Machine Learning Algorithms Application for Fixed-Income Market Analysis: Cross Countries Comparisons

Abstract

The research aim is to distinguish the global fixed income market structure and dynamics basing on such core indicator as yields to maturity of government bonds. The processes taken part within or linked with fixed income market are of core importance for the global world economic system as a whole, thus have to be analyzed and researched constantly to ensure decision-making processes on all levels of hierarchy. The following tasks have been implemented to achieve the aim: initial observation set of yields data for world government bonds of different maturities has been formed; initial dataset dimension has been reduced based on principal components analysis technique; fixed-income market structure has been analyzed in the obtained three-dimensional level-slope-curvature space for each researched country; set of countries has been classified onto homogenous groups in multi-dimensional space basing on k-means algorithm; the resulting groupings composition and structure have been analyzed.