University of Washington - Department of Statistics
Computational Finance Seminar
The first part of this talk will discuss the use of robust statistics in finance. We briefly describe what it means for a statistic to be robust, and then provide striking applications to estimating betas, correlations and cross-section regressions. The second part of the talk begins by briefly describing the three main types of factor models used in finance, namely time series factor models, cross-section regression/fundamental factor models, and factor analysis models. Then we discuss the current status and long-term goals of a project to design and implement a versatile data structure and associated object-oriented methods for both least squares and robust fitting of factor models in S-PLUS. The methodology includes convenient graphical display methods for exploring the goodness of fit of the model. The basic data structure is a three-dimensional cube, and the initial focus is on populating the cube with combined CRSP and COMPUSTAT data and fitting cross-section regression/fundamental factor models. The over-arching long-term goal is to have a product that will support research on the use of robust factor models in finance, including the use of factor models for portfolio optimization and risk management.
*This is joint work with Chris Green, Eric Aldrich and Heiko Bailer.