Tracker funds offer an attractive balance between risk and return, by providing the profit of the index, with the reduced risk associated with the broad market cover. An effectively designed tracker fund will achieve best tracking of the index with minimal running and trading costs. This paper aims to investigate the use of improved optimisation methods for the design and maintenance of tracker funds.
Most current methods of tracker fund optimisation use quadratic programming (QP), due to its simple formulation and efficient solution. However, the explicit tracking of the return of the index and the optimal selection of the subset of shares composing the fund is not directly available using these methods. This paper investigates ways to overcome the shortcomings of current methods by using genetic algorithms (GA). A GA based tracker fund optimisation method is applied to Financial Times Stock Exchange 100 data using computer simulations.
Tracking performance is presented and compared to QP. Results show the advantage of the new method for various conditions of tracker fund subset size and update rates. In particular, there is an improved performance when evaluating the errors in optimising returns of the index.
The paper intentionally sets out to use commercially available software to implement the optimisation approaches, thus demonstrating that the advantages of using GAs are easily realisable and do not require tailor made software.
The paper provides a direct comparison between the established approach of QP and a GA. The implementation uses commercially available software and is therefore easily realisable in practice.
CitationDownload as .RIS
Emerald Group Publishing Limited
Copyright © 2006, Emerald Group Publishing Limited