Paper: arXiv 2411.11847
Authors: Pierdomenico Duttilo
Abstract
This PhD Thesis presents an investigation into the analysis of financial returns using mixture models, focusing on mixtures of generalized normal distributions (MGND) and their extensions. The study addresses several critical issues encountered in the estimation process and proposes innovative solutions to enhance accuracy and efficiency. In Chapter 2, the focus lies on the MGND model and its estimation via expectation conditional maximization (ECM) and generalized expectation maximization (GEM) algorithms. A thorough exploration reveals a degeneracy issue when estimating the shape parameter. Several algorithms are proposed to overcome this critical issue. Chapter 3 extends the theoretical perspective by applying the MGND model on several stock market indices. A two-step approach is proposed for identifying turmoil days and estimating returns and volatility. Chapter 4 introduces constrained mixture of generalized normal distributions (CMGND), enhancing interpretability and efficiency by imposing constraints on parameters. Simulation results highlight the benefits of constrained parameter estimation. Finally, Chapter 5 introduces generalized normal distribution-hidden Markov models (GND-HMMs) able to capture the dynamic nature of financial returns. This manuscript contributes to the statistical modelling of financial returns by offering flexible, parsimonious, and interpretable frameworks. The proposed mixture models capture complex patterns in financial data, thereby facilitating more informed decision-making in financial analysis and risk management.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The thesis involves advanced statistical estimation algorithms (ECM, GEM), complex distribution theory (Generalized Normal, mixtures, HMMs), and significant theoretical derivations, warranting a high math score. It also shows strong empirical rigor through extensive simulation studies, real-world index applications (Euro Stoxx 50, S&P500), model comparisons (AIC, BIC, KS, AD), and proposal of models with direct implications for risk management and volatility estimation.
Research Flowchart
flowchart TD
subgraph Research
G["Research Goal: Model financial returns<br>with Generalized Normal Distributions"]
D["Data: Stock Market Indices<br>& Financial Returns"]
M["Methodology: Mixture Models &<br>ECM/GEM Algorithms"]
C["Computational Processes:<br>Degeneracy Fixes & Optimization"]
F["Key Findings: Flexible,<br>Parsimonious Frameworks"]
end
G --> D
D --> M
M --> C
C --> F
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The thesis proposes mixture models of generalized normal distributions (MGND), constrained MGND (CMGND), and generalized normal distribution hidden Markov models (GND‑HMM) for modelling financial returns.
Source sentence
This PhD Thesis presents an investigation into the analysis of financial returns using mixture models, focusing on mixtures of generalized normal distributions (MGND) and their extensions.
Assumes 4
- Daily log returns are computed for each stock according to equation 2.4. data
Source sentence
Daily log returns are computed for each stock according to equation 2.4.
- The MGND model can objectively detect market turmoil days using a two‑component mixture. model
Source sentence
The use of the MGND model introduces an objective and data‑driven way to identify financial market turmoil days.
- The Markov chain in GND‑HMM is homogeneous (transition probabilities constant over time). model
Source sentence
The Markov chain is homogeneous if the transition probabilities do not depend on
- Returns and volatility are estimated with GARCH models including detected turmoil days as exogenous dummy variables. model
Source sentence
returns and volatility of several market indices are estimated with generalized autoregressive conditional heteroskedasticity (GARCH) models including the detected turmoil days as exogenous dummy variables.
Data
- STOXX50E daily closing prices market-prices public
Source sentence
Data on daily closing prices of the 52 stocks that belong to the STOXX50E are collected from Yahoo Finance for the period from 01 January 2010 to 30 September 2023
- S&P 500 daily closing prices market-prices public
Source sentence
Data on daily closing prices of the two indices are collected from Yahoo Finance. The time period taken for the study is from April 2, 2007 to August 15, 2023.
- Traditional and ESG indices market-prices unknown
Source sentence
Table 3.5: Basic statistics of traditional and ESG indices.
Costs
Not applicable
Finds 4
- Algorithm 1 fails to prevent degeneracy of the shape parameter (ν1) with AVG(bν1)=18.43, while Algorithms 5 and 6 achieve AVG(bν1)=6.21. statistical
Source sentence
Algorithm 1 fails to prevent the degeneracy of the shape parameter, with AVG(bν1) = 18.43
- For sample size N=2000 the degeneracy problem of ν1 disappears and all algorithms provide accurate estimates. statistical
Source sentence
2000, all algorithms obtain a more accurate estimation of ν1. For large samples, the degeneracy problem of ν1 disappears and therefore Algorithm 1 works as expected.
- The MGND model objectively identifies financial market turmoil days, improving accuracy over subjective approaches. performance
Source sentence
The use of the MGND model introduces an objective and data‑driven way to identify financial market turmoil days. Its flexibility to accommodate different shapes of data distributions improves the accuracy in identifying turmoil days.
- Daily returns of S&P 500 and STOXX50E are not normally distributed, exhibiting very high kurtosis and negative skewness. statistical
Source sentence
Table 4.5 demonstrates that daily returns of the S&P500 and STOXX50E are not normally distributed because they are characterised by very high kurtosis and negative…
Checked by others
Supported by
- Comparing Bitcoin and Ethereum tail behavior via Q-Q analysis of cryptocurrency returns
Compare the findings
This paper: Daily returns of S&P 500 and STOXX50E are not normally distributed, exhibiting very high kurtosis and negative skewness.
That paper: Both Bitcoin and Ethereum return distributions are heavy‑tailed, deviating significantly from normality.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.