This thesis investigates the use of realized volatility features from high frequency data in com- bination with neural networks to improve forecasts of the yield curve of government bonds. I use high frequency data on futures of four U.S. Treasury securities to estimate the Nelson-Siegel yield curve and realized variance of its parameters over the period of 25 years. The estimated parameters are used in prediction of the level, slope and curvature of the yield curve using an LSTM neural network and compared to the Dynamic Nelson-Siegel model. Results show that the use of realized variance and neural network outperforms autoregressive methods in prediction of the level and curvature in daily and monthly forecasts. The yield curve of government bonds itself has a predictive power on multiple macroeconomic variables, therefore improvements in its forecastability may have broader implications on forecasting the overall state of the economy.

Purpose: This thesis investigates the relationship concerning traditional safe haven assets, gold and US 10-year treasury bonds during periods of market instability, specifically during the economic concerns raised by the COVID-19 pandemic. It assesses the hedging and safe haven properties of these assets and their dynamic nature throughout two periods of unconventional monetary and fiscal policy measures by the Federal Reserve & US Congress respectively. Furthermore, the study explores a unique divergence between the price movements of the two assets, as well as potential changes in their properties and relationships. Theoretical Perspective: The study is anchored in theoretical concepts based on previous research such as Modern Portfolio Theory, Safe Haven Theory and Hedging Theory. These theories explain asset behaviours during financial turmoil and the relationship between gold and US 10-year treasury bonds during financial crises. The research gap and research questions were formulated based on the information gathered. Methodology: The research employs a quantitative, explanatory approach, anchoredin objectivism and realism, focusing on testing established theories through empirical data. Using a deductive methodology, it investigates potential changes in the dynamic between traditional safe haven assets, gold and US 10-year treasury bonds. Empirical Foundation: Based on a thorough literature review, this study integrates insights from past research and with new data emerging from the pandemic's influence on financial markets and subsequent policy action. The empirical evidence is integrated through quantitative analysis, leveraging ARCH/GARCH models and quantile regression to understand asset performance amid market shocks and policy changes. Conclusion: The findings indicate that gold did not initially act as a hedge against bonds but did so against other assets such as Oil, USD, and BTC during the height of COVID-19. In the recovery phase, this relationship shifted, with gold emerging as a hedge against bonds while its hedging capacity against Oil and Real Yield was negated. Additionally, gold's role as a safe haven against bonds was consistently unsupported across both periods studied. Furthermore, a portfolio analysis revealed a shift in investment strategy, from a balanced gold-bonds mix during the crisis to a sole preference for gold in the recovery phase, adapting to the evolving market conditions and policy changes.