Alternative Approaches to Portfolio Selection

The main objective of this thesis is to present and implement two alternative approaches to portfolio selection alongside the classical mean-variance portfolio selection model developed by H. Markowitz. Furthermore, we aim to perform the portfolio construction using the Python programming language and then compare the performance of constructed portfolios over a 17 months investment period. We implement the standard mean-variance approach, its simplification that uses the Single-Index model together with the Capital Asset Pricing Model and a second alternative approach based on the hierarchical clustering for which we develop two portfolio selection strategies that we then combine with several well-known clustering methods. We found that the best portfolio of the standard mean-variance approach can outperform the original mean-variance framework's best portfolio. Moreover, the first strategy based on hierarchical clustering resulted in volatile portfolios that did not provide adequate returns to compensate for the risk associated with them. The second strategy provided stable results that were comparable with the mean-variance framework, yet the construction of its models was much simpler.