Eigenvalues of the stock market covariance matrix
Specialeforsvar ved Marianne Kjeldbjerg
Titel: EIGENVALUES OF THE STOCK MARKET COVARIANCE MATRIX
Resume: During the past decade the availability of financial data has increased tremendously. This increase leads to a large amount of available data for analyzing the uncertainties in the stock market. One tool for extracting information in huge data sets is PCA.We examine how PCA can be used to reduce the complexity of stock data, and then we explore the estimation uncertainty that the use of PCA introduces. When PCA is conducted, the eigenvalues and eigenvectors are estimated. A large part of the thesis concerns the behavior of the estimated eigenvalues. We both study dierent simulation
experiments, and actual observed data prices. Financial data however can be di cult to model and understand, not only are the dependencies di cult to model, they also change over time. Through a study of the eigenvalues, we investigate which kind of information the eigenvalues capture about the dependence structure
Vejleder: Thomas Mikosch
Censor: Søren Asmussen, Aarhus Universitet