Singular spectrum analysis forecasting for financial time series

Abstract

SSA is a relatively new non-parametric data-driven technique in time series analysis, has been developed and applied to many practical problems across different fields. This paper focuses on the technique of Singular Spectrum Analysis (SSA), its application for financial time series, and also represents results of numerical experiments done by author. The main algorithm of SSA consists of two complementary stages: decomposition and reconstruction; both stages include two separate steps. The performance of the SSA technique is assessed by applying it to the close prices of “AS Tallink Grupp” stock. Results in this work are obtained from creation trajectory matrix of given time series and finding eigenvalues and eigenvectors; construction of the principal and reconstructed components of the time series; applying forecasting algorithm to the time series; interpretation of obtained results. In this thesis for numerical experiments we use the software Matlab.