An integrated generalized TODIM model for portfolio selection based on financial performance of firms
Introduction
The modern portfolio theory was originally instituted by Markowitz [1] in 1952. The classical mean–variance (MV) model is also viewed as a pioneer of portfolio selection. Portfolio selection is a problem in the designing of an optimal allocation of limited funds among several alternatives. It is closely related to our lives and has wide application prospects in project portfolio selection [2], [3], [4], energy research and development [5], [6], and the stock market [7]. Different academic studies, such as behavioral finance [8], operational research [9], and intelligent optimization technology [10], have also been conducted to boost the development of portfolio selection [11]. Among these methods, the multi-criteria decision-making (MCDM) method is a useful tool because investors tend to analyze the characteristics and evaluate alternatives with several objectives or criteria in portfolio selection.
MCDM is an active research field in operational research that aims to address decision-making problems that involve multiple criteria, and it can primarily satisfy the inherent multi-criteria nature of portfolio selection problems [12]. Under the MCDM framework, besides the two basic criteria of return and risk, one can consider several other crucial financial criteria, such as the return on equity and net profit margin. Furthermore, MCDM has the advantage of considering the preferences of any particular investor. Thus, there is potential for more realistic models to be built by exploiting the MCDM for the portfolio selection problem [13]. Portfolio selection in the MCDM framework mainly includes two stages: financial performance evaluation of the firms and portfolio allocation of the stocks. Firms’ financial performance is of great interest to institutional investors, led by funds, social insurance security, and securities. This feature is consistent with the notion of value investment. Financial performance generally refers to the financial ratios of firms [13], [14], [15]. The financial statement offers guiding significance to investors in acquiring information on financial position, operating results, and investment value. The fusion of all pertinent financial factors can well describe the financial performance of firms. This is in accordance with the MCDM paradigm, which greatly facilitates the portfolio allocation process.
Regarding the multi-criteria context of portfolio selection, various MCDM methods have been used such as the TOPSIS and DEA [16], DEMATEL, ANP, and VIKOR [17], and DEA cross-efficiency [18], [19], [20]. However, most of them are based on linear utility measures and thus, do not consider the influence of decision makers’ (DMs’) psychological factors. In practice, DMs’ psychological factors (e.g., subjective cognition, experimental judgment) notably influence the decision-making results. The TODIM (TOmada de Decisão Iterativa Multicritério) method, initiated by Gomes and Lima [21], is a typical behavioral MCDM method. It offers a robust framework for describing a DM’s subjective psychological behaviors, such as reference dependence, loss aversion, and diminishing sensitivity [22]. In recent decades, the TODIM method has drawn much attention and has been extensively applied in various fields [23], [24], [25]. In 2018, Llamazares [26] proposed a generalized TODIM method using a simplified form of dominance function. Compared to the classical TODIM method, the generalized TODIM method can produce more stable and consistent decision results. The investment decision process involves fluctuations in investors’ psychology and behavior. Therefore, it is suitable to employ the generalized TODIM method to model investors’ subjective psychological behaviors in the portfolio selection process.
In this study, we propose an integrated MCDM framework for portfolio selection based on financial analysis (financial indexes), machine learning (affinity propagation clustering, APC), objective weighting method (entropy weight method), behavioral MCDM model (generalized TODIM method), and an extended MV portfolio selection model (multi-objective optimization programming). The COVID-19 pandemic has actively promoted the development of the medical industry, which has received significant attention from global investors. Thus, we document these methods in detail and apply them to medical portfolio selection in the Chinese stock market. The motivations and solutions are summarized as follows.
- •
First, multidimensional financial data is used in this study. Usually, a single data source obtained from the stock price, such as mean, variance, skewness, and maverick, is employed [27], [28], [29], [30]. However, stock prices are vulnerable to good and bad news, fluctuations in international markets, and economic policies. Using the stock price to assess firm performance is one-sided. The better a firm’s financial performance, the higher is its investment value. Here, data in the financial statements are compiled to assess the firms’ financial performance. The use of a multidimensional dataset enables stock portfolio selection in a broader framework.
- •
Second, the affinity propagation clustering (APC) algorithm is used for feature reduction. There are 22 financial indices in the dataset, and there is always some relativity between them. The feature reduction technique makes the analysis more concise while retaining a relatively high accuracy in showing the original features. APC is a machine learning technique [31] that transforms high-dimensional data into low-dimensional data through an unsupervised learning process [32], [33], [34], [35]. By employing APC as a feature reduction technique, the number of features is effectively reduced.
- •
Third, an integrated generalized TODIM method is applied to the MCDM analysis. The objective entropy weighting method is used to calculate the criteria weights. The generalized TODIM method [26] is applied to derive firms’ financial performance, which has simple mathematics, solid decision theory foundation, flexible dominance functions, and considers the DM’s psychological behaviors [36]. Handling financial performance evaluation using the integrated generalized TODIM method can make the decision result more objective and accurate.
- •
Fourth, a novel multi-objective portfolio selection model considering both financial and stock market performance is designed, which acts as an extension of the classical MV model. The compromise solution is used to solve multi-objective optimization programming. Using generalized TODIM and multi-objective portfolio selection models, we can capture investors’ attitudes in both the evaluation and decision processes, making the results more comprehensive and flexible.
Through the aforementioned analysis, this study develops an integrated portfolio selection model in the MCDM framework and applies it to Chinese medical stock portfolio selection. The main contributions of this study are as follows:
- •
A multi-source decision framework that includes financial analysis, machine learning, behavioral decision, and portfolio optimization is constructed for portfolio selection. This framework can solve decision and evaluation problems with multi-source data processing, high-dimensional data analysis, and behavioral decision-making modeling, which has strong practicability and expansibility.
- •
Analyzing the financial performance of a firm is conducive to the long-term investment philosophy. Normalizing these financial indices using different piecewise linear value functions conforms to their characteristics. Feature reduction can reduce computational complexity while retaining a suitably high accuracy in presenting the original criteria.
- •
The novel MV portfolio selection framework based on the generalized TODIM method provides meaningful lessons for solving portfolio selection problems using many other MCDM methods. The proposed framework is an organic combination of behavioral decision theory and portfolio optimization, and has many potential applications in the field of economic management.
The remainder of the paper is organized as follows. Section 2 offers a literature review. Section 3 describes the basic methods employed in this study. In Section 4, a novel multi-objective portfolio selection model combining generalized TODIM and classical MV models, considering financial and stock market performance, is proposed. Section 5 applies the proposed model to the medical stock portfolio selection in the Chinese stock market. The sensitivity and comparative analyses are presented in this section. Finally, Section 6 concludes the study.
Section snippets
Literature review
Portfolio optimization has been extensively studied. Several approaches have been proposed for providing an optimal decision strategy for portfolio selection. These methods include mathematical models, such as uncertain tools [37], [38], behavioral finance [39], [40], and operations research models [41], [42]. In recent years, with the popularization of artificial intelligence technology, several deep learning approaches such as recurrent neural network (RNN) [43], variational RNN [44], and
Affinity propagation clustering algorithm
Affinity propagation clustering (APC) algorithm is a well-known unsupervised clustering algorithm initialed by Frey and Dueck [31]. Unlike classical K-means [56] and fuzzy C-means [57] clustering methods, APC does not need to set the cluster number in advance. In APC, any sample dataset is a potential clustering center, and the initial cluster center can be selected arbitrarily. APC can avoid the influence of outliers to the greatest extent possible, thus improving the reliability of the
The evaluation criteria system for the portfolio selection
Xidonas et al. [12] noted that financial ratios are the most appropriate criteria for evaluating corporate performance. To discover the prevalently used criteria in MCDM models for financial problems, Almeida-Filho et al. [59] conducted an in-depth analysis of the main financial criteria by conducting a systematic literature review (SLR). The 17 criteria categories are summarized in Fig. 2.
The SLR results in [59] show that profitability exists in 446 of the articles analyzed. Liquidity and debt
Background and problem description
The COVID-19 pandemic and global economic recession have shrunk world trade and financial markets, and many industries are losing competitiveness. However, the medical industry has received significant attention from global investors over two years. This industry plays a critical role in preventing the spread of the novel coronavirus, relieving health disasters, protecting public health, improving quality of life, and promoting economic development and social progress. To explore the effective
Conclusions, limitations, and future studies
With the increasing maturity and benign regulations of the Chinese stock market, value investment has been accepted by many institutional and individual investors. Long-term stock market value investments also show superior returns than cash deposits. This study proposes an integrated MCDM framework to solve portfolio selection based on the financial performance of firms. Our model comprehensively addresses data collection, data processing, behaviors modeling, and portfolio optimization. First,
CRediT authorship contribution statement
Qun Wu: Methodology and model design, Formal analysis, Writing & editing. Xinwang Liu: Validation, Funding acquisition, Review & editing, Supervision. Jindong Qin: Validation, Funding acquisition, Review & editing, Supervision. Ligang Zhou: Validation, Funding acquisition, Review & editing. Abbas Mardani: Validation, Review & editing. Muhammet Deveci: Validation, Review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors are thankful to the Editor in Chief, Professor Jie Lu, associate editor and anonymous reviewers for their valuable comments and constructive suggestions with regard to this paper. The work was supported by National Natural Science Foundation of China (Nos. 72071045, 72071151, 71771051, 72171002, 71701158), Scholarship from China Scholarship Council (202106090193), Natural Science Foundation for Distinguished Young Scholars of Anhui Province, China (1908085J03), Natural Science
References (71)
- et al.
Modeling project preferences in multiattribute portfolio decision analysis
European J. Oper. Res.
(2017) - et al.
Pharmaceutical R&D project portfolio selection and scheduling under uncertainty: A robust possibilistic optimization approach
Comput. Ind. Eng.
(2021) - et al.
Combining multiple criteria analysis, mathematical programming and Monte Carlo simulation to tackle uncertainty in research and development project portfolio selection: A case study from Greece
European J. Oper. Res.
(2021) - et al.
A portfolio model for siting offshore wind farms with economic and environmental objectives
European J. Oper. Res.
(2018) - et al.
Robust portfolio decision analysis: An application to the energy research and development portfolio problem
European J. Oper. Res.
(2020) - et al.
Grouped data, investment committees & multicriteria portfolio selection
J. Bus. Res.
(2021) - et al.
Behavioral mean-variance portfolio selection
European J. Oper. Res.
(2018) - et al.
Clustering and portfolio selection problems: A unified framework
Comput. Oper. Res.
(2020) - et al.
Assistant reference point guided evolutionary algorithm for many-objective fuzzy portfolio selection
Swarm Evol. Comput.
(2021) - et al.
Portfolio decision analysis: Recent developments and future prospects
European J. Oper. Res.
(2021)
IPSSIS: An integrated multicriteria decision support system for equity portfolio construction and selection
European J. Oper. Res.
A multicriteria methodology for equity selection using financial analysis
Comput. Oper. Res.
A multicriteria DSS for stock evaluation using fundamental analysis
European J. Oper. Res.
Greenness index: IPO performance and portfolio allocation
Res. Int. Bus. Finance
A fuzzy hybrid project portfolio selection method using data envelopment analysis, TOPSIS and integer programming
Expert Syst. Appl.
Combined DEMATEL technique with a novel MCDM model for exploring portfolio selection based on CAPM
Expert Syst. Appl.
Use of DEA cross-efficiency evaluation in portfolio selection: An application to Korean stock market
European J. Oper. Res.
Improving DEA cross-efficiency optimization in portfolio selection
Expert Syst. Appl.
Regret theory-based fuzzy multi-objective portfolio selection model involving DEA cross-efficiency and higher moments
Appl. Soft Comput.
A DEMATEL-TODIM based decision framework for PV power generation project in expressway service area under an intuitionistic fuzzy environment
J. Cleaner Prod.
Pythagorean fuzzy VIKOR approaches based on TODIM for evaluating internet banking website quality of ghanaian banking industry
Appl. Soft Comput.
An analysis of the generalized TODIM method
European J. Oper. Res.
A mean-maverick game cross-efficiency approach to portfolio selection: An application to Paris stock exchange
Expert Syst. Appl.
Data envelopment analysis based fuzzy multi-objective portfolio selection model involving higher moments
Inform. Sci.
A hybrid approach for portfolio selection with higher-order moments: Empirical evidence from shanghai stock exchange
Expert Syst. Appl.
Energy efficiency evaluation and energy saving based on DEA integrated affinity propagation clustering: Case study of complex petrochemical industries
Energy
A novel DEACM integrating affinity propagation for performance evaluation and energy optimization modeling: Application to complex petrochemical industries
Energy Convers. Manage.
An affinity propagation clustering based particle swarm optimizer for dynamic optimization
Knowl.-Based Syst.
Energy efficient building envelope using novel RBF neural network integrated affinity propagation
Energy
A linguistic distribution behavioral multi-criteria group decision making model integrating extended generalized TODIM and quantum decision theory
Appl. Soft Comput.
A new approach to the bi-criteria multi-period fuzzy portfolio selection
Knowl.-Based Syst.
Portfolio optimization using higher moments in an uncertain random environment
Inform. Sci.
Portfolio optimization with behavioural preferences and investor memory
European J. Oper. Res.
Utilizing dependence among variables in evolutionary algorithms for mixed-integer programming: A case study on multi-objective constrained portfolio optimization
Swarm Evol. Comput.
A parallel variable neighborhood search algorithm with quadratic programming for cardinality constrained portfolio optimization
Knowl.-Based Syst.
Cited by (49)
The open banking era: An optimal model for the emergency fund
2024, Expert Systems with ApplicationsA single-valued neutrosophic CIMAS-CRITIC-RBNAR decision support model for the financial performance analysis: A study of technology companies
2024, Socio-Economic Planning SciencesPortfolio optimization based on the pre-selection of stocks by the Support Vector Machine model
2024, Finance Research LettersGeneralized TODIM method based on symmetric intuitionistic fuzzy Jensen–Shannon divergence
2024, Expert Systems with ApplicationsReformative ROCOSD–ORESTE–LDA model with an MLP neural network to enhance decision reliability
2024, Knowledge-Based Systems