Stablecoin Liquidity as a Crypto-Native Regime Signal: State-Dependent Density Forecasts for BTC, ETH, and SOL

Abstract This study examines whether stablecoin liquidity, network congestion, and macro-financial stress predicts major cryptoasset returns in a regime-dependent manner. Using daily data for BTC, ETH, and SOL from April 15, 2020, to March 17, 2026, the analysis estimates homogeneous and non-homogeneous hidden Markov models to capture shifts across latent low-, medium-, and high-volatility states. The paper’s central contribution is to reframe stablecoin activity not as a constant linear return predictor, but as a crypto-native signal of changing market regimes. The final model selection supports BTC NHHM (3), ETH HHM (3), and SOL HHM (3) as the preferred specifications. These models rank first by average log score and continuous ranked probability score for their respective assets, indicating superior density-forecast performance, although mean squared forecast error gains are not uniform. The coefficient evidence is strongest for BTC and ETH. In medium-volatility regimes, stablecoin volume growth is positively associated with returns, whereas stablecoin transaction-count growth is negatively associated with returns, suggesting that different dimensions of stablecoin activity capture distinct liquidity and stress channels. BTC also exhibits stronger macro-stress sensitivity in the high-volatility regime. SOL shows clear regime separation and density-forecast gains, but its mean-equation coefficients are less precise. Overall, the findings show that stablecoin liquidity helps characterize latent crypto-market risk states and improves distributional forecasting, creating value for researchers studying crypto-market transmission and for investors managing dynamic volatility exposure. Keywords: stablecoins; cryptocurrency returns; hidden Markov models; regime switching; density forecasting; Bitcoin; Ethereum; Solana