prikaz prve stranice dokumenta Estimation of latent factors from high-dimensional financial time series based on unsupervised learning.
  Download
PDF 46.4 MB
doctoral thesis
Estimation of latent factors from high-dimensional financial time series based on unsupervised learning.
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2020. urn:nbn:hr:168:732129
Begušić, Stjepan
University of Zagreb
Faculty of Electrical Engineering and Computing
Department of Electronic Systems and Information Processing

Institutional repository: FER Repository

Cite this document

Begušić, S. (2020). Estimation of latent factors from high-dimensional financial time series based on unsupervised learning. (Doctoral thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:732129

Begušić, Stjepan. "Estimation of latent factors from high-dimensional financial time series based on unsupervised learning.." Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2020. https://urn.nsk.hr/urn:nbn:hr:168:732129

Begušić, Stjepan. "Estimation of latent factors from high-dimensional financial time series based on unsupervised learning.." Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2020. https://urn.nsk.hr/urn:nbn:hr:168:732129

Begušić, S. (2020). 'Estimation of latent factors from high-dimensional financial time series based on unsupervised learning.', Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 17 November 2024, https://urn.nsk.hr/urn:nbn:hr:168:732129

Begušić S. Estimation of latent factors from high-dimensional financial time series based on unsupervised learning. [Doctoral thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2020 [cited 2024 November 17] Available at: https://urn.nsk.hr/urn:nbn:hr:168:732129

S. Begušić, "Estimation of latent factors from high-dimensional financial time series based on unsupervised learning.", Doctoral thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2020. Available at: https://urn.nsk.hr/urn:nbn:hr:168:732129

Metadata
TitleEstimation of latent factors from high-dimensional financial time series based on unsupervised learning.
Title (croatian)Procjena latentnih faktora iz visokodimenzionalnih financijskih vremenskih nizova primjenom nenadziranoga učenja.
AuthorStjepan Begušić
MentorZvonko Kostanjčar (mentor)
Committee memberZvonko Kostanjčar (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
(Department of Electronic Systems and Information Processing)
Zagreb
Defense date and country2020, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Data Processing
Universal decimal classification
(UDC)		004 - Computer science and technology. Computing. Data processing
Abstract
Unsupervised learning methods have been increasingly used for detecting latent factors in high-dimensional time series, with many applications, especially in financial risk modelling. Most latent factor models assume that the factors are pervasive and affect all of the time series. However, some factors may affect only certain assets in financial markets, due to their clustering within countries, asset classes, or sector classifications. In this thesis, high-dimensional financial time series with pervasive and cluster-specific latent factors are considered. For the assumed latent factor model, an iterative method for clustering and latent factor estimation is proposed. A model selection algorithm is also developed, based on the spectral properties of asset correlation matrices and asset graphs. Based on the estimated latent factor structures, a covariance matrix estimator is also proposed, decomposing the security return covariance into the pervasive latent factor component, cluster-specific latent factor component, and a sparse idiosyncratic risk component. The covariance matrix estimates are used in a portfolio optimization scenario, focusing on risk-based portfolios. Moreover, a new portfolio optimization method based on the risk contributions of the identified latent factors and security clusters is also developed. A simulation study with known data generating processes demonstrates that the proposed latent factor estimation and clustering method outperforms other clustering methods and provides estimates with a high degree of accuracy. Moreover, the model selection procedure is also shown to provide stable and accurate estimates for the number of clusters and latent factors. In addition, risk-based portfolios using the estimated latent factor structures are tested on datasets of asset returns from global financial markets using a backtesting approach. The results demonstrate that the clustering approach and estimated latent factors yield relevant information, improve risk modelling and reduce volatility in the out-of-sample portfolio returns.
Abstract (croatian)
Metode nenadziranog učenja se sve više koriste za procjenu latentnih faktora u visokodimenzionalnim vremenskim nizovima, s mnogim primjenama, pogotovo u modeliranju financijskih rizika. Većina modela s latentnim faktorima pretpostavlja da su faktori široki i da utječu na sve razmatrane vremenske nizove. No, neki faktori mogu utjecati samo na određene vrijednosnice u financijskim tržištima, zbog njihovog grupiranja s obzirom na države, klase imovine ili sektorsku pripadnost. U ovoj disertaciji razmatraju se modeli sa širokim latentnim faktorima i latentnim faktorima specifičnima za grupe u visokodimenzionalnim financijskim vremenskim nizovima. Za pretpostavljeni faktorski model je predložena iterativna metoda za procjenu latentnih faktora i grupa iz podataka. Također je razvijen i algoritam za odabir modela, zasnovan na spektralnim svojstvima korelacijske matrice vrijednosnica i grafa vrijednosnica. Koristeći procijenjene latentne faktorske strukture, predložen je i procjenitelj za matricu kovarijance, koji dekomponira kovarijancu u komponente širokih latentnih faktora, latentnih faktora specifičnih za grupe, i rijetku komponentu s vlastitim rizicima. Procjene kovarijanci su korištene u optimizaciji portfelja, s naglaskom na portfelje zasnovane na rizicima. Također, predložena je i nova metoda za optimizaciju portfelja zasnovana na doprinosima riziku identificiranih grupa i latentnih faktora. Simulacijski okvir u kojem su predložene metode testirane s poznatim procesima generiranja podataka pokazuje da predložena metoda za procjenu latentnih faktora i grupiranje vremenskih nizova nadmašuje ostale razmatrane metode za procjenu latentnih faktora i algoritme grupiranja. Također je pokazano da procedura za odabir modela daje stabilne i točne procjene broja grupa, širokih latentnih faktora i latentnih faktora specifičnih za grupe. Portfelji zasnovani na riziku koji koriste procijenjene latentne faktorske strukture su testirani na skupovima povijesnih povrata vrijednosnica s globalnih financijskih tržišta. Rezultati ukazuju na to predložene metode i pristup grupiranju i procjeni latentnih faktora donose bitne informacije te poboljšavaju modeliranje rizika uz smanjenje volatilnosti portfelja izvan uzorka procjene.
Keywords
Keywords (croatian)
Languageenglish
URN:NBNurn:nbn:hr:168:732129
Promotion2020
Study programmeTitle: Computer Science
Study programme type: university
Study level: postgraduate
Academic / professional title: Doktor znanosti (Doktor znanosti)
Type of resourceText
Extent105 str. ; 30 cm
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2020-12-14 17:23:38