Original research in dynamic correlation and skewness modeling

Peer-reviewed methodology, practitioner-tested tools, and ongoing inquiry into how multi-asset portfolios behave when markets turn.

Our research program sits at the intersection of academic rigor and institutional relevance. Every paper, model, and dataset we publish is designed to survive both peer review and production deployment.

A whitepaper examining how correlation skewness intensifies during high-volatility periods and what that means for portfolio construction.

Regime-Dependent Correlation Asymmetries in Equity Markets

This paper documents a persistent pattern: equity correlations become more asymmetric precisely when volatility rises, yet standard risk tools assume stability.

We analyze daily returns across developed and emerging equity markets from 2000 through 2024, identifying a structural break in correlation dynamics during stress periods that Gaussian frameworks systematically miss.

The skew-normal copula approach we introduce captures this asymmetry without imposing the computational burden of full non-parametric estimation, making it feasible for real-time risk systems.

Empirical results show that portfolios optimized under symmetric correlation assumptions exhibit materially higher drawdown frequency than those accounting for regime-dependent skewness.

We provide full model specifications, estimation code, and replication datasets to enable independent validation by practitioners and academics.

The framework has been adopted by risk departments at three major financial institutions and is currently under review for inclusion in industry stress-testing guidelines.

Dynamic Skewness Models for Multi-Asset Portfolio Construction

Research brief exploring how time-varying correlation asymmetries reshape risk parity and diversification assumptions across asset classes.

Risk parity assumes balanced risk contributions. Dynamic skewness breaks that balance in ways static models cannot detect.

The 2020 Market Dislocation: Correlation Skewness and Early Warning Signals

The March 2020 drawdown provided a natural experiment in correlation asymmetry. Our post-hoc analysis reveals signals that were visible weeks before the cascade accelerated.

High-frequency estimation

We develop estimators that operate on intraday return data, capturing correlation dynamics at timescales invisible to daily or monthly frameworks. The approach uses realized covariance decompositions that separate symmetric and asymmetric components without assuming a specific return distribution.

Regime-switching frameworks

Our models explicitly distinguish between normal and stress correlation regimes, estimating transition probabilities and regime-specific skewness parameters jointly rather than imposing them exogenously.

Copula-based tail dependence

We employ skew-normal and skew-t copulas to model asymmetric tail dependence directly, avoiding the mean-variance simplifications that obscure left-tail risk in standard approaches.

Validation protocols

Every estimator undergoes out-of-sample backtesting, synthetic scenario analysis, and implementation stress-testing before publication. We document failure modes explicitly rather than burying them in appendices.

Correlation matrix heatmap visualization

Stay current with our research

Cookie preferences

We use cookies to improve your experience

Your privacy matters

This website uses cookies and similar technologies to enhance functionality, analyze traffic, and support our research communications. You can choose which categories you accept. Essential cookies cannot be disabled as they are required for the website to function properly.

Essential cookies

Required for core website functionality, security, and accessibility. These cannot be disabled.

Analytics cookies

Help us understand how visitors interact with our content and improve the website experience.

For more information about how we handle your data, please review our privacy practices and cookie policy.