Multicriteria portfolio construction with Python (Q2154552)

Actually the authors are practitioners in portfolio management, more specifically multi criteria decision analysis (MCDA). For such a purpose, they used Python programming language. The aim of this book is to link theory and practice in such an area. The target readers of this book are ``diversified'', including fund managers, risk managers, investment advisors, bankers, private investors, analytical scientists, operations researchers scientists, and computer engineers, just to name several. But this book could also be a text book for advanced undergraduate or post-graduate students in investment analysis, portfolio engineering, decision science, computer science, or financial engineering. There are eight chapters, going from a brief introduction to Chapter 8, which summarises the conclusions of the book. An appendix provides some experimental results on concrete data, e.g. NYSE securities, NASDAQ, Paris Stock Exchange, and so on. Chapter 2 starts with the introduction of used terms concerning portfolio expected return and portfolio risk. The risk is to be detailed according to whether the securities are correlated or not. Then \textit{H. Markowitz}'s method [``Portfolio selection'', J. Finance 7. No. 1, 77--91 (1952; \url{doi:10.1111/j.1540-6261.1952.tb01525.x})] (mean-variance model) is introduced, dealing with the so-called ``efficient portfolio'' versus ``efficient frontier''. This method is based on two criteria, which are not efficient to incorporate other goals than performance and risk, and to take in account the investors' risk tolerance. So the authors propose more flexible methods. Thus, Chapter 3 introduces some new methods to care for the inadequacies of the previous mean-variance model. Multi-criteria decision analysis (MCDA) is introduced: discrete multi-criteria decision; continuous optimal methods; introduction to multi-objective programming problems. Chapter 4 provides a review of published research in the fields of portfolio management and MCDA: a wide set of references are classified in three lists. The first one concerns discrete MCDA, with four different approaches (Promethee, Topsis, Electree, AHP); the second one, the multi-objective mathematical programming (MMP); and finally the goal programming (GP). In Chapter 5, a construction methodology is developed, mainly based on two phases. The first one involves the multi-criteria portfolio selection problem, based on four different methodologies (ELECTRE, PROMETHEE, MAUT, TOPSIS). The second one concerns the problem of portfolio optimization. The concrete utilization of Python is described in Chapter 6: the architecture of the system, the programming language and the library are exposed as the MCDA platform, and finally the use of Python is detailed according to the different methodologies previously proposed (ELECTRE, PROMETHEE, MAUT, TOPSIS). Chapter 7 provides a use of the concerned methodology on concrete data, namely ``empirical testing'': NYSE, NASDAQ, Paris, Tokyo, on four calendar years, split in three areas (technology, energy, financial). A cumulative ranking of the securities is produced from the first phase. Then the second phase goes to portfolio optimization. Finally, to evaluate the accuracy of the methodology, a validation process is done, based on out-of-sample data. Finally, the input data and results of this empirical testing are provided in the appendix.