfolio value and still sum to one for a fully invested net portfolio, but they are signed. A short position contributes negatively to portfolio return and therefore contributes its beta multiplied by a negative weight.

For example, if a portfolio has weights of \(1.20\) and \(-0.20\), the investor is long assets worth 120% of net portfolio value and short assets worth 20%. The weights sum to one. The beta calculation uses \(1.20\) and \(-0.20\), not the absolute position sizes divided by their total.

Worked Calculations

Worked Example: A Two-Asset Long-Only Portfolio

A hypothetical portfolio invests 60% in Asset A and 40% in Asset B. Their betas, each measured against the same market return over the same period, are \(1.20\) and \(0.30\), respectively. The market-return variance is \(0.0400\). Calculate the portfolio beta both from the weighted betas and by checking the covariance calculation.

Method and assumptions. The holdings are set at the start of the period, the portfolio return is the weighted sum of the two asset returns, and the weights are long-only and fully invested. Thus \(w_A=0.60\), \(w_B=0.40\), and \(w_A+w_B=1.00\).

Weighted-beta calculation. Multiply each beta by its portfolio weight and add:

\beta_P=(0.60)(1.20)+(0.40)(0.30)=0.72+0.12=0.84.

Covariance check. From \(\beta_i=\operatorname{Cov}(R_i,R_M)/\operatorname{Var}(R_M)\), the asset covariances with the market are \(\operatorname{Cov}(R_A,R_M)=(1.20)(0.0400)=0.0480\) and \(\operatorname{Cov}(R_B,R_M)=(0.30)(0.0400)=0.0120\). The portfolio covariance is \((0.60)(0.0480)+(0.40)(0.0120)=0.0288+0.0048=0.0336\). Therefore \(\beta_P=0.0336/0.0400=0.84\), agreeing with the weighted-beta result.

Check and common mistake avoided. The weights sum to one, and the direct covariance calculation gives the same beta. A common error is to average the two betas equally, obtaining \(0.75\); that ignores the unequal portfolio weights.

Worked Example: A Long Position Financed Partly by a Short

A hypothetical investor holds 125% of net portfolio value in Asset C and shorts 25% in Asset D. Their betas against the same market return are \(1.40\) and \(0.60\). The market-return variance is \(0.0500\). Find the portfolio beta and verify it using covariances.

Method and assumptions. Assume the positions are set at the start of the period and the portfolio return is linear in the signed holdings. The weights are \(w_C=1.25\) and \(w_D=-0.25\), with \(1.25-0.25=1.00\). Keep the short weight negative.

Weighted-beta calculation. Apply the formula using the signed weights:

\beta_P=(1.25)(1.40)+(-0.25)(0.60)=1.75-0.15=1.60.

Covariance check. The constituent covariances with the market are \(\operatorname{Cov}(R_C,R_M)=(1.40)(0.0500)=0.0700\) and \(\operatorname{Cov}(R_D,R_M)=(0.60)(0.0500)=0.0300\). Hence the portfolio covariance is \((1.25)(0.0700)+(-0.25)(0.0300)=0.0875-0.0075=0.0800\). Dividing by the market variance gives \(\beta_P=0.0800/0.0500=1.60\).

Check and common mistake avoided. The covariance check agrees, and the net weights sum to one. A tempting but incorrect calculation uses absolute holdings and renormalizes them: \(\big((1.25)(1.40)+(0.25)(0.60)\big)/(1.25+0.25)=1.90/1.50\approx1.27\). That treats the short as a positive holding. The short’s return contribution has the opposite sign, so its weight must be \(-0.25\), not \(+0.25\).

Worked Example: A Long-Only Portfolio With a Negative-Beta Asset

A hypothetical long-only portfolio places 25% in Asset E, 50% in Asset F, and 25% in Asset G. Their betas against the same market return are \(1.60\), \(0.80\), and \(-0.40\), respectively. The market-return variance is \(0.0900\). Calculate the portfolio beta and confirm it from the covariances.

Method and assumptions. Assume a common return period, beginning-of-period weights, and a weighted-sum portfolio return. The weights are \(0.25\), \(0.50\), and \(0.25\); they are all nonnegative and sum to \(1.00\).

Weighted-beta calculation. The negative beta of Asset G is retained as given:

\beta_P=(0.25)(1.60)+(0.50)(0.80)+(0.25)(-0.40) =0.40+0.40-0.10=0.70.

Covariance check. The respective covariances with the market are \((1.60)(0.0900)=0.1440\), \((0.80)(0.0900)=0.0720\), and \((-0.40)(0.0900)=-0.0360\). The portfolio covariance is \((0.25)(0.1440)+(0.50)(0.0720)+(0.25)(-0.0360)=0.0360+0.0360-0.0090=0.0630\). Thus \(\beta_P=0.0630/0.0900=0.70\).

Check and common mistake avoided. The direct and weighted calculations agree. Negative beta does not mean that the asset has a negative portfolio weight: Asset G is held long, so its portfolio weight is positive, while its beta contribution is negative. Do not replace the asset’s beta by its absolute value.

Interpreting the Result and Recognizing Its Limits

With long-only weights that sum to one, portfolio beta is bounded by the lowest and highest constituent betas. This follows because it is a weighted average: no nonnegative combination with weights summing to one can be below every input or above every input. A negative-beta asset can lower the portfolio beta even when it is held long, as the third example showed.

Mixed-weight portfolios do not have the same bound. A short position subtracts the shorted asset’s beta contribution, so the portfolio beta can be outside the range of constituent betas. This is not an error in the formula; it reflects leveraged net exposure. In a fully invested portfolio, signed weights sum to one, but their absolute values can sum to more than one.

The formula also depends on consistent measurement. Each constituent beta must use the same market return and time period. A beta estimated against one market proxy cannot simply be combined with another beta estimated against a different proxy as though they shared a denominator and benchmark. The portfolio weights must correspond to the returns being combined. If weights are reset during a period, the portfolio’s actual period return may not equal a single fixed-weight sum of full-period constituent returns; use weights and returns consistent with the actual holding and rebalancing process.

Finally, portfolio beta is a measure of market-related covariance, not a complete description of portfolio risk. The weighted-beta identity does not imply that portfolio variance is the weighted average of constituent variances. Portfolio variance also depends on covariances among the assets. Use beta to summarize market sensitivity, and use the appropriate portfolio-risk calculation when the question concerns total variability.

Key takeaway: First verify the return period, common market benchmark, and portfolio weights. Then calculate \(\beta_P=\sum_i w_i\beta_i\), preserving negative signs for short positions and for negative constituent betas. Check that long-only weights sum to one, or that signed weights sum to one for a fully invested net portfolio.

Check Your Understanding

For each problem, use betas measured against the same market return over the same period, and assume portfolio returns are linear in the stated holdings.

A fully invested portfolio holds 70% in an asset with beta \(1.10\) and 30% in an asset with beta \(0.20\). Calculate its beta. (Answer: \(0.83\))
A long-only portfolio invests 40% in an asset with beta \(1.50\), 35% in an asset with beta \(0.70\), and 25% in a risk-free asset with beta \(0\). Calculate its beta. (Answer: \(0.845\))
A net portfolio is long 120% in an asset with beta \(1.25\) and short 20% in an asset with beta \(0.50\). Verify that the weights sum to one and calculate the portfolio beta. (Answer: weights sum to \(1.00\); portfolio beta \(=1.40\))
A long-only portfolio has weights \(0.20\), \(0.50\), and \(0.30\), with respective betas \(1.80\), \(0.90\), and \(-0.20\). Calculate the portfolio beta and state whether it lies within the range of constituent betas. (Answer: portfolio beta \(=0.75\); yes, it lies between \(-0.20\) and \(1.80\))
A portfolio is long 110% in an asset with beta \(1.30\) and short 10% in an asset with beta \(0.40\). A candidate divides the sum of the absolute-weight beta contributions by \(1.20\). Give the correct portfolio beta and explain the sign correction. (Answer: \(1.39\); use \(1.10(1.30)-0.10(0.40)\), because the short weight is negative, not positive.)

What You'll Learn