Crypto has been sold as diversification and has often traded as leveraged risk-on. Both descriptions have supporting periods, which is exactly the problem with quoting one number.
Why a single figure misleads
Correlation is computed over a window, and crypto’s relationship with equities shifts between regimes rather than drifting gently. Average a period of tight co-movement with a period of decoupling and you get a middling figure that describes neither — the same failure as blending funding across venues.
Worse, correlation is not stable through stress. The pattern across risk assets generally is that correlations rise toward one exactly when diversification is being relied on, and crypto is no exception.
The asymmetry worth knowing
The co-movement tends to be strongest on sharp risk-off days and weakest during ordinary drift. So crypto can look diversifying for months and behave like a high-beta equity position on the one day that matters.
Any portfolio argument that relies on the calm-period correlation is relying on a number that stops applying under stress.
What actually drives the co-movement
Shared exposure to liquidity conditions and risk appetite — the mechanism in crypto and the rates cycle — plus a shared marginal buyer. When the same participants allocate to both, both move on the same allocation decisions.
That is a more useful model than a correlation coefficient, because it tells you when to expect co-movement rather than how much there was on average.
Measuring it honestly
Rolling windows rather than one number, so the regime shifts are visible. And the clock problem: crypto trades continuously while equities do not, so daily correlation depends on which closing times you align — a crypto day and an equity day are not the same interval, and pretending otherwise is the same class of error as annualising crypto on 252 days.