Correlation Breakdown: Why Your 'Uncorrelated' Assets Suddenly Move Together in a Crash
6 min read
Loading...
6 min read
You hold two assets that barely move together. Maybe it's an IDX bank stock next to a global index fund, or gold sitting alongside your equity portfolio. In calm markets, that's exactly what you want: when one falls, the other holds steady, so your total portfolio doesn't fall as far. Then a real crash hits, and both of them fall together anyway, on the same day, for the same reason. That isn't a coincidence, and it isn't your tracking app getting the math wrong. It has a name: correlation breakdown, and it tends to show up at the exact moment diversification is supposed to protect you the most.
Correlation is a number between -1 and +1 that describes how closely two assets' returns move together. Near +1, they rise and fall together. Near 0, their movements have no consistent relationship. Near -1, one tends to rise when the other falls. NetWort has covered the basics of correlation and diversification in more depth in Two Stocks That Rise and Fall Together Aren't Actually Diversifying You, which is worth reading first if the concept itself is new.
Correlation breakdown is a specific pattern inside that idea: two assets can hold a low correlation for months, then see that correlation jump sharply higher during a market downturn specifically, not during a rally. This isn't a vague market saying. Francois Longin and Bruno Solnik's 2001 study in The Journal of Finance, "Extreme Correlation of International Equity Markets" (Vol. 56, No. 2, pp. 649-676), tested exactly this using extreme value theory on international equity returns. They found that the statistical behavior of returns in the negative tail, the worst days, differs from the positive tail, the best days: markets that look only loosely linked most of the time can show a much stronger relationship specifically when everything is falling.
The clearest attributable figures come from Two Sigma's research note "Asset Class Correlations: Return to Normalcy?", which tracked monthly returns across 74 securities: 42 country equity indices, 16 government bond markets, 9 currencies and 7 commodities.
| Period | Global equity-to-equity correlation | Equity-to-commodity correlation |
|---|---|---|
| Before the 2008 crisis | roughly 0.40 | roughly 0.0 |
| During the 2008 crisis | roughly 0.70 | roughly 0.6 |
Source: Two Sigma, "Asset Class Correlations: Return to Normalcy?"
Country equity markets that historically moved somewhat independently of each other nearly doubled their pairwise correlation once the crisis hit. Commodities, which had shown essentially no consistent relationship with equities before 2008, started moving with them at a correlation of roughly 0.6, a level that would normally suggest two assets share a real economic link, not two unrelated markets.
The 2020 COVID crash is the cleanest recent example, because it hit fast enough to see the whole pattern inside a few weeks. The S&P 500 fell 33.9%, from a closing high of 3,386.15 on 19 February 2020 to a closing low of 2,237.40 on 23 March 2020, a decline that took the 2008 crisis roughly 17 months to reach but took COVID-19 about five weeks.
Gold is the textbook example of a "safe haven," an asset that is supposed to hold up, or even rise, when stocks are falling. During the same stretch, gold fell too, by roughly 12% over the course of March 2020, according to World Gold Council research on the period. The World Gold Council's own explanation wasn't that gold's long-term case had changed. It was liquidity: as stocks, credit and other holdings fell hard, investors facing margin calls and redemption requests sold whatever was easiest to sell quickly, and gold is one of the most liquid assets in the world. It got sold to raise cash for other losses, not because anyone stopped believing in it.
The mechanism isn't mysterious once you see it from the seller's side rather than the asset's side. A crash usually starts with one shock: a credit freeze, a margin call wave, a rush of fund redemptions. Investors and funds holding many different assets at once don't sell just the asset connected to that shock. They sell whatever is liquid enough to sell fast, across their entire book, to raise cash or meet a margin call. That single, shared reason to sell is what pushes many different assets down together, regardless of how differently they behaved the month before.
It's worth being honest that not every researcher reads the 2008 and 2020 numbers the same way. Kristin Forbes and Roberto Rigobon's 2002 paper in The Journal of Finance, "No Contagion, Only Interdependence: Measuring Stock Market Comovements," argued that simple correlation coefficients are biased upward whenever volatility itself rises, so part of an apparent correlation "spike" during a crisis can be a statistical side effect of choppier markets rather than proof that the underlying relationship truly changed. For an investor actually holding both assets, the distinction matters less than it sounds: whether the cause is a real shift or a volatility-driven measurement effect, the practical experience is the same. Both assets fell together, right when one of them was supposed to hold the line.
This isn't only a 2008-and-2020 story. NetWort's own Bitcoin's Correlation With the Nasdaq Swung From -0.68 to +0.72 in Two Weeks documented the same mechanism happening in February 2026: bitcoin's 30-day rolling correlation with the Nasdaq moved from -0.68 to +0.72 in about two weeks, during a tech-driven selloff. A pair that had been moving in opposite directions started moving in lockstep the moment a shared shock hit both of them. You can check Bitcoin's own asset page for its current correlation with the S&P 500 directly, rather than assuming it holds at whatever level it showed last month.
The lesson isn't to stop diversifying. It's to know that a low correlation measured in calm markets is not a guarantee, it's a snapshot that can move hard exactly when a shock hits. NetWort's Portfolio Health view shows your portfolio's real beta, volatility and maximum drawdown calculated from your actual holdings, so you can see how much of your total risk is concentrated in positions that tend to answer to the same shocks, rather than assuming it from how the assets have behaved recently.