Abstract This paper investigates how different sources of market volatility affect impermanent loss (IL) in decentralized liquidity provision. Using daily data from Uniswap V2, our analysis distinguishes between two types of liquidity pools: volatile-volatile (V–V) pairs, composed of two risky assets, and volatile-stable (V–S) pairs, which include a stablecoin. The empirical approach combines quantile regression (QRM), autoregressive models with exogenous inputs (ARX), principal component analysis (PCA), and dominance decomposition to assess volatility transmission of both crypto-native and traditional financial volatility indices. Our results indicate that IL responds asymmetrically to market volatility and reveals an autoregressive structure, highlighting the importance of considering cumulative IL exposure when developing hedging strategies for liquidity providers (LPs). Our findings also reveal that V-S pools have a buffering effect, mitigating the conversion of external volatility into IL despite the same exposure to shocks. We integrate these findings using a structural sensitivity map that classifies the relevance of each market volatility index to IL risk. The study contributes to the understanding of the dynamics of IL risk, suggesting that it is a controllable risk with variables endogenous to Automated Market Makers and provides rationales for LPs managing risk exposure in volatile environments.
Monte et al. (2026) studied this question.