Portfolio Analysis

Abstract

In this dissertation, the techniques of portfolio analysis have been considered and the merits of the various models found in the literature are discussed. Portfolio analysis involves the determination of the most suitable combination of specific shares that should be included in the portfolio of an investor at any point in time. The main purpose of this study has been to compare the suitability of the following five methods of portfolioselection: 1) The Filter Method, 2) The Valuation Method, 3) The Markowitz Model, 4) The Sharpe Model, 5) A Monte Carlo Technique. Of these methods the Markowitz and the Sharpe approaches might be termed as classical models - and the Valuation and Monte Carlo methods as developments proposed and tested by the author. The ad-hoc methods were used as a crude strategy against which the efficiency of the more complex method could be measured empirically. Incidental to the main study, an examination of certain of the constructs of Investment Analysis was necessary(e.g. @ test of the "Random walk hypothesis"). The Markowitz and the Sharpe models have each been developed into a composite routine. The algorithms were compared with programmes utilising more general quadratic programming routines. The Markowitz and Sharpe models, as developed, appeared to be consistently better than other methods. Since the Sharpe model is considerably cheaper to operate than Markowitz's model, it may be concluded that Sharpe's model is the most efficient of those methods tested.