Logo do repositório

CLSBE - Dissertações de Mestrado / Master Dissertations

URI permanente para esta coleção:

http://hdl.handle.net/10400.14/2528

Navegar

Percorrer CLSBE - Dissertações de Mestrado / Master Dissertations por assunto "1/N"

A mostrar 1 - 1 de 1
  • Cryptocurrencies : a way to diversify a financial portfolio
    Publication . Kraft, Isabel Marie; Fouquau, Julien
    Cryptocurrencies have attracted growing attention from investors worldwide, prompting debate over whether they can serve as effective diversifiers and enhance portfolio performance. This study addresses this question and further evaluates whether a DCC-GARCH model improves portfolio forecasting compared to a simple historical average approach. Using daily data from November 2017 to March 2025, I first estimate historical bivariate DCC- GARCH models between four cryptocurrencies (Bitcoin, Ethereum, Cardano, and Litecoin) and five traditional financial assets (government bonds, global and emerging equity indices, gold, and oil). Results suggest that, while correlations tend to increase during periods of market stress, cryptocurrencies generally exhibit time-varying but, on average, low correlations with traditional assets, supporting their role as diversifiers. Next, I split the sample into in- and out-of-sample periods with daily rolling windows to construct various portfolios with and without cryptocurrencies, using mean and (co-)variance forecasts derived from both multivariate DCC-GARCH and historical averages. I find that including cryptocurrencies in traditional portfolios only partially improves performance, depending on the period, cryptocurrency, and portfolio strategy. In risk-focused portfolio strategies like Maximum-Diversification and Risk-Parity, adding cryptocurrencies often leads to improved risk- adjusted performance, while Mean-Variance and 1/N strategies produce mixed results. Minimum-Variance portfolios consistently exclude cryptocurrencies due to their high volatilities. Among the analysed cryptocurrencies, Bitcoin emerges as the most valuable addition. Finally, I find that portfolios constructed with DCC-GARCH forecasts do not outperform those constructed with historical averages.
    2025-10-17Dissertação de mestrado Acesso embargado Ver mais