Unveiling Systematic Risks: PCA Analysis to High-Frequency and Low-Frequency Factor Data Open Access

Abstract

Systematic risks have long been analyzed through factor models. This paper applies quadratic Principal Component Analysis (PCA), as proposed by Pelger (2019), to extract stable systematic risk factors from high-frequency data and compares them with factors identified from low-frequency data. The findings suggest that industry factors like technology, financials, energy, and industrials consistently dominate the explained variance, highlighting their central role in systematic risk. While for the low-frequency data, accounting-related factors emerge as the primary drivers of variance. However, during portfolio optimization, the factors with the highest optimized weights are notably different from the factors in the leading principal components. These weights are more firm-specific and diversified across various categories, emphasizing the importance of idiosyncratic drivers in practical investment strategies. This study contributes to the literature by highlighting the divergence between systematic risk decomposition and optimized portfolio construction, offering valuable insights for risk management and investment strategies. It underscores the importance of integrating firm-specific factors with broader systematic components to enhance portfolio performance and suggests future research directions in dynamic modeling and event-driven analysis.

Table of Contents

Section 1: Introduction                                                                                                  1

Section 2: Methodology                                                                                                 3

2.1 PCA Overview                                                                                                              3

2.2 Comparison between Factors                                                                                   4

2.3 Mean-Variance Portfolio Optimization                                                                  5

Section 3: Dataset                                                                                                           6

Section 4: Empirical Results                                                                                         7

4.1 Latent Factor Estimation                                                                                           7

4.2 State-varying Factors                                                                                                15

4.3 Portfolio Optimization                                                                                              20

Section 5: Conclusion                                                                                                    26

Section 6: Further Discussion                                                                                     27

About this Honors Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Emory College
Department	Quantitative Science
Degree	B.B.A.
Submission	Honors Thesis
Language	English
Research Field	Statistics Economics, Finance
Keyword	Machine Learning High Frequency Data Systematic Risk Principal Component Analysis
Committee Chair / Thesis Advisor	Xiong, Ruoxuan, Emory University
Committee Members	Mann, William Giles, Emory University Lyle, Matthew R., Emory University

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Unveiling Systematic Risks: PCA Analysis to High-Frequency and Low-Frequency Factor Data ()	2024-12-11 23:50:00 -0500	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions