Crypto correlation regimes

Definition

Crypto’s correlation to traditional assets is a REGIME variable, not a parameter. Early samples showed near-zero correlation to equities and no loading on standard asset-pricing factors (Liu-Tsyvinski 2021) — the origin of the “uncorrelated diversifier” claim. Post-2020, as institutional participation grew, spillovers to equities rose materially (IMF 2022), and the 2022 tightening cycle ran bitcoin and the Nasdaq down together — high positive correlation exactly when diversification was needed. The platform treats any crypto correlation number as a dated, regime-conditional measurement (port-correlation-budgets).

How it works / structure

• The documented arc: pre-2020 near-independence (Liu-Tsyvinski’s sample), 2020-2021 rising co-movement with risk assets as the holder base institutionalized (IMF spillover evidence), 2022 strongly positive correlation through the liquidity contraction (regime-rate-environments), and post-ETP-era integration with equity-market flow machinery (crypto-etps) — each phase a different correlation regime with a structural explanation.
• The driver logic: correlation follows the MARGINAL HOLDER. A retail/crypto-native holder base produces idiosyncratic dynamics; leveraged institutions funding crypto and equities from the same balance sheet transmit common liquidity shocks (the IMF note’s channel) — which is why correlation rises in stress and eases in calm.
• Conditional structure: crypto-equity correlation is liquidity-regime-dependent — elevated in tightening and vol-spike regimes (regime-volatility), lower in calm — the general stressed-correlation pattern (risk-correlation-exposure) expressed at maximum amplitude.
• Within-class correlation: crypto assets correlate highly with each other, especially in drawdowns — a multi-coin book is closer to one position than its line count suggests (port-diversification-math).

When it applies

Portfolio construction with crypto sleeves (the correlation input must be the stressed value, not the full-sample average), diversification-claim evaluation (any “uncorrelated asset” thesis must date its evidence and state its regime), macro theses using crypto as a liquidity-sensitivity gauge, and hedging design — an equity hedge sized off calm-period crypto correlation underhedges the regime where it matters.

Risk profile & failure modes

• Full-sample averaging: blending the independent early era with the integrated era produces a correlation no regime actually exhibits — the classic estimation error here.
• Diversification that vanishes: allocations justified by early-era evidence carried 2022’s joint drawdown; the claim failed precisely at its test (port-correlation-budgets stress rows exist for this).
• Regime-call overconfidence: the integration arc is not destiny — correlation could re-decouple with holder-base or policy shifts; both directions are theses, neither is a fact.
• Within-class illusion: diversifying across coins while their pairwise stressed correlations approach one.

Evidence & limits

Early-sample independence is peer-reviewed (Liu-Tsyvinski 2021); the spillover increase is IMF-documented; the 2022 joint drawdown is public record. Forward correlation is not knowable — this entry supplies the regime frame and the measurement discipline, not a number. Correlation estimates on crypto inherit session-mismatch artifacts (crypto-sessions-24-7) — alignment choices change the estimate and must be pinned.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Bitcoin’s 90-day correlation to the Nasdaq exceeds 0.4 during the next sustained policy-tightening phase (liquidity-channel thesis)” — falsified by the conditional series.
• “Average pairwise 30-day correlation among the five largest coins exceeds 0.8 during the next 20%+ class drawdown (within-class convergence thesis)” — falsified by the episode’s correlation matrix.

Cross-references

• The budget it feeds: port-correlation-budgets, risk-correlation-exposure, port-diversification-math
• The regimes that drive it: regime-rate-environments, regime-volatility, macro-fed-balance-sheet
• Asset-class frame: ext-crypto; measurement hazards: crypto-sessions-24-7