Abstract

This thesis contributes to the topic of developing methods for bond price predictions over a time horizon of one to five days. This is done by first representing yields as a term structure of spot rates by introducing an alternative Nelson-Siegel model which includes a linearized estimation procedure provided with a new approach in determining a λ-parameter. We then proceed by comparing this representation of yields to a flat spread structure across maturities. This is followed by applying the latter approaches to the U.S. Treasury, Goldman Sachs and Verizon Communications during the years 2017 and 2018. In order to assess the accuracy of bond price predictions, we compare three distinct approaches for the representation of interest rates and yields. In addition, we want to know if more complex yield curve modeling provides better price predictions than more naive approaches.