A0287
Title: Diagnosing cryptocurrency security vulnerability through time series decomposition
Authors: Julien Chevallier - University Paris 8 (LED) (France) [presenting]
Abstract: Cryptocurrency markets have experienced repeated systemic breakdowns over the past decade. Notable examples include the Mt. Gox collapses (2011-2014), the COVID-19-driven "312" flash crash, the "519" crash in 2021 following environmental concerns, the crypto winter of 2017-2018, the collapse of the Terra/Luna ecosystem and the FTX exchange in 2022, as well as the combined impact of Grayscale ETF outflows and tariff conflicts in early 2025. These episodes were triggered by factors such as regulatory shocks, exchange failures, excessive leverage, macroeconomic instability, and automated liquidations, often resulting in billions in losses within hours and trillions in erased market capitalization. Mathematical techniques are proposed for diagnosing such market failures through the decomposition of time series data. The framework integrates nonlinear signal analysis and regime-shift detection to identify critical transitions before major breakdowns occur. Across security vulnerability episodes, the method highlights consistent patterns in volatility structure, trading flow distortions, and abnormal amplitude shifts preceding failure events. These patterns provide insight into latent instability and offer potential as early diagnostic indicators. A novel lens is contributed for understanding crypto market fragility, offering tools to dissect chaotic behaviors not captured by conventional risk metrics.
