Walter Scott, Jr. College of Engineering

Graduate Exam Abstract

Ramin Zahedi
M.S. Final
Oct 30, 2007, 10:30 am
ECE conference room
Stochastic Dynamic Programming Approximation Techniques for the Portfolio Management Problem
Abstract: The portfolio management problem is concerned with how an investor should divide wealth among different assets. This problem was first formulated by Markowitz in 1952. Since then, a variety of methods have been introduced for solving this problem. However, only a small number of these techniques are useful in the real market.

In this research, we develop a model that is based on stochastic dynamic programming. In this model, we use two well known time series models, ARMA and GARCH, and build on policies that are commonly used in the market for trading. We show that if the behavior of the market is appropriately modeled, then our model can be used to develop a portfolio management algorithm that performs better than a variety of techniques available in the market.
Adviser: Prof. Edwin Chong
Non-ECE Member: Prof. Jan Hannig (Statistics)
Member 3: Prof. Peter Young (ECE)
Addional Members: -
Program of Study: