Spectral Risk Parity

Written By Chris Smith

ADIA Lab Research Paper Series

Authors: Marcos López de Prado, Alexander Lipton, Koushik Balasubramanian

Date Published: February 2026

Utilizing concepts from random matrix theory (RMT), we propose a spectral approach to constructing risk parity portfolios. This portfolio tracks the principal component portfolio while explicitly penalizing the exposure to noise. The principal components and noisy components are identified using the Marčenko-Pastur distribution associated with the covariance (correlation) matrix. Such portfolios are robust to noise by construction. We provide empirical evidence to show the out-performance over the equal-risk-contribution (ERC) and minimum variance (MinVar) portfolios.

DOWNLOAD THIS PAPER
Chris Smith
Previous

Causality and Factor Investing: A Primer
Next

The False Discovery Rate in Finance: Identification Failure and Search-Adjusted Estimation