Performance evaluation of volatility estimation methods for Exabel

TL;DR

This paper compares various volatility estimation methods for a large universe of companies and portfolios, finding that a GARCH-based approach using entity returns performs best in ranking volatility.

Contribution

It introduces a comprehensive comparison of volatility estimation strategies across a large dataset, highlighting the effectiveness of GARCH models with direct entity returns.

Findings

GARCH model with entity returns outperforms other methods in ranking volatility

The study covers data from 2010 to 2021 across 28,629 companies and 858 portfolios

Direct entity return-based GARCH estimation provides the most accurate volatility rankings

Abstract

Quantifying both historic and future volatility is key in portfolio risk management. This note presents and compares estimation strategies for volatility estimation in an estimation universe consisting on 28 629 unique companies from February 2010 to April 2021, with 858 different portfolios. The estimation methods are compared in terms of how they rank the volatility of the different subsets of portfolios. The overall best performing approach estimates volatility from direct entity returns using a GARCH model for variance estimation.