A negative (or inverse) correlation describes a relationship between two variables in which they tend to move in opposite directions: when one rises, the other tends to fall, and vice versa. In statistics a perfectly negative linear relationship is represented by a correlation coefficient of −1.0; zero means no linear relationship and +1.0 means a perfect positive linear relationship.
Key Takeaways
– Negative correlation means two series typically move opposite one another.
– The Pearson correlation coefficient (r) ranges from −1 to +1 and is the usual numeric measure.
– Investors use negatively correlated assets to lower portfolio volatility; however, correlations change over time and can break down in stress periods.
– Outliers, sampling choices (time window, frequency), and non‑linear relationships can distort correlation estimates.
How Negative Correlation Works
Correlation measures linear co‑movement. If returns of Asset A and Asset B have a negative correlation, positive returns in A tend to be accompanied by negative returns in B. This does not imply causation: correlation is descriptive, not explanatory. Also, most real‑world correlations are imperfect (between −1 and +1).
Negative Correlation and the Correlation Coefficient
The standard statistic is the Pearson correlation coefficient:
r = cov(X, Y) / (σX · σY)
where cov(X, Y) is the sample covariance and σX, σY are sample standard deviations. Interpretation:
– r ≈ −1: nearly perfect inverse linear relationship
– r ≈ 0: no linear relationship
– r ≈ +1: nearly perfect positive linear relationship
Important
– Correlation is sensitive to sample period and return frequency (daily vs monthly returns can give different r).
– Correlation measures linear relationship only. Non‑linear dependence can be missed.
– Correlations are not constant; they often rise during market stress.
Watching for Outliers
One or a few extreme observations can materially change r. Options to reduce outlier influence:
– Inspect scatterplots and time series before trusting a single r value.
– Use robust techniques (winsorize, truncate, or use rank correlations such as Spearman).
– Compute rolling correlations to see how relationships change over time.
Negative Correlation and Investing
In investing, negative correlation is useful because assets that move oppositely can offset each other’s losses, reducing portfolio volatility and smoothing returns. Classic examples often cited (but not guaranteed) include:
– Stocks vs. high‑quality government bonds (often mildly negative over long periods).
– Cash or short‑term Treasuries vs. risky assets in crisis (liquidity and safety flows can be inverse).
– Some currencies vs. domestic equities in certain regimes.
Note: these relationships can change, especially during crises when correlations often converge upward.
Negative Correlation and Portfolio Diversification
Diversification works by combining assets whose returns are not perfectly positively correlated. If components have low or negative pairwise correlations, the portfolio variance can be substantially below a weighted average of individual variances. Strategic asset allocation seeks to balance correlations across large portfolios.
Fast Fact
Perfectly negative correlation (r = −1) is rare in real markets; useful negative correlations are typically partial (e.g., r between −0.2 and −0.6).
Example: Constructing a Portfolio With Negative Correlations (Illustrative)
Suppose you want a simple 3‑asset illustrative portfolio:
– 60% equities (Stock ETF)
– 30% bonds (Bond ETF)
– 10% gold (Gold ETF)
Assume (illustrative) pairwise correlations:
– Equity–Bond = −0.20
– Equity–Gold = 0.00
– Bond–Gold = +0.10
Calculate the weighted pair contributions:
w_eq·w_bond·ρ_eq,bond = 0.6×0.3×(−0.20) = −0.036
w_eq·w_gold·ρ_eq,gold = 0.6×0.1×0.00 = 0.000
w_bond·w_gold·ρ_bond,gold = 0.3×0.1×0.10 = 0.003
Sum = −0.036 + 0 + 0.003 = −0.033
Average pairwise correlation (weighted by pairwise weight sum):
pair weight sum = 0.6×0.3 + 0.6×0.1 + 0.3×0.1 = 0.27
weighted average ρ ≈ −0.033 / 0.27 ≈ −0.12
Interpretation: the portfolio shows a modest net negative average pairwise correlation, which helps reduce variance versus a portfolio of perfectly positively correlated assets. This is illustrative; use actual return histories and volatilities for real decisions.
How to Calculate the Weighted Average Correlation (Practical)
A practical, commonly used approach to get a portfolio‑level sense of correlation:
1. Collect return series for the assets at the chosen frequency (e.g., monthly).
2. Compute the correlation matrix (pairwise ρij) and the weight vector w.
3. Compute the weighted sum of pairwise correlations: S = sum_{i
– Negative correlation: move opposite (r < 0). - Zero correlation: no linear relation (r ≈ 0). - Perfect positive/negative (r = ±1): theoretical extremes rarely observed. - Linear (Pearson) vs rank (Spearman, Kendall) vs other dependence measures (copulas). Is Negative Correlation Better Than Positive Correlation?
Neither is intrinsically “better” — it depends on objectives:
- If your goal is to reduce portfolio volatility, assets that are negatively or weakly correlated with your risky holdings are valuable. - If you seek exposure to a particular return source (e.g., equities), positively correlated assets may amplify returns but increase risk. - Effective portfolio construction balances return objectives, risk tolerance, liquidity, and correlations — not simply chasing negative r. Practical Steps Checklist (Summary)
1. Define investment horizon and data frequency. 2. Gather historical return data for candidate assets. 3. Clean data, check for outliers and structural breaks. 4. Compute correlation and covariance matrices; visualize. 5. Compute weighted average correlations and portfolio variance (use full covariance matrix for exact risk). 6. Build candidate allocations; simulate/backtest with rolling windows. 7. Stress test allocations in extreme scenarios and crisis periods. 8. Use robust estimators when data are noisy; avoid overfitting to past correlations. 9. Monitor and rebalance; update estimates as new information arrives. The Bottom Line
Negative correlation is a powerful concept for diversification and risk management: assets that move oppositely can smooth portfolio returns and reduce volatility. But correlations are estimates, sensitive to sample choices, outliers, and market regimes. Use proper data handling, rolling analysis, robust statistics, and full covariance‑based risk calculations rather than relying on a single pairwise correlation number. Always combine correlation analysis with a clear investment plan, risk controls, and stress‑testing. Source
Material and definitions adapted from Investopedia — Negative Correlation .