The mistake was not owning bonds. It was treating the relationship between stocks and bonds as a law of nature instead of a number that moves. That number is asset correlation, and it decides whether diversification works or only looks like it works.
Think of two lifeboats. If they hang from separate ships, you are protected. If both are tied to the same ship, you own two lifeboats and one point of failure. Asset correlation tells you which situation you are in.
This guide shows you what asset correlation is, how to calculate it, how it changes portfolio risk, and how to test whether your own holdings are truly independent. Examples use US dollars and broad global indices, so you can swap in your own local equivalents.
Actual screenshot of Investopedia's correlation primer (updated 7 May 2026), captured 22 October 2026.
What Is Asset Correlation?
Asset correlation measures how the returns of two investments move relative to each other. The standard tool is the Pearson correlation coefficient, written as r or rho. It always lands between -1 and +1, and it is calculated from returns over matching periods, not from raw prices.
A reading of +1 means two assets move in perfect lockstep. A reading of 0 means no linear relationship. A reading of -1 means they move as mirror images. In real portfolios, most pairs sit somewhere between -0.3 and +0.9.
The math divides the covariance of the two return series by the product of their standard deviations. Covariance tells you whether the assets tend to be above or below average together. Dividing by volatility rescales the answer so you can compare very different pairs on one scale. For a refresher on the basics, see the Investopedia correlation guide.
These ranges are typical, not fixed. Every pair shifts with the period you measure, which is the core idea behind the rest of this article.
Why Asset Correlation Matters for Portfolio Risk
Portfolio volatility is not the weighted average of each holding's volatility. It also depends on correlation. For two assets, the variance equals each weight squared times its own variance, plus two times the weights times the correlation times both volatilities. The lower the correlation, the smaller that last term, and the calmer the portfolio.
Here is an illustration using assumptions, not forecasts. Take a 60/40 mix where stocks have 16% annual volatility and bonds have 6%. Only the correlation changes between rows.
2.1 points of annual volatility separate the same 60/40 mix at zero correlation (9.9%) from the same mix at perfect correlation (12.0%). Nothing about the holdings changed. Only the relationship did.
History shows both outcomes. In 2008 the S&P 500 returned about -37% while the US Aggregate bond index returned about +5.2%, so bonds did their job. In 2022 the S&P 500 fell about 18.1% and the same bond index fell about 13.0%, so a 60/40 portfolio lost roughly 16%. Rising inflation and rising interest rates hurt both sides at once.
-18.1% and -13.0% is the 2022 stock and bond result, a reminder that a negative correlation is a condition, not a promise. For the broader picture, see our guide to spreading risk with diversification.
Actual screenshot of FRED's 10-year Treasury yield, 5.12% on 20 October 2026, captured 22 October 2026.
Rates matter because they sit behind the stock-bond link. With the 10-year yield at 5.12% on 20 October 2026 and the S&P 500 closing at 7,757.17 on 21 October 2026 (both from FRED), investors earn far more from bonds than a decade ago. That changes the trade-off, but it does not guarantee the sign of the correlation.
A quick way to feel the effect: imagine you hold only two assets, and in a bad month stocks drop 8%. If bonds are negatively correlated, they might rise 2% and cut your portfolio loss. If they are positively correlated, they might drop 3% and deepen it. Same assets, same month, very different outcome. This is why professional risk teams track the correlation matrix as closely as they track individual volatility.
How to Measure and Use Asset Correlation
You do not need a quant desk. A spreadsheet or a few lines of code are enough. Follow these steps for any set of holdings.
Step 1: Pull Matching Return Series
Download at least 36 months of monthly total returns for each asset, using the same dates and the same currency. Monthly data smooths out the timing gaps between markets in different time zones. Daily data over short windows is noisy and can mislead.
Step 2: Build the Correlation Matrix
In a spreadsheet, use the CORREL function on two columns of returns. In Python, compute percentage changes on a price table and call the corr method. Either way you get a grid where every cell is one pair, and the diagonal is always 1.
Step 3: Run a Rolling Window
Repeat the calculation over a 36-month window that slides forward one month at a time, then plot the line. A pair that sits at -0.3 for years and jumps to +0.5 in a single stress period is telling you its diversification benefit is unstable.
Step 4: Test the Bad Months Only
Filter for the worst 10% of months for stocks and calculate correlation on those months alone. This downside correlation is usually higher than the full-sample figure, and it is the number that matters when you are actually losing money.
Step 5: Act on What You Find
If two holdings correlate above 0.8, size them as one position. Look for assets that respond to different drivers, such as growth, inflation, and interest rates. Then keep the mix on target with a clear portfolio rebalancing strategy. If you want to equalize risk instead of dollars, see risk parity explained.
A worked example. Suppose you compare a global stock fund and a government bond fund over 36 months and the CORREL result is 0.35. That is a weak positive link, so the pair still diversifies. Now rerun it on only the 4 worst months for stocks and the figure rises to 0.62. The gap between 0.35 and 0.62 is the hidden cost of assuming calm-market behavior will continue in a sell-off, and it is the number to plan around.
Keep a short log of each reading, the window used, and the date. Over a few years that log becomes your own evidence about which relationships are stable and which are not, and it stops you from relying on a headline statistic from a research note that may have used a different period.
Real Examples: Three Stress Tests
2008 financial crisis. Equities collapsed and government bonds rallied as investors fled to safety. The stock-bond correlation was negative and diversification worked. Many other pairs, such as US and foreign stocks, moved closer together than their long-run averages.
March 2020. The S&P 500 fell about 34% in just 23 trading days, from 19 February to 23 March. Investors sold almost everything to raise cash, and assets that normally diverge fell together for a stretch. Correlations spiked toward 1, then eased once central banks stepped in.
2022 inflation shock. Central banks raised rates quickly. Stocks fell about 18.1% and bonds about 13.0%. The positive correlation was not a glitch. It was the predictable result of inflation driving both.
What to take from this: each crisis had a different driver, so the safe haven changed each time. Check the rolling reading for your own pairs rather than trusting last decade's average.
Currency adds another layer. A foreign stock fund priced in a different currency can show a lower or higher correlation to your home market depending on whether the currency is hedged. Always compute correlations in the currency you actually spend, otherwise the number describes someone else's portfolio.
Common Mistakes With Asset Correlation
Mistake 1: Treating One Number as Permanent
A single correlation figure from a long history hides large swings. Always look at the rolling line and the crisis windows, not just the headline average.
Mistake 2: Correlating Prices Instead of Returns
Two assets that both trend upward will show a high price correlation even if their monthly moves are unrelated. This is called spurious correlation. Always use returns.
Mistake 3: Counting Tickers Instead of Drivers
Owning ten funds feels diversified. If most of them hold the same large technology stocks, you own one bet ten times. Check holdings overlap, not fund names.
Mistake 4: Ignoring Crisis Correlation
Correlations tend to rise when markets fall, which is exactly when you need diversification most. Measure downside correlation and stress-test your plan with a maximum drawdown estimate.
Mistake 5: Chasing Low Correlation Alone
A low correlation is worthless if the asset has poor expected return or high costs. Judge each addition on return, cost, and correlation together. The Sharpe ratio helps you compare them.
Actual screenshot of the SEC Investor.gov definition of diversification, captured 22 October 2026.
The SEC's plain-language definition on Investor.gov summarizes the goal: if one investment loses money, others may make up for it. Correlation is how you check whether that is actually true.
Frequently Asked Questions
What is a good correlation for diversification?
Readings below 0.4 give meaningful benefit, and negative correlation assets give the most. Remember that the benefit shrinks fast above 0.8, where two holdings behave almost as one.
Is a correlation of zero the goal?
Not necessarily. Zero means no linear link, but the asset still needs a positive expected return to earn its place. A slightly negative or low positive reading with solid returns is often better than a perfect zero with weak returns.
How often should I recalculate the correlation between stocks and bonds?
Review the rolling 36-month figure at least once a year, and again after any major shift in inflation or interest rates. Those shifts are what flipped the sign in 2022.
Which negative correlation assets are worth considering?
Government bonds, gold, and certain trend-following strategies have shown negative or low correlation to stocks in some periods. None of them has done so in every period, so treat each as a tool to test, not a guarantee.
Does asset correlation work for crypto and other newer assets?
It works the same way, but the history is short and the relationships have shifted quickly. Treat any reading from fewer than 36 months of data as a rough guide, and widen your safety margin accordingly.
Key Takeaways
- Asset correlation is a number between -1 and +1 that measures how two return series move together.
- Portfolio volatility falls as correlation falls, even when the holdings stay the same.
- Stocks and bonds fell together in 2022, so a negative correlation is a condition, not a promise.
- Use monthly returns, a 36-month rolling window, and a crisis-only test.
- Count drivers of risk, not tickers, and size highly correlated holdings as one position.
- Return to the lifeboat image: two boats on the same ship are one point of failure.
What to Watch Next
- v Does the rolling 36-month stock-bond correlation stay positive while the 10-year yield is above 5%?
- v Do equity pairs tighten again if the S&P 500 drops from its 7,757.17 level?
- v Does gold or any other hedge keep a low correlation in the next sharp sell-off?
- v Does your own portfolio's downside correlation exceed its full-period correlation?
References
This article is for education only and is not financial advice.