Skip to Main Content

Risk & Quant

Correlation

Correlation is a unitless measure, from −1 to +1, of the strength and direction of the linear relationship between two series, such as two assets’ returns.
Also Called
  • Correlation Coefficient
  • Pearson Correlation
  • r
By InsiderFinance TeamUpdated Sources & Methodology

Change Correlation and Compare Portfolio Volatility

Hold 60% equities at 18% volatility and 40% bonds at 6%. At correlation −0.20, portfolio volatility is 10.58%.
CorrelationCross TermVarianceVolatility
1.000.0051840.0174213.20%
0.500.0025920.0148312.18%
0.0000.0122411.06%
−0.20−0.0010370.0112010.58%
−1.00−0.0051840.0070568.40%
Hold 60% equities at 18% volatility and 40% bonds at 6%. At correlation −0.20, portfolio volatility is 10.58%.
On This Page
  1. What is correlation?
  2. How is correlation calculated?
  3. Correlation example: a 60/40 portfolio
  4. Correlation, beta and hedge ratios
  5. Implied correlation and dispersion trades
  6. Limitations of correlation
  7. Bottom line
  8. FAQ

Key Takeaways

  • Correlation scales covariance by both standard deviations, producing a unitless number between −1 and +1.
  • For portfolio-return risk, calculate correlation from matched return observations. Price-level relationships answer a different statistical question.
  • With two positive weights and fixed individual volatilities, lower correlation reduces portfolio variance. Opposite-signed weights reverse that cross-term relationship.
  • Pearson correlation measures linear association. Inspect the paired observations to understand the relationship it summarizes.
  • Compare correlations across the selected windows and market conditions to see how the portfolio’s co-movement changes.
  • Cboe derives implied correlation gauges from index and component option volatilities. Use the selected index’s construction and horizon when interpreting the reading.

What is correlation?

Correlation is a measure of the strength and direction of the linear relationship between two series, on a scale from −1 to +1. Pearson correlation divides the covariance of the two series by the product of their standard deviations, so the result has no units. In finance, return correlation shows how two assets move together, and it shapes how much combining them can reduce portfolio volatility.

With nonzero dispersion in both series, the result lies between −1 and +1. Positive values describe an upward fitted relationship; negative values describe a downward one. The comparison concerns deviations from each series’ mean. It does not count how often both raw returns have the same sign.

How is correlation calculated?

For portfolio-return risk, use returns matched by calendar and observation interval. Calculate covariance and standard deviations using one consistent divisor convention.

Dividing covariance by the two standard deviations removes the units. Expressing the same returns as percentages or decimals therefore leaves correlation unchanged.

Zero correlation means zero linear association in this calculation. It does not establish independence. For example, a symmetric variable and its squared value can be dependent while having zero covariance.

Formula

r(x, y) = Cov(x, y) / (StdDev(x) × StdDev(y)); two-asset portfolio variance = wA² × σA² + wB² × σB² + 2 × wA × wB × r × σA × σB
  • Use covariance and both standard deviations from the same paired observations and divisor convention. Both standard deviations must be nonzero.
  • StdDev is the standard deviation of the same return series used in the covariance.
  • wA and wB are portfolio weights, and σA and σB are the assets’ return standard deviations.
  • r is unitless, so the same value applies whether returns are expressed in decimals or percent.

Correlation example: a 60/40 portfolio

A portfolio holds 60% in an equity sleeve with 18% annualized return volatility and 40% in a bond sleeve with 6% annualized volatility. The realized correlation between the two return series is −0.20.

  • Equity term: 0.60² × 0.18² = 0.01166
  • Bond term: 0.40² × 0.06² = 0.000576
  • Cross term: 2 × 0.60 × 0.40 × (−0.20) × 0.18 × 0.06 = −0.001037
  • Portfolio variance = 0.01166 + 0.000576 − 0.001037 = 0.01120, giving volatility of 10.58%.
  • Same weights at correlation +1.00: 0.60 × 18% + 0.40 × 6% = 13.20% volatility

With weights and individual volatilities fixed, changing correlation from +1.00 to −0.20 reduces portfolio volatility from 13.20% to 10.58%, a 2.62 percentage-point reduction. Correlation changes only the cross term in this comparison.

The cross term is 0.005184 × correlation. At −0.20, it equals −0.001037. Total variance is 0.01120 and volatility is 10.58%.

At +1 correlation, volatility is 13.20%. The reduction is 2.62 percentage points. At −1 correlation, the unequal volatility-weighted legs leave 8.40% volatility.

For positive weights, reducing correlation lowers variance. For a long and a short weight, the sign of that effect reverses. Use the actual exposures when evaluating the relationship.

Correlation, beta and hedge ratios

Beta equals correlation times asset volatility divided by benchmark volatility. Correlation measures standardized association; beta supplies direction and scale against the benchmark.

For a return hedge, covariance with the hedge divided by hedge-return variance gives the minimum-variance hedge ratio. Position values and contract specifications convert the result into tradable notional.

A correlated pair also has a price spread and a return relationship. Examine the spread’s behavior separately when selecting a pairs-trading rule.

Implied correlation and dispersion trades

Cboe’s implied correlation gauges use index and component option volatilities. The selected index defines the component universe, expiry and moneyness convention.

Dispersion positions combine index and component volatility exposures. Assess the actual legs and changes in implied and realized inputs together. A lower correlation reading alone does not determine the trade’s total P&L.

Limitations of correlation

  • Plot the paired observations beside the coefficient. Correlation captures only linear association: a zero reading does not establish independence, and a strong nonlinear relationship can still produce a low value.
  • Compare readings across windows and intervals. The same pair can show different readings on daily and weekly returns, over different lookbacks and across market conditions.
  • Check how much a few extreme observations drive the estimate, since a single shock can change the reading.
  • Use beta for scale and business analysis for cause. Correlation measures neither: a high reading does not identify a common economic driver.
  • Stress-test diversification for a selloff. Shared selling or financing changes can raise co-movement among affected assets, so diversification estimates can weaken when they matter most.

Bottom line

Correlation is a unitless summary of how two return series move together linearly, and it drives the cross term in portfolio variance. Use it on matched returns, compare it across windows and market conditions, and pair it with volatilities and position weights before relying on it for diversification or hedging.

Clear answers

Frequently Asked Questions

Sources and Further Reading

  1. FINRA: Asset Allocation and Diversification
  2. Cboe: Implied Correlation Index
  3. Fidelity: 3 Steps to Diversify Your Investment Portfolio

How entries are sourced, calculated and updated: editorial methodology.

Continue learning