{"id":{"repo_id":"mcmaster","oai_identifier":"oai:macsphere.mcmaster.ca:11375/33779"},"canonical_url":"https://search.ndltd.org/etd/mcmaster/oai:macsphere.mcmaster.ca:11375/33779","repository":{"repo_id":"mcmaster","name":"McMaster University","base_url":"https://macsphere.mcmaster.ca/server/oai/request"},"display":{"title":"Three Essays on Modelling Financial Time Series: A Time-Varying Bayesian Nonparametric Approach","abstract":"This thesis develops a time-varying Bayesian nonparametric framework for modelling the full distribution of financial time series. Unlike models that allow conditional moments to vary while maintaining a fixed innovation law, the proposed time-dependent Dirichlet process mixture (DDPM) lets mixture weights change when the data support a distributional break. The framework is designed to distinguish persistent distributional changes from volatility clustering while allowing location, asymmetry, tail thickness, multimodality, and cross-asset dependence to evolve without fixing the number of mixture components in advance. Chapter 1 introduces a reformulated DDPM and joint Markov chain Monte Carlo sampler. Simulations show that it identifies substantial variance and shape changes while producing weak break evidence and negligible predictive gains for minor shifts. An application to monthly three-month U.S. Treasury bill rates combines the DDPM with stochastic volatility and separates lower-frequency distributional changes from localized volatility dynamics. Chapter 2 embeds the DDPM in a multivariate GARCH model for equity and long-term Treasury bond returns during COVID-19. The model identifies changes in the joint innovation distribution beyond conditional covariance dynamics and improves predictive-density and tail-risk forecasts relative to parametric and static nonparametric benchmarks. It also improves global minimum-variance and utility-based portfolio outcomes during the crisis evaluation period. Chapter 3 develops an order-invariant multivariate specification and applies it to five assets around U.S. tariff announcements. The initial April 2025 announcement is associated with strong break evidence, whereas later tariff news is primarily absorbed through within-regime covariance adjustments. The model delivers gains in cumulative log-predictive likelihood and improves value-at-risk and expected shortfall forecasts across fixed portfolios. Overall, the thesis shows that allowing flexible financial distributions to evolve over time provides information beyond conditional moments. The proposed framework offers a way to identify structural distributional changes and improves the measurement and forecasting of density, dependence, and tail risk.","abstract_html":"This thesis develops a time-varying Bayesian nonparametric framework for modelling the full distribution of financial time series. Unlike models that allow conditional moments to vary while maintaining a fixed innovation law, the proposed time-dependent Dirichlet process mixture (DDPM) lets mixture weights change when the data support a distributional break. The framework is designed to distinguish persistent distributional changes from volatility clustering while allowing location, asymmetry, tail thickness, multimodality, and cross-asset dependence to evolve without fixing the number of mixture components in advance. Chapter 1 introduces a reformulated DDPM and joint Markov chain Monte Carlo sampler. Simulations show that it identifies substantial variance and shape changes while producing weak break evidence and negligible predictive gains for minor shifts. An application to monthly three-month U.S. Treasury bill rates combines the DDPM with stochastic volatility and separates lower-frequency distributional changes from localized volatility dynamics. Chapter 2 embeds the DDPM in a multivariate GARCH model for equity and long-term Treasury bond returns during COVID-19. The model identifies changes in the joint innovation distribution beyond conditional covariance dynamics and improves predictive-density and tail-risk forecasts relative to parametric and static nonparametric benchmarks. It also improves global minimum-variance and utility-based portfolio outcomes during the crisis evaluation period. Chapter 3 develops an order-invariant multivariate specification and applies it to five assets around U.S. tariff announcements. The initial April 2025 announcement is associated with strong break evidence, whereas later tariff news is primarily absorbed through within-regime covariance adjustments. The model delivers gains in cumulative log-predictive likelihood and improves value-at-risk and expected shortfall forecasts across fixed portfolios. Overall, the thesis shows that allowing flexible financial distributions to evolve over time provides information beyond conditional moments. The proposed framework offers a way to identify structural distributional changes and improves the measurement and forecasting of density, dependence, and tail risk.","abstract_has_math":false,"creators":["Khoshro, Hamidreza"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Business Administration","school":null,"contributors":[],"advisors":["Maheu, John"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:37:46Z","subjects":["Bayesian nonparametrics","financial time series","time-varying distributions","Dirichlet process mixtures","density estimation","stochastic volatility","multivariate GARCH"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.71548/25549"],"render_values":[{"text":"https://doi.org/10.71548/25549","href":"https://doi.org/10.71548/25549","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11375/33779","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/11375/33779","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://macsphere.mcmaster.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amacsphere.mcmaster.ca%3A11375%2F33779","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Maheu, John"]},{"key":"dc:contributor.department","label":"Department","values":["Business Administration"]},{"key":"dc:creator","label":"Author","values":["Khoshro, Hamidreza"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-10-14T12:54:26Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Bayesian nonparametrics","financial time series","time-varying distributions","Dirichlet process mixtures","density estimation","stochastic volatility","multivariate GARCH"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11375/33779","https://doi.org/10.71548/25549"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This thesis develops a time-varying Bayesian nonparametric framework for modelling the full distribution of financial time series. Unlike models that allow conditional moments to vary while maintaining a fixed innovation law, the proposed time-dependent Dirichlet process mixture (DDPM) lets mixture weights change when the data support a distributional break. The framework is designed to distinguish persistent distributional changes from volatility clustering while allowing location, asymmetry, tail thickness, multimodality, and cross-asset dependence to evolve without fixing the number of mixture components in advance. Chapter 1 introduces a reformulated DDPM and joint Markov chain Monte Carlo sampler. Simulations show that it identifies substantial variance and shape changes while producing weak break evidence and negligible predictive gains for minor shifts. An application to monthly three-month U.S. Treasury bill rates combines the DDPM with stochastic volatility and separates lower-frequency distributional changes from localized volatility dynamics. Chapter 2 embeds the DDPM in a multivariate GARCH model for equity and long-term Treasury bond returns during COVID-19. The model identifies changes in the joint innovation distribution beyond conditional covariance dynamics and improves predictive-density and tail-risk forecasts relative to parametric and static nonparametric benchmarks. It also improves global minimum-variance and utility-based portfolio outcomes during the crisis evaluation period. Chapter 3 develops an order-invariant multivariate specification and applies it to five assets around U.S. tariff announcements. The initial April 2025 announcement is associated with strong break evidence, whereas later tariff news is primarily absorbed through within-regime covariance adjustments. The model delivers gains in cumulative log-predictive likelihood and improves value-at-risk and expected shortfall forecasts across fixed portfolios. Overall, the thesis shows that allowing flexible financial distributions to evolve over time provides information beyond conditional moments. The proposed framework offers a way to identify structural distributional changes and improves the measurement and forecasting of density, dependence, and tail risk."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Doctor of Philosophy (PhD)"]},{"key":"dc:title","label":"Title","values":["Three Essays on Modelling Financial Time Series: A Time-Varying Bayesian Nonparametric Approach"]},{"key":"others:access_status","label":"Others Access Status","values":["open.access"]}]}],"canonical_facts":{"dc:contributor.advisor":["Maheu, John"],"dc:contributor.department":["Business Administration"],"dc:creator":["Khoshro, Hamidreza"],"dc:date.accessioned":["2026-10-14T12:54:26Z"],"dc:date.issued":["2026"],"dc:description.abstract":["This thesis develops a time-varying Bayesian nonparametric framework for modelling the full distribution of financial time series. Unlike models that allow conditional moments to vary while maintaining a fixed innovation law, the proposed time-dependent Dirichlet process mixture (DDPM) lets mixture weights change when the data support a distributional break. The framework is designed to distinguish persistent distributional changes from volatility clustering while allowing location, asymmetry, tail thickness, multimodality, and cross-asset dependence to evolve without fixing the number of mixture components in advance. Chapter 1 introduces a reformulated DDPM and joint Markov chain Monte Carlo sampler. Simulations show that it identifies substantial variance and shape changes while producing weak break evidence and negligible predictive gains for minor shifts. An application to monthly three-month U.S. Treasury bill rates combines the DDPM with stochastic volatility and separates lower-frequency distributional changes from localized volatility dynamics. Chapter 2 embeds the DDPM in a multivariate GARCH model for equity and long-term Treasury bond returns during COVID-19. The model identifies changes in the joint innovation distribution beyond conditional covariance dynamics and improves predictive-density and tail-risk forecasts relative to parametric and static nonparametric benchmarks. It also improves global minimum-variance and utility-based portfolio outcomes during the crisis evaluation period. Chapter 3 develops an order-invariant multivariate specification and applies it to five assets around U.S. tariff announcements. The initial April 2025 announcement is associated with strong break evidence, whereas later tariff news is primarily absorbed through within-regime covariance adjustments. The model delivers gains in cumulative log-predictive likelihood and improves value-at-risk and expected shortfall forecasts across fixed portfolios. Overall, the thesis shows that allowing flexible financial distributions to evolve over time provides information beyond conditional moments. The proposed framework offers a way to identify structural distributional changes and improves the measurement and forecasting of density, dependence, and tail risk."],"dc:description.degree":["Doctor of Philosophy (PhD)"],"dc:identifier.uri":["https://hdl.handle.net/11375/33779","https://doi.org/10.71548/25549"],"dc:language.iso":["en"],"dc:subject":["Bayesian nonparametrics","financial time series","time-varying distributions","Dirichlet process mixtures","density estimation","stochastic volatility","multivariate GARCH"],"dc:title":["Three Essays on Modelling Financial Time Series: A Time-Varying Bayesian Nonparametric Approach"],"dc:type":["Thesis"],"others:access_status":["open.access"]},"updated_at":"2026-11-18T19:37:46Z"}