Grid Technology in Financial Planning - A Methodology for Portfolio Structuring

This report outlines the development of a grid-enabled computational methodology for solving the portfolio optimization problem. Unlike the classical treatment of a set of mean-variance optimized portfolios as a deterministic efficient frontier, we adopt a novel treatment of the frontier as a distribution of points on a statistical front. This is made possible by our stochastic algorithms, specifically an evolutionary programming (EP) approach that is able to derive meaningful portfolios for the statistical front. More importantly, the EP approach is inherently a parallel search algorithm, making it amenable to exploit the immense computational resources that a grid computing architecture can provide. In order to derive statistically meaningful front, one requires a great number of points to form the efficient band. With a grid computing framework, one is able to farm out such computationally expensive tasks and derive "interesting" portfolios to create the band. Such a capability lays the necessary backbone for developing further analytic methodology for portfolio rebalancing, an important aspect of financial planning. This is significant since in recent years, financial planning has permeated into the mindset of the population. The level of awareness is set to increase with the population's growing affluence.