Author(s): Sumit Kumar
PCA (Principal Component Analysis) reduces the dimensionality of an input dataset while also ensuring that it preserves maximum information. In the present work, we conducted a PCA on US treasury Bonds. We took a data set of 9 treasury bonds of various maturities and computed the principal factors that explain the maximum variances. This study suggested a set of hedges that effectively hedge the bond portfolio's significant risk without taking an off-setting position with all the bond holdings. This methodology of creating a hedge against the interest rate movement will reduce the trading desk's hedging cost and increase operational efficiency, thereby reducing operational risk. The extraction of Eigenvalues and Eigenvectors from the data produced 9 PCs (Principal Components), of which the first two explain 99.137% of all variances in the bond yields. Analyzing the correlations between the first two PCs and the initial variables revealed that the best bonds to hedge in the portfolio are the five-year and 7-year maturity bonds.