Faculty Advisor or Committee Member

Dalin Tang, Advisor

Faculty Advisor or Committee Member

Eugene Yablonski

Faculty Advisor or Committee Member

Bogdan Vernescu

Co-advisor

Eugene Yablonski

Identifier

etd-050405-180040

Abstract

Factor models are very useful and popular models in finance. In this project, factor models are used to examine hidden patterns of relationships for a set of stocks. We calculate the weekly rates of return and analyze the correlation among those variables. We propose to use Principal Factor Analysis (PFA) and Maximum-likelihood Factor Analysis (MLFA) as a data mining tool to recover the hidden factors and the corresponding sensitivities. Prior to applying PFA and MLFA, we use the Scree Test and the Proportion of Variance Method for determining the optimal number of common factors. Then, rotation for PFA and MLFA were performed to improve the first order approximations. PFA and MLFA were used to extract three underlying factors. It was determined that the MLFA provided a more accurate estimation for weekly rates of return

Publisher

Worcester Polytechnic Institute

Degree Name

MS

Department

Mathematical Sciences

Project Type

Thesis

Date Accepted

2005-05-04

Accessibility

Unrestricted

Subjects

Factor Analysis, Principal Factor, Maximum-likelihood, Stock Performance, Stocks, Prices$vMathematical models, Factor analysis

Share

COinS