On the Structure of Risk Contribution: A Leave-One-Out Decomposition into Inherent and Correlation Risk

Battle Count: 87

Highly relevant for quantitative trading and portfolio management. The ICD decomposition provides actionable diagnostics for position-level risk analysis, hedging effectiveness assessment, and stress-period attribution. It helps distinguish whether risk reduction requires exposure reduction (inherent risk) or diversification/offsetting (correlation risk). The framework is directly implementable in risk reporting systems and supports hierarchical attribution across asset classes, sectors, and strategies. Particularly useful for long-short hedge fund risk management and risk parity portfolio construction.

Moderate complexity. The core decomposition requires only standard portfolio weights, asset volatilities, and covariance estimates already available in most risk systems. The leave-one-out representation is algebraically straightforward. However, robust extensions (shrinkage estimators, random matrix filtering, asynchronous covariance, exponential decay weighting) add implementation complexity. The time-varying analysis requires rolling window computations. No specialized software or optimization solvers are needed beyond standard covariance estimation.

The paper provides complete mathematical derivations and formulas for the ICD decomposition. However, empirical illustrations rely on Bloomberg price data (commercial, not freely available). No code repository is provided. The methodology is analytically transparent and implementable with standard covariance estimates, but exact reproduction of empirical results requires Bloomberg access. Convergence simulations are described but code is not shared.