Scholarlink

<p>This study investigates the time-varying interdependencies among multiple financial markets through a novel framework that integrates return-based features into the traditional connectedness model and constructs a graphical neural network encompassing assets from stock, cryptocurrency, commodity, and exchange-traded fund (ETF) markets over the period from January 1, 2018, to November 10, 2024. The proposed time-varying parameter vector autoregressive model with graph neural networks (TVP-VAR-GNN) is implemented in investigating the connectedness structure across markets and further analyse the inter market linkages during the normal and turbulent phases of COVID-19 pandemic and the Russia–Ukraine conflict. The findings reveal Bitcoin and Ethereum as major transmitters of shocks in the financial network, whereas Tether and the FTSE.SA primarily act as receivers. Bitcoin and Ethereum, forming the same cluster illustrates a resilient bidirectional relationship during normal as well as turbulent time periods. Similarly, technological assets such as Alphabet and IBM exhibit strong interconnections throughout the sample, indicating substantial mutual dependencies. Moreover, the neural network analysis highlights that silver, gold, and crude oil, together with equities from BRICS economies, display safe-haven characteristics, offering potential hedging opportunities during the market turmoil periods. Overall, the study offers valuable insights for fund managers and policymakers regarding systemic risk management, portfolio diversification, and the formulation of informed investment strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Random Walk down Cross-Asset Networks: A Deep Learning Tour of Volatility Transmission between Digital and Conventional Markets

  • 期刊论文

  • 发表日期：

    2026年1月24日

查看全文
  • Hitesh Kumar Sahu,
  • Ipsita Parida,
  • Avishek Bhandari,
  • Nihar Ranjan Jena,
  • Maddikera Vijay

摘要

This study investigates the time-varying interdependencies among multiple financial markets through a novel framework that integrates return-based features into the traditional connectedness model and constructs a graphical neural network encompassing assets from stock, cryptocurrency, commodity, and exchange-traded fund (ETF) markets over the period from January 1, 2018, to November 10, 2024. The proposed time-varying parameter vector autoregressive model with graph neural networks (TVP-VAR-GNN) is implemented in investigating the connectedness structure across markets and further analyse the inter market linkages during the normal and turbulent phases of COVID-19 pandemic and the Russia–Ukraine conflict. The findings reveal Bitcoin and Ethereum as major transmitters of shocks in the financial network, whereas Tether and the FTSE.SA primarily act as receivers. Bitcoin and Ethereum, forming the same cluster illustrates a resilient bidirectional relationship during normal as well as turbulent time periods. Similarly, technological assets such as Alphabet and IBM exhibit strong interconnections throughout the sample, indicating substantial mutual dependencies. Moreover, the neural network analysis highlights that silver, gold, and crude oil, together with equities from BRICS economies, display safe-haven characteristics, offering potential hedging opportunities during the market turmoil periods. Overall, the study offers valuable insights for fund managers and policymakers regarding systemic risk management, portfolio diversification, and the formulation of informed investment strategies.