Regime-Dependent Volatility Sensitivity

This study examines whether regime-dependent volatility sensitivity and its impact on trend-following profitability are universal across asset classes. Using a common sample of daily data for Bitcoin, Gold, and the S&P 500 from 2015 to 2026, we evaluate a standardized volatility-targeted EMA trend strategy under regime-conditional stress tests. To account for liquidity collapses, we implement dynamic transaction cost modeling, penalizing execution friction at 3x the baseline during high-volatility expansions. We document a structural divergence: net of stressed transaction costs, Bitcoin maintains a statistically invariant and robust performance profile across volatility regimes, whereas traditional assets suffer severe degradation. Statistical tests confirm that trend performance is highly regime-sensitive for traditional assets (p < 0.001). For Bitcoin, we fail to reject the null hypothesis (p = 0.195), and supporting Bayes Factor analysis confirms the equivalence of trend performance across volatility states. These findings provide empirical evidence consistent with the hypothesis that volatility is an asset-dependent signal rather than a homogeneous risk factor. Consequently, portfolio construction frameworks that mechanically de-leverage during volatility expansions— such as volatility targeting or risk parity—may be structurally misaligned for assets exhibiting positive volatility return dependence. We propose conditioning algorithmic exposure on asset-specific volatility characteristics rather than applying universal defensive filters.