Crypto Market Cascades Challenge Existing Prediction Models

Recent research has identified seven liquidation cascades in the crypto market that challenge existing prediction models. These events, which occurred between 2022 and 2025, demonstrate that most frameworks built on gradual pre-crash 'critical slowing down' are not reliable across regimes.

The study analyzed minute-by-minute mechanics inside the cascades and found that onset dynamics are abrupt and scale-robust rather than classically critical. In five of seven cascades, price-based critical slowing down was observed, but it failed in the two tariff-shock events. This suggests a two-type classification of endogenous build-up versus exogenous shock.

The research also highlights the importance of microstructure and exchange design as first-order drivers of realized outcomes, not just investor positioning or macro news flow. The study's findings have significant implications for risk models and trading desks, emphasizing the need to treat regime identification as a first step and prioritize forward-looking liquidity risk measures.