Why Calculate Beta From Paired Returns?
A historical beta estimate summarizes how an asset’s returns have varied alongside returns on a chosen market proxy over a specified set of matched periods. It is a calculation from observed data, not a guarantee that the same relationship will hold in the future. The key is to compare returns from the same periods and measure how their deviations from their respective averages move together.
The calculation uses sample covariance and sample market variance. Both use the same observations and the same sample denominator, so that denominator cancels in their ratio. This makes the arithmetic manageable, but it does not make centering optional: beta depends on deviations from the sample means, not on raw returns measured from zero.
A Reliable Calculation Process
Use one row per matched period. First calculate each series’ mean, then subtract the appropriate mean from every return. Multiply the paired deviations to form the covariance cross-products, and square the market deviations to form the market-variance terms. Divide each total by \(n-1\), and divide the resulting covariance by the market variance.
The sample denominator \(n-1\) is used for both sample quantities. It is important when reporting covariance and variance themselves, but it cancels when calculating beta. The cancellation is valid because the same periods are used in both calculations.
Confirm that each asset return is matched with the market return from the same period, and that both series use the same return units.
Find \(\bar R_A\) and \(\bar R_M\) from the observations included in the table.
For every period, calculate \(R_{A,t}-\bar R_A\) and \(R_{M,t}-\bar R_M\).
Sum paired deviation products for covariance and squared market deviations for market variance; divide each sum by \(n-1\).
Calculate \(\hat{\beta}=s_{AM}/s_M^2\). Check whether its sign and size are consistent with the paired observations.
Why the Beta Formula Holds
The formula is the ordinary least-squares slope when asset returns are related to market returns using an intercept. This justification explains why the covariance-to-variance ratio captures the direction and scale of the sample linear relationship. It does not require fitting or interpreting a full regression output to perform the historical beta calculation here.
To see the result, write the fitted asset return in period \(t\) as \(a+bR_{M,t}\), where \(a\) is an intercept and \(b\) is the slope. Least squares chooses \(a\) and \(b\) to minimize the sum of squared residuals. With an intercept, the fitted line passes through the sample means, so the slope can be found by minimizing the sum of squared differences between centered asset returns and \(b\) times centered market returns:
Differentiate with respect to \(b\), set the derivative to zero, and solve. If the market returns are not all identical, the centered market squared deviations sum to a positive number, so division is valid:
The last equality follows because both sample quantities divide their centered sums by \(n-1\). A positive beta means the centered returns tend to move in the same direction; a negative beta means they tend to move in opposite directions. The magnitude describes the slope in the chosen return units. It is not a percentage and is not a measure of total asset risk.
Worked Example: Centering Returns Before Taking the Ratio
A hypothetical investment’s returns and the returns of a market proxy are recorded over five matched periods. All figures below are percentage points. Calculate the sample covariance, sample market variance, and historical beta. Then check the result against the paired data.
| Period | Market return \(R_{M,t}\) | Asset return \(R_{A,t}\) |
|---|---|---|
| 1 | -1 | 0.2 |
| 2 | 0 | 0.2 |
| 3 | 1 | 2.2 |
| 4 | 2 | 4.2 |
| 5 | 3 | 4.2 |
Method and common mistake avoided. Calculate the means and centered quantities before taking a ratio. Using uncentered returns in place of deviations from the means would not calculate beta.
Step 1: Find the means. The market-return total is \(-1+0+1+2+3=5\), so \(\bar R_M=5/5=1\). The asset-return total is \(0.2+0.2+2.2+4.2+4.2=11\), so \(\bar R_A=11/5=2.2\).
Step 2: Center each observation. The market deviations are \(-2,-1,0,1,2\). The asset deviations are \(-2,-2,0,2,2\). Their paired products are \(4,2,0,2,4\), with total \(12\). The squared market deviations are \(4,1,0,1,4\), with total \(10\).
Step 3: Calculate the sample quantities. There are \(n=5\) observations, so the sample denominator is \(n-1=4\). Thus \(s_{AM}=12/4=3\) and \(s_M^2=10/4=2.5\), in squared percentage-point units.
Step 4: Calculate beta. The estimate is \(\hat{\beta}=3/2.5=1.20\). The centered-sum ratio gives the same result: \(12/10=1.20\).
Check. The estimate is positive, consistent with the generally positive pairing of centered returns. Its value above 1 reflects a positive sample slope greater than 1. This check supports the direction and scale; it does not establish that the estimate is precise or will persist.
The Raw-Return Trap
A tempting shortcut is to divide the sum of asset-market raw return products by the sum of squared raw market returns. That calculation omits the means and is generally not the covariance-to-variance ratio. The distinction can be made explicit using the example above.
The raw asset-market products are \((-1)(0.2)= -0.2\), \((0)(0.2)=0\), \((1)(2.2)=2.2\), \((2)(4.2)=8.4\), and \((3)(4.2)=12.6\). They sum to \(23\). The raw squared market returns sum to \(1+0+1+4+9=15\). The raw ratio is therefore \(23/15\), approximately \(1.5333\), not the beta estimate.
The correct centered totals are \(12\) for the cross-products and \(10\) for the market squared deviations, giving \(\hat{\beta}=12/10=1.20\). The different raw and centered ratios show why the calculation must use deviations from each series’ mean. In this example, dividing the raw totals by \(n-1=4\) would give \(5.75/3.75=23/15\) as well; using a common denominator does not repair the missing centering.
Worked Examples With Different Scales
Worked Example: Sample Covariance and Variance in Percentage Points
For four matched periods, a hypothetical asset has returns \(-1,0,1,2\) percentage points, while its market proxy has returns \(-2,0,2,4\) percentage points. Calculate the sample covariance, sample market variance, and beta.
Method and common mistake avoided. Use the sample means and deviations, and retain the sample denominator when reporting covariance and variance. Do not confuse those reported quantities with the beta ratio, where the common denominator cancels.
Step 1: Find the means. The market mean is \((-2+0+2+4)/4=1\). The asset mean is \((-1+0+1+2)/4=0.5\).
Step 2: Center and calculate products. The market deviations are \(-3,-1,1,3\), and the asset deviations are \(-1.5,-0.5,0.5,1.5\). The paired products are \(4.5,0.5,0.5,4.5\), summing to \(10\). The squared market deviations are \(9,1,1,9\), summing to \(20\).
Step 3: Divide by \(n-1\). Here \(n-1=3\), so \(s_{AM}=10/3\) and \(s_M^2=20/3\), both in squared percentage-point units.
Step 4: Take the ratio. The beta estimate is \((10/3)/(20/3)=10/20=0.50\).
Check. Across the observations, the centered asset deviations are half the corresponding centered market deviations. A slope of \(0.50\) is therefore consistent with the table. The sample covariance and variance have not been rounded; the exact ratio is \(0.50\).
Worked Example: Checking a Beta Greater Than One
A hypothetical asset and market proxy have the following returns, in percentage points, over five matched periods. Calculate beta and verify whether its size is consistent with the deviations.
| Period | Market return | Asset return |
|---|---|---|
| 1 | 0 | -4 |
| 2 | 2 | -1 |
| 3 | 4 | 2 |
| 4 | 6 | 5 |
| 5 | 8 | 8 |
Method and common mistake avoided. Calculate centered cross-products and market squares, rather than judging beta from the asset’s average return or from the largest observation alone.
Step 1: Find the means. The market mean is \((0+2+4+6+8)/5=4\). The asset mean is \((-4-1+2+5+8)/5=10/5=2\).
Step 2: Center the observations. Market deviations are \(-4,-2,0,2,4\); asset deviations are \(-6,-3,0,3,6\). The paired products are \(24,6,0,6,24\), summing to \(60\). Market squared deviations are \(16,4,0,4,16\), summing to \(40\).
Step 3: Calculate both sample quantities. Since \(n-1=4\), the sample covariance is \(60/4=15\) and the sample market variance is \(40/4=10\).
Step 4: Calculate and check beta. Beta is \(15/10=1.50\), equivalently \(60/40=1.50\). Each asset deviation is \(1.5\) times its paired market deviation, so the calculated slope matches the data exactly.
Units, Interpretation, and Limits
If both return series are multiplied by the same conversion factor, beta is unchanged. For example, converting both percentage-point returns to decimal returns divides each by 100. Covariance and market variance are then each divided by \(100^2\), so their ratio is unchanged. If only one series is converted, the ratio changes by the corresponding scale factor; consistent units are essential.
The arithmetic requires matched observations and positive sample market variance. It does not require a normal distribution to calculate the descriptive sample beta. However, the estimate describes the selected historical observations and chosen market proxy. It should not be interpreted as a causal effect or as a certainty about future returns.
- Positive beta: sample asset and market deviations tend to have the same sign.
- Negative beta: the centered returns tend to move in opposite directions.
- Beta near zero: the sample linear co-movement relative to market variance is small; this does not mean the asset has no risk.
- Beta greater than one: the estimated positive slope is greater than one in the stated return units.
A useful final audit is to check the means, the centered cross-product total, the centered market squared total, and the sign of the result separately. A plausible answer should follow from the paired deviations, not merely from the raw return levels.
Check Your Understanding
For each problem, use matched returns as stated and calculate beta from centered deviations.
- Over three periods, market returns are \(1,3,5\) percentage points and asset returns are \(0,2,4\) percentage points. Find the sample covariance, market variance, and beta. (Answer: sample covariance \(=4\), sample market variance \(=4\), beta \(=1.00\))
- Over four periods, market returns are \(2,4,6,8\) percentage points and asset returns are \(3,4,5,6\) percentage points. Calculate the beta and state whether the centered association is positive or negative. (Answer: beta \(=0.50\), positive)
- Market returns in decimal form are \(-0.02,0,0.02,0.04\), and asset returns are \(-0.01,0,0.01,0.02\). Calculate beta, and explain whether converting both series to percentage points would change it. (Answer: beta \(=0.50\); converting both series to percentage points does not change beta)
- For five periods, market returns are \(1,2,3,4,5\) percentage points and asset returns are \(4,1,3,2,5\) percentage points. Calculate the centered cross-product total, market squared-deviation total, and beta. (Answer: cross-product total \(=3\), market squared-deviation total \(=10\), beta \(=0.30\))
- Over four matched periods, market returns are \(0,2,4,6\) percentage points and asset returns are \(5,1,3,7\) percentage points. Find the sample covariance, sample market variance, and beta. (Answer: sample covariance \(=8/3\), sample market variance \(=20/3\), beta \(=0.40\))