Elsevier

Knowledge-Based Systems

Volume 58, March 2014, Pages 86-97
Knowledge-Based Systems

Combining VIKOR-DANP model for glamor stock selection and stock performance improvement

https://doi.org/10.1016/j.knosys.2013.07.023Get rights and content

Abstract

This study proposes a multiple attributes decision making (MADM) method for solving glamor stock selection problem based on fundamental analysis. Traditional analyzes rely on choosing key financial ratios in making comparison, or by observing the trends of change in various financial variables (also termed as criteria or signals in this study). However, most of the criteria for stock selection have inter-dependent/interactive characteristics. In practice, investors often have to make compromising decisions when target stocks indicate conflicting performance outcomes in different criteria. Traditional methods have difficulty in making decision while facing inter-dependent criteria and compromise alternatives. Thus, this study proposes a combined MADM method to retrieve financial experts’ knowledge for glamor stock selection. The proposed method not only helps to identify the ideal glamor stock, but the pertaining insights may also be used for the management teams of glamor stocks to prioritize their improvement plans. In addition, this study provides an empirical case in analyzing five glamor stocks of semiconductor industry in Taiwan. The result indicates that the proposed method for glamor stock selection is effective and provides meaningful implications for investors and management teams to refer. The selected top ranking stock consistently outperformed the other four glamor stocks in 32 month and 44 month holding periods from May 2009 to December 2012 with statistical significance, which indicated the effectiveness of the proposed model.

Introduction

In this study, we aim to propose a combined multiple attributes decision making (MADM) method to deal with the glamor stock selection problem, which has caused increasing interest in both academic and practical fields [1]. The stock selection problem was stemmed from the efficient market hypothesis (EMH), and it could be traced back to the original work of efficient capital markets [2]. From the assumption of EMH, investors cannot use available information to formulate a strategy that consistently outperforms the market [3], [4], [5]. To examine the EMH, some previous studies tended to apply various investing styles to test the hypothesis. Value investing and growth stock (also termed as glamor stock in financial literature) investing are two of the commonly examined strategies [6], [7]. These two investing strategies use simple valuation criteria separately to form different investment portfolios. The value stock portfolio was found to outperform the diversified portfolio or market index in the long-term [7], which has gathered attention from both academic and practical fields. But the emerging recognition towards growth stock investing strategy also caught investors’ interests in practice. From the definition of previous study [8], glamor stocks are expected by market to perform well in the future, embracing relatively low B/M (book to market value) or E/P (earning to price) ratio. Although the discussion of growth stock investing in academic research has formed a foundation for investment practice [9], while making real investment decision, investors still have to make selection among the growth stock portfolio because of limited capital constraint. In making stock selection, fundamental analysis (FA) is a main approach in evaluating a stock for its financial worthiness by using historical data [10]. However, relatively little attention has been given to the use of FA for glamor stock investment. Thus, this study focuses on examining the proposed model as a basis for glamor stock selection, and provides directions for the management teams of glamor stocks to improve their stock performance.

Some previous attempts in applying quantitative models were conducted to predict stock performance by using financial data. On one hand, the artificial intelligence (AI) approach is broadly applied, such as the artificial neural network (ANN) technique often uses financial data as input variables to predict future stock performance [11], [12], [13]. To retrieve rules for stock price prediction, integrated fuzzy system and ANN was also introduced [14]. Despite the superior modeling capability of the ANN, the black-box processing characteristic often impedes the way to gain insights of the influential effects for investors to comprehend. On the other hand, the multiple attributes decision-making (MADM) approach has been considered to solve the stock selection problem recently. One of the mainstreams is the data envelopment analysis (DEA) approach, which has gained research interests for the portfolio selection problem [15]. Although the DEA method seems to provide a useful tool for portfolio selection, it still lacks of capability to explore the relationships among financial variables for more in-depth analysis. In order to help fill the gap, this study proposes a combined MADM model to solve the glamor stock selection problem, aiming at exploring the plausible interrelationships among the considered criteria for making better decision.

The main concern of glamor stocks is that they are relatively expensive, if the glamor stocks fail to maintain the market’s confidence, it often leads to significant price corrections. Thus, the glamor stocks selection is more challenging compared with the generic stock selection problem. Research has shown that stock market tends to naively extrapolate recent fundamentals of glamor stocks [16], which attributed to cognitive biases [6]. To consider the “naïve extrapolation” effect of growth stock investing, Mohanram [1] included this dimension (naïve extrapolation) in his proposed G-score model with the other financial criteria. In this study, we follow the conceptual framework of the traditional G-score model to regard “naïve extrapolation” as a major dimension for glamor stock selection. Although the findings of the G-score model contributed to identify relevant financial information regarding growth stock investing, the model chose a simplified approach to transform relevant financial information into binary inputs, and formed a logistic regression model to separate gainers from losers. In the regression model of the G-score, dependent variables were set to be either one or zero, comparing a firm’s financial ratios with its contemporary industry average as signals. Although the G-score model showed the possibility to use FA for growth stock investing, it still has some limitations and potential drawbacks.

To improve the limitations of previous studies [1], [10], [11], [12], [13], this study proposes a combined MADM model for four advantages. Firstly, the designed transformation of performance scores for each stock in different criteria (see Section 4) may measure the relative strength of each stock more adequately. The traditional G-score model cannot measure the distance of a stock’s current performance with its industry average in each criterion. The extreme values cannot be reflected in a proportional manner. For example, suppose the average ROA (return over total asset) of the electronic industry is 1%, and the current ROA of Stock A and B are 19% and 4% respectively. In the G-score model, both of the two stocks’ ROA signals will be set to one; but in reality, the Stock A’s ROA is about five times better than the Stock B’s result. In our proposed model, this drawback can be overcome by measuring the distance between the aspired/ideal level and current status of financial variables for each stock. Secondly, the logistic regression model adopted by previous research [1] assumed that variables are independent, which is not realistic in practice. The proposed Decision-Making Trial and Evaluation Laboratory (DEMATEL) method applied in this study may preserve the interrelationships among financial variables/criteria, which could be used to adjust the influential weights based on the basic concept of the analytical network process (ANP), termed as the DANP (DEMATEL-based ANP) in this study [17]. Thirdly, the proposed model can decompose criteria into cause and effect groups for gaining more insights. Compared with previous studies that used the DEA or the ANN technique [10], [11], [12], [13], the proposed model can explore the interrelationships among criteria, provides more information for management teams to make strategic plans. Lastly, the integration of VIKOR method can provide the priority of improvement gaps for growth stocks, which may be regarded as a managerial tool to probe the direction for improvement to increase growth stocks’ market value.

The rest of the study is organized as follows: in Section 2, this research reviews the concept of FA and growth stock investing. Also, some related MADM methods for solving the stock selection problem are introduced, including DEMATEL, DANP and VIKOR that are applied in the proposed model. In Section 3, a combined MADM framework is proposed for solving the stock selection problem. Section 4 provides an empirical example to rank the five selected glamor stocks and compare their stock performance with the output that our model suggests. Section 5 discusses the interrelationships of the evaluated dimensions and criteria, and provides implications to conclude this study.

Section snippets

Preliminaries

This section briefly reviews the concepts adopted by the research, such as FA, growth stock investing and related MADM methods for the stock selection problem.

Constructing the combined VIKOR-DANP model

This section introduces the conceptual framework of the G-score model for growth stock investing and the proposed hybrid methods, including DEMATEL technique (to form the relationship-structure model), DANP (find out influential weights for criteria with dependence and feedbacks from DEMATEL technique using the basic concept of ANP) (see Appendix A Stage1), and VIKOR (to form ranking scores and explore priority to reduce gaps for achieving the aspired level) (see Appendix A Stage2).

Empirical case for evaluating glamor stocks in semiconductor industry

In this section, an empirical study is presented to illustrate the application of the proposed VIKOR-DANP model to evaluate glamor stocks from Taiwan’s world-leading semiconductor industry as an empirical case.

Conclusions and remarks

In summary, three main outcomes were achieved: (1) The proposed VIKOR-DANP model for glamor stock selection reduced the limitations of the conventional regression models; the independence assumption (among variables) could be removed, and the probability distribution of the dependent variable does not need to be assumed; (2) In the proposed model, Stock A was selected from the five target stocks and outperformed the other four stocks in the subsequent holding period (May 2009 to December 2012),

References (35)

  • C.R. Wu et al.

    Evaluating business performance of wealth management banks

    European Journal of Operational Research

    (2010)
  • S. Opricovic et al.

    Compromise solution by MCDM methods: a comparative analysis of VIKOR and TOPSIS

    European Journal of Operational Research

    (2004)
  • P.S. Mohanram

    Separating winners from losers among low book-to-market stocks using financial statement analysis

    Review of Accounting Studies

    (2005)
  • E.J. Fama

    Efficient capital markets: a review of theory and empirical work

    Journal of Finance

    (1970)
  • R. Bird et al.

    The case for market inefficiency: investment style and market pricing

    Journal of Asset Management

    (2005)
  • E.F. Fama et al.

    Multifactor explanations of asset pricing anomalies

    Journal of Finance

    (1996)
  • E.F. Fama et al.

    The capital asset pricing model: theory and evidence

    Journal of Economic Perspectives

    (2004)
  • Cited by (84)

    • A model for evaluating green credit rating and its impact on sustainability performance

      2021, Journal of Cleaner Production
      Citation Excerpt :

      The neutrosophic sets powerfully handled the indeterminate and inconsistent data through including three aspects which are truth, indeterminate, and false degrees (Nabeeh et al., 2019b). Recently, the integration of MCDM methods used in complicated problems with various criteria and alternatives (Alam-Tabriz et al., 2014; Kuan and Chen, 2014; Shen et al., 2014). The proposed N-MCDMF integrates the neutrosophic with MCDM to support the decision-makers with a highly systematic framework in the uncertain and inconsistent environmental conditions.

    • Analytic network process: An overview of applications

      2020, Applied Mathematics and Computation
    • An uncertain possibility-probability information fusion method under interval type-2 fuzzy environment and its application in stock selection

      2019, Information Sciences
      Citation Excerpt :

      Markowitz [8] originated the mean-variance model, which is known as the foundation of modern portfolio theory. Other studies discussed the limitations of measuring investment risk, and developed a lot of multiple criteria decision making(MCDM) methods [9–12]. For the stock selection problems with MCDM method, Chen [13] applied data envelopment analysis (DEA) models to select stock.

    View all citing articles on Scopus
    View full text