Paper: arXiv 2611.15619

Authors: Edoardo Otranto

Abstract

Covariance models describe the dynamics of conditional covariance matrices as functions of past variances and covariances and are widely used to model the joint evolution of financial asset returns. Their specification faces two major challenges: the curse of dimensionality, whereby the number of parameters increases rapidly with the matrix dimension, and the need to ensure the positive definiteness of the estimated covariance matrices. Moreover, commonly used parsimonious parameterizations, such as scalar and diagonal specifications, fail to capture complex interactions among variances and covariances across assets and cannot accommodate the abrupt changes that often characterize financial markets. We propose a new class of parameterizations that reconciles parsimony with flexibility by allowing distinct, time-varying coefficients for each variance and covariance while requiring only a limited number of parameters. We derive sufficient conditions to ensure the positive definiteness of the resulting covariance matrices and assess the proposed approach through Monte Carlo simulations and an empirical application. The results show that the proposed models can capture complex dependence structures while preserving computational tractability.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents a novel class of parameterizations for realized covariance matrix models, addressing key challenges of dimensionality and positive definiteness with rigorous mathematical derivations. It includes Monte Carlo simulations and an empirical application, demonstrating strong empirical rigor. The methodology is clearly explained, contributing to a high overall score.

Research Flowchart

  flowchart TD
    A[Research Goal: Reconcile Parsimony with Flexibility in Realized Covariance Matrix Models] --> B{Challenges: Curse of Dimensionality & Positive Definiteness};
    B --> C[Methodology: New Class of Parsimonious & Flexible Parameterizations];
    C --> D{Key Features: Distinct, Time-Varying Coefficients for each Variance/Covariance};
    D --> E[Computational Process: Derive Sufficient Conditions for Positive Definiteness];
    E --> F[Validation: Monte Carlo Simulations & Empirical Application (Financial Asset Returns)];
    F --> G[Outcomes: Models Capture Complex Dependence Structures & Preserve Computational Tractability];

Cited References (38)

  1. Realized covariance models with time-varying parameters and spillover effects (2026) — link
  2. VOLatility Archive for Realized Estimates (VOLARE) — → 2602.19732
  3. A matrix-variate log-normal model for covariance matrices (2026) — link
  4. Realized BEKK-CAW Models (2022) — link
  5. Contraction Analysis of Discrete-Time Stochastic Systems (2021) — link
  6. Realized volatility forecasting: Robustness to measurement errors (2021) — link
  7. Quasi‐maximum likelihood estimation of conditional autoregressive Wishart models (2020) — link
  8. Nonlinearities and regimes in conditional correlations with different dynamics (2020) — link
and 30 more
  1. Realized GARCH models: Simpler is better (2020) — link
  2. Modeling and forecasting (un)reliable realized covariances for more reliable financial decisions (2018) — link
  3. High-Dimensional Copula-Based Distributions with Mixed Frequency Data (2016) — link
  4. Exploiting the errors: A simple approach for improved volatility forecasting (2016) — link
  5. Financial Risk Measurement for Financial Risk Management (2011) — link
  6. Modelling and Forecasting Multivariate Realized Volatility (2011) — link
  7. The Conditional Autoregressive Wishart Model for Multivariate Stock Market Volatility (2010) — link
  8. A Generalized Dynamic Conditional Correlation Model: Simulation and Application to Many Assets (2009) — link
  9. A Simple Approximate Long-Memory Model of Realized Volatility (2008) — link
  10. Functions of matrices - theory and computation (2008) — link
  11. New Introduction to Multiple Time Series Analysis (2007) — link
  12. A simple multivariate ARCH model specified by random coefficients (2006) — link
  13. Testing and Valuing Dynamic Correlations for Asset Allocation (2006) — link
  14. Dynamic Conditional Correlation a Simple Class of Multivariate GARCH Models (2000) — link
  15. Matrix Variate Distributions (1999) — link
  16. Multivariate Simultaneous Generalized ARCH (1995) — link
  17. Specification, Estimation, and Evaluation of Smooth Transition Autoregressive Models (1994) — link
  18. Strict Stationarity of Generalized Autoregressive Processes (1992) — link
  19. Modeling realized covariance matrices: A class of hadamard exponential models (2023)
  20. Long term component dynamic models for realized covariance matrices (2014)
  21. Forecasting multivariate realized stock market volatility (2011) — link
  22. The Model Confidence Set ∗ (2009)
  23. ESTIMATION , INFERENCE AND SPECIFICATION ANALYSIS (2008) — link
  24. Multivariate GARCH Models: A Survey (2003) — link
  25. Riskmetrics Technical Document, 4th ed. (1996)
  26. GARCH for groups (1996)
  27. A conditional heteroskedastic model for speculative prices and rates of return (1987)
  28. Large sample estimation and hypothesis testing (1986) — link
  29. Maximum Likelihood Estimation of Misspecified Models (1982) — link
  30. which provides a reasonable compromise between low and high variability. 6 In the case of daily financial time series, the sample size is very often larger than 1000. 13

1 of 38 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2570 linked papers.