THIS THESIS ADVANCES THE APPLICATION OF NETWORK SCIENCE METHODS TO FINANCIAL TIME SERIES ANALYSIS THROUGH SEVERAL METHODOLOGICAL AND EMPIRICAL CONTRIBUTIONS THAT FUNDAMENTALLY ENHANCE OUR UNDERSTANDING OF MARKET STRUCTURE AND INTER- CONNECTEDNESS. CHAPTER 1 ESTABLISHES THE METHODOLOGICAL FOUNDATION BY SYSTEMATICALLY REVIEW- ING NETWORK SCIENCE CONCEPTS AND THEIR EVOLUTION IN FINANCIAL CONTEXTS, TRACING THE DEVELOPMENT FROM CLASSICAL GRAPH THEORY TO MODERN FINANCIAL NETWORKS. IT PROVIDES A COMPREHENSIVE TAXONOMY OF METHODOLOGIES WITH PARTICULAR EMPHASIS ON STOCK CORRELATION NETWORKS, A CENTRAL TOPIC IN THE LITERATURE AND THE FOCUS OF CHAPTER 2. CHAPTER 2 ADDRESSES CRITICAL YET LARGELY NEGLECTED CHALLENGES IN STOCK CORRELATION NETWORK ANALYSIS THROUGH THE MOST EXTENSIVE MULTI-FREQUENCY EMPIRICAL INVESTI- GATION IN THE LITERATURE. WE SHOW THAT CONVENTIONAL METHODS INTRODUCE SEVERE DISTORTIONS WHEN APPLIED TO NOISY FINANCIAL DATA AND PROPOSE ROBUST COUNTERPARTS ACHIEVING UNPRECEDENTED IMPROVEMENTS IN COMMUNITY DETECTION PERFORMANCES. MOREOVER, WE ESTABLISH THAT JOINTLY CONSIDERING MULTIPLE INFORMATION SOURCES RE- VEALS AN INTRINSIC CAPACITY FOR MARKET SELF-ORGANIZATION INVISIBLE TO TRADITIONAL AP- PROACHES. THE RESULTING COMMUNITIES ACHIEVE THE HIGHEST SECTOR ALIGNMENT EVER DOCUMENTED WHILE GENERATING SUPERIOR RISK-ADJUSTED RETURNS, VALIDATED THROUGH THE EMPIRICAL ANALYSIS OF THE LARGEST AMERICAN STOCK MARKET OBSERVED FOR 20 YEARS AT HIGH FREQUENCIES. CHAPTER 3 PROVIDES THE FIRST LARGE-SCALE APPLICATION OF STOCHASTIC BLOCK MODELS TO FINANCIAL TIME SERIES, IMPLEMENTING GENERATIVE NETWORK RECONSTRUCTION ACROSS TWENTY YEARS OF S&P 500 DATA. THIS EXTENSIVE INVESTIGATION ESTABLISHES THE VI- ABILITY OF GENERATIVE APPROACHES AT PRACTICAL SCALES. WE ALSO INTRODUCE A NOVEL VALIDATION FRAMEWORK ADDRESSING A FUNDAMENTAL LIMITATION IN FINANCIAL NETWORK RECONSTRUCTION: EVALUATING RECONSTRUCTION QUALITY WITHOUT A GROUND TRUTH NET- WORK. OUR FRAMEWORK REVEALS CLEAR SIGNATURES OF MAJOR ECONOMIC EVENTS, DEMON- STRATING THE CAPACITY TO CAPTURE MEANINGFUL STRUCTURAL TRANSITIONS DURING CRISES. COLLECTIVELY, THIS THESIS DEMONSTRATES THAT ADVANCED NETWORK METHODOLOGIES, RIG- OROUSLY VALIDATED THROUGH COMPREHENSIVE EMPIRICAL ANALYSIS, EXTRACT ECONOMI- CALLY MEANINGFUL MARKET STRUCTURES WITH DIRECT IMPLICATIONS FOR ASSET ALLOCATION, RISK MANAGEMENT, AND SYSTEMIC RISK ASSESSMENT.

ADVANCES IN NETWORK METHODS FOR FINANCIAL TIME SERIES ANALYSIS

Peluso, Alfonso
2026

Abstract

THIS THESIS ADVANCES THE APPLICATION OF NETWORK SCIENCE METHODS TO FINANCIAL TIME SERIES ANALYSIS THROUGH SEVERAL METHODOLOGICAL AND EMPIRICAL CONTRIBUTIONS THAT FUNDAMENTALLY ENHANCE OUR UNDERSTANDING OF MARKET STRUCTURE AND INTER- CONNECTEDNESS. CHAPTER 1 ESTABLISHES THE METHODOLOGICAL FOUNDATION BY SYSTEMATICALLY REVIEW- ING NETWORK SCIENCE CONCEPTS AND THEIR EVOLUTION IN FINANCIAL CONTEXTS, TRACING THE DEVELOPMENT FROM CLASSICAL GRAPH THEORY TO MODERN FINANCIAL NETWORKS. IT PROVIDES A COMPREHENSIVE TAXONOMY OF METHODOLOGIES WITH PARTICULAR EMPHASIS ON STOCK CORRELATION NETWORKS, A CENTRAL TOPIC IN THE LITERATURE AND THE FOCUS OF CHAPTER 2. CHAPTER 2 ADDRESSES CRITICAL YET LARGELY NEGLECTED CHALLENGES IN STOCK CORRELATION NETWORK ANALYSIS THROUGH THE MOST EXTENSIVE MULTI-FREQUENCY EMPIRICAL INVESTI- GATION IN THE LITERATURE. WE SHOW THAT CONVENTIONAL METHODS INTRODUCE SEVERE DISTORTIONS WHEN APPLIED TO NOISY FINANCIAL DATA AND PROPOSE ROBUST COUNTERPARTS ACHIEVING UNPRECEDENTED IMPROVEMENTS IN COMMUNITY DETECTION PERFORMANCES. MOREOVER, WE ESTABLISH THAT JOINTLY CONSIDERING MULTIPLE INFORMATION SOURCES RE- VEALS AN INTRINSIC CAPACITY FOR MARKET SELF-ORGANIZATION INVISIBLE TO TRADITIONAL AP- PROACHES. THE RESULTING COMMUNITIES ACHIEVE THE HIGHEST SECTOR ALIGNMENT EVER DOCUMENTED WHILE GENERATING SUPERIOR RISK-ADJUSTED RETURNS, VALIDATED THROUGH THE EMPIRICAL ANALYSIS OF THE LARGEST AMERICAN STOCK MARKET OBSERVED FOR 20 YEARS AT HIGH FREQUENCIES. CHAPTER 3 PROVIDES THE FIRST LARGE-SCALE APPLICATION OF STOCHASTIC BLOCK MODELS TO FINANCIAL TIME SERIES, IMPLEMENTING GENERATIVE NETWORK RECONSTRUCTION ACROSS TWENTY YEARS OF S&P 500 DATA. THIS EXTENSIVE INVESTIGATION ESTABLISHES THE VI- ABILITY OF GENERATIVE APPROACHES AT PRACTICAL SCALES. WE ALSO INTRODUCE A NOVEL VALIDATION FRAMEWORK ADDRESSING A FUNDAMENTAL LIMITATION IN FINANCIAL NETWORK RECONSTRUCTION: EVALUATING RECONSTRUCTION QUALITY WITHOUT A GROUND TRUTH NET- WORK. OUR FRAMEWORK REVEALS CLEAR SIGNATURES OF MAJOR ECONOMIC EVENTS, DEMON- STRATING THE CAPACITY TO CAPTURE MEANINGFUL STRUCTURAL TRANSITIONS DURING CRISES. COLLECTIVELY, THIS THESIS DEMONSTRATES THAT ADVANCED NETWORK METHODOLOGIES, RIG- OROUSLY VALIDATED THROUGH COMPREHENSIVE EMPIRICAL ANALYSIS, EXTRACT ECONOMI- CALLY MEANINGFUL MARKET STRUCTURES WITH DIRECT IMPLICATIONS FOR ASSET ALLOCATION, RISK MANAGEMENT, AND SYSTEMIC RISK ASSESSMENT.
16-giu-2026
Inglese
CORETTO, Pietro
Università degli Studi di Salerno
File in questo prodotto:
File Dimensione Formato  
abstract (2).pdf

accesso aperto

Licenza: Tutti i diritti riservati
Dimensione 56.02 kB
Formato Adobe PDF
56.02 kB Adobe PDF Visualizza/Apri
main.pdf

accesso aperto

Licenza: Tutti i diritti riservati
Dimensione 29.32 MB
Formato Adobe PDF
29.32 MB Adobe PDF Visualizza/Apri

I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.14242/385495
Il codice NBN di questa tesi è URN:NBN:IT:UNISA-385495