Crypto assets can trade with little relationship to stock markets for months and then move in lockstep with them during a decline. The pattern is a feature of how stress propagates rather than a change in what the assets are.
Calm periods are driven by internal factors
When conditions are stable, crypto prices respond to sector-specific developments: protocol upgrades, flows between assets, regulatory news and liquidity conditions on exchanges.
These have little connection to corporate earnings or interest rate expectations, so measured correlation with equity indices is low.
That period is when the diversification argument looks strongest, because the assets genuinely are responding to different information.
Stress transmits through positions, not fundamentals
Many participants hold both crypto and conventional risk assets, often with borrowed money against a combined pool of collateral.
A loss anywhere in that pool triggers a demand for more collateral, and meeting it means selling something.
What gets sold is what can be sold quickly at an acceptable price, which frequently means crypto because it trades continuously and settles immediately.
Continuous trading makes it the first outlet
Crypto markets run at all hours, so a shock arriving when equity markets are closed is expressed in crypto first.
Participants needing to reduce exposure act where they can act, which concentrates initial selling into the market that happens to be open.
This produces moves that look like a reaction to the equity market before the equity market has traded, which is a sequencing effect rather than a leading indicator.
Liquidity conditions affect everything at once
Both crypto and equities are sensitive to the general availability and cost of funding, since leverage in both depends on it.
When funding tightens, positions are reduced across the board, and assets with no cash flows are usually reduced first because there is no valuation anchor to argue against selling.
The shared driver is the funding environment rather than any economic link between the assets themselves.
Correlation is unstable, so averages mislead
A long-run correlation figure blends calm periods of near independence with brief episodes of very tight co-movement.
The average describes neither state accurately, and it understates exactly the behaviour that matters most to someone holding both.
The useful question is not how correlated the assets are on average but how they behave together when positions are being reduced, which is when a diversification assumption is actually tested.