سیستم پشتیبانی از تصمیم گیری برای تخصیص دارایی استراتژیک

Strategic asset allocation is a crucial activity for any institutional or individual investor. Given a set of asset classes, the problem concerns the definition and management over time of the best asset mix to achieve favorable returns subject to various uncertainties, policy and legal constraints, and other requirements. Although a considerable attention has been placed by the scientific community to address this problem by proposing sophisticated optimization models, limited effort has been devoted to the design of integrated framework that can be systematically used by financial operators. The paper presents a decision support system which integrates simulation techniques for forecasting future uncertain market conditions and sophisticated optimization models based on the stochastic programming paradigm. The system has been designed to be accessed via web and takes advantages of the increased computational power offered by high performance computing platforms. Real-world instances have been used to assess the performance of the decision support system also in comparison with more traditional portfolio optimization strategies.

Optimization-based decision support systems (DSS) provide a powerful tool in many industrial and operative contexts. By the formulation of optimization models it is possible to mathematically represent complex decision making problems and control, by the corresponding solutions, critical issues such as cost, risk, performance indexes and so forth. When encapsulated in decision support systems, the advantage related to the adoption of quantitative methodologies is further enhanced by the recent advances in computer science and information technology. The impact deriving from the adoption of DSS becomes even more evident in all those operative contexts characterized by a high level of complexity due, for example, to the presence of uncertainty, to the need to analyze huge amount of data and to operate in a short time. All these features are dominant in the field of financial management where the adoption of efficient and effective DSS can assist the complex decision making process by improving the quality and the effectiveness of provided solutions. Nevertheless, recent global crisis and spectacular breakdowns have further emphasized the need to complement expertise and experience with the suggestions provided by advanced systems based on the integration of decision models and algorithms. The paper contributes in this direction by proposing a system designed to support financial operators in portfolio management activities. Different applications, from personal financial planning, to pension fund management, to strategic and tactical asset allocation can be casted as Asset–Liability Management problems (ALM) (see the contributions in [12], [14], [33] and [38]). In this paper, the attention is focused on the strategic asset allocation problem (SAA, for short). Given a time horizon, the problem refers to the allocation and management of a portfolio of asset classes with the aim of maximizing the portfolio return while controlling the risk exposure. SAA is a crucial problem that every investor (at individual or institutional level) has to deal with. Its practical relevance is the motivation behind the considerable attention devoted to this topic by the scientific community. Starting from the Markowitz's seminal contribution [27] based on a static buy and hold portfolio strategy, other modeling frameworks [6], [9], [26], [28], [29] and [32] have been proposed and deeply investigated in the last decades. A special attention deserves the stochastic programming framework (SP, for short) [8] and [22] because of its high modeling power stemming from the integration of dynamic and uncertain aspects also accounting for the possibility to accommodate specific requirements (policy restrictions, trading constraints, and so forth). On the other hand, the complexity of the SP methodology has limited its wide diffusion in the non-academic environment. The considerations reported above represent the main motivation of the present contribution aimed at proposing a user-friendly DSS relying on the integration of SP as decision engine with simulation techniques for the generation of future uncertain market conditions and solution algorithms powered by the adoption of advanced computing platforms. Other DSS designed for financial management problems have been proposed in the last decade. A system which integrates multi-dimension databases, on-line analytical processing (OLAP) tools, procedural and declarative modeling languages has been proposed in [15]. In [30] the authors have proposed DSSALM, a tool to simulate interest rate risk and forecast the amounts of future assets and liabilities. Another contribution is [16], where the authors propose a web-based DSS which integrates portfolio management models with OLAP tools to handle multidimensional data structures and PVM network environment as high-performance computational framework. The proposed DSS differs from other systems for the following features: •the system core relies on innovative methodologies for the SAA problem and takes advantage of the computational power offered by high-performance computing environment. This last feature is particularly important for applications which are “space”-sensitive: a huge amount of data should be stored, managed and analyzed in order to get robust recommendations. In addition “time” can play a key role in terms of competitive advantage; • the DSS provides a tool for risk management and allows benchmark analysis with other portfolio strategies; •the system can be easily extended to other problems and functionalities, thanks to its modular structure; •customers can access the system by a user-friendly interface and can easily modify parameters and settings; •no investment or maintenance cost for final users is required because of the web technology used. The rest of the paper is organized as follows. Section 2 describes the SAA problem and provides a high level snapshot of the methodological framework. Section 3 presents the system architecture, by illustrating the main technical choices and the functional modules which implement the proposed methodological steps. Section 5 presents a system demo from an end-user perspective. Computational experiments are reported in Section 6: extensive numerical tests have been carried out in order to assess the performance of the proposed system also in comparison with other approaches. Some concluding remarks end the paper.

The paper presents an integrated approach to support financial operators in the strategic asset allocation process. This is a crucial activity very difficult to carry out because of the intrinsic stochastic and dynamic nature of the problem. In addition, the selection of a portfolio of financial instruments has to generally satisfy a large number of constraints of different nature. Starting from the seminal contribution of Markowitz, a significant effort has been devoted by the scientific community to propose sophisticated mathematical models able to encompass all the relevant features of the problem. Starting from static mathematical models based on the use of the variance as risk measure, dynamic and stochastic models based on the stochastic programming paradigm have been proposed. Particular attention has also been devoted to the choice of appropriate risk measures. However, the existing methodologies are not widely used in practice by financial operators because they are too complex and require a lot of input data that should be preliminarily processed. The paper presents a DSS which implements sophisticated mathematical models and integrates simulation and optimization techniques. The complexities of the system core are hidden to the end-user that can access via web by a user-friendly interface. Comparison with other portfolio optimization strategies is also provided. This represents an important feature for a financial operator who can capitalize his/her own expertise in the field. In addition, the DSS has been designed to take advantage of the increased computational power offered by high-performance platforms. Besides a concurrent use of the system, the implementation on a HPC infrastructure allows to obtain more robust and reliable solutions in shorter times. Finally, the system can be easily modified to account for a wider class of decision problems of the ALM family and can be improved in terms of functionalities for a more general support to the investment planning activity.