Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1598-7248 (Print)
ISSN : 2234-6473 (Online)
Industrial Engineering & Management Systems Vol.17 No.3 pp.464-478
DOI : https://doi.org/10.7232/iems.2018.17.3.464

Credibilitic Multiperiod Mean Semivariance Portfolio Selection with Transaction Costs

Peng Zhang*, Bi-Yu Peng
School of Economics and Management, South China Normal University, Guangzhou. China
Corresponding Author, E-mail: zhangpeng300478@aliyun.com
September 13, 2017 May 21, 2018 June 7, 2018

ABSTRACT


This paper proposes a credibilitic multiperiod mean-simivariance portfolio selection model with transaction costs, borrowing constraints, threshold constraints and cardinality constraints. In this model, the return rate of asset and the risk are quantified by credibilitic expected value and credibilitic semivariance, respectively. Threshold constraints limit the amount of capital to be invested in each stock and prevent very small investments in any stock. Based on credibilitic theories, the model is transformed into a mix integer dynamic optimization problem with path dependence. The proposed model is approximated to dynamic programming model. A novel discrete iteration method is designed to obtain the optimal portfolio strategy, and is proved linearly convergent. Finally, an example is given to illustrate the behavior of the proposed model and the designed algorithm using real data from the Shanghai Stock Exchange.



초록


    1. INTRODUCTION

    Markowitz (1952) originaly presented mean variance single-period portfolio selection. In real world, the portfolio strategies are often multi-period. The investor usually needs to rebalance his position from time to time. After Markowitz’s pioneering work, various researchers extended his work to multiperiod case. Among others, the representative works include Li and Ng (2000), who extend the originally mean-variance problem to dynamic discrete-time setting. Li and Ng (2000) embed the original problem into a tractable auxiliary problem, then the analytical portfolio strategy and mean-variance efficient frontier are derived. The further study about dynamic extension of the Markowitz’s model includes, e.g., Zhu et al. (2004) discussed a generalized multiperiod mean-variance model with bankruptcy risk control, and showed that under the classical multiperiod mean-variance portfolio optimization model, the bankruptcies mostly occur in the early periods. Bielecki et al. (2005) study discrete time and continuous-time MV models, respectively. Cui et al. (2014) reformulated the multiperiod mean-variance model, and derived the optimal investment strategy and efficient frontier by using the mean-field approach. Note that all the portfolio models mentioned above do not allow cash additions or withdrawals. Cui et al. (2017) investigate analytically effects of portfolio constraints on time consistency of efficiency for convex cone constrained markets. Li and Li (2012) investigate a multiperiod asset-liability portfolio optimization problem which intends to control the probability of bankruptcy. Gao et al. (2015) considers the time cardinality constrained mean-variance dynamic portfolio selection problem in markets with correlated returns and in which the number of time periods to invest in risky assets is limited. Wu and Zeng (2015) studid a generalized multiperiod mean variance port-folio selection problem within the game theoretic framework for a defined-contribution pension scheme member.

    All previous literatures are proposed on the basis on probability theory. They view the uncertainty associated with risky assets as random uncertainty. If using probabilistic approaches to handle the uncertainty of the financial markets, it is necessary to estimate the probability distribution of return for each risky asset. Due to the everchanging financial markets, the probability distribution of return of risky asset usually cannot be predicted accurately. Even though the probability distribution can be estimated, there is no guarantee that the return rates truly obey it. Actually, in real world, there are many non-random factors that affect the financial markets such as economics, policies, laws, regulations and people’s psychological factors. In particular, the influence of people’s psychological factors cannot be neglected in portfolio decision-making. In many cases, investors are usually provided with information which is characterized by vague linguistic descriptions such as high risk and low profit. Based on the facts mentioned above, to estimate the risky assets returns, it is necessary to take experts’knowledge and investors’subjective opinions into consideration. A possibility distribution can be regarded as an alternative to a probability distribution, in which the un-quantifiable factors can be reflected easily. In fact, in many situations, it might be easier to estimate the possibility distributions of returns on risky assets than the corresponding probability distributions. In contrast to the probability theory, fuzzy set theory needs less information in the decision process. Thus, it is worthwhile to use fuzzy set theory to investigate the uncertainty of financial markets. Following the widely used fuzzy set theory in Zadeh (1999), researchers have used the fuzzy set theory to investigate portfolio selection problems under uncertain environment. Numerous models have been proposed by using different approaches. Li et al. (2010) proposed several fuzzy mean variance skewness models in fuzzy environment. Qin (2017) proposed a random fuzzy mean absolute deviation portfolio selection model. Some researchers extend the single-period fuzzy portfolio selection into multiperiod setting. By using experts’judgments, Sadjadi et al. (2011) formulated a fuzzy multiperiod portfolio selection model with different borrowing and lending rates. Zhang et al. (2012, 2014), and Liu et al. (2012, 2013) respectively proposed several kinds of multiperiod fuzzy portfolio selection models. Zhang and Zhang (2014) proposed a multiperiod mean absolute deviation fuzzy portfolio selection model with cardinality constraints.

    Though possibility measure has been widely used in portfolio selection, it has limitation. One great limitation is that possibility measure is not self-dual. Using possibility measure which has no self-duality property, we can find that two fuzzy events with different occurring chances may have the same possibility value. These results are quite awkward and will confuse the decision maker. Thus, Huang (2008) presented mean-semivariance model based on the self-dual credibility measure. Li et al. (2010) proposed cross-entropy minimization credibilitic model and so forth, Zhang et al. (2014) proposed a credibilitic multi-period mean-variance portfolio selection model. Mehlawat (2016) proposed a multiperiod mean entropy credibilitic portfolio selection model.

    In many existing models are proposed on the framework of Markowitz’s mean variance with cardinality constraints, threshold constraints and so on. These constraints come from real-world practice where the administration of a portfolio made up of many assets is clearly not desirable because of transaction costs, complexity of management, or policy of the asset management companies. Because of its practical relevance, this cardinality constrained Markowitz model, and some variations have been intensively studied in the last decade. Especially from the computational viewpoint, some researchers proposed exact solution methods (i.e., Bertsimas and Shioda, 2009; Li et al., 2006; Shaw et al., 2008; Murray and Shek, 2012; Cesarone et al., 2013; Cui et al., 2013; Sun et al., 2013; Le Thi et al., 2009; Le The and Moeini, 2014). Since exact solution methods are able to solve only a fraction of practically useful cardinality constrained Markowitz models, many heuristic algorithms have also been proposed (ie., Fernández and Gómez, 2007; Ruiz-Torrubiano and Suarez. 2010; Anagnostopoulos and Mamanis, 2011; Woodside-Oriakhi et al., 2011; Deng et al., 2012). In these studies, it appears that the computational complexity for the solution of the cardinality constrained Markowitz model is much greater than the one required by the classical Markowitz model or by several other of its refinements. Indeed, the standard Markowitz model is a convex quadratic programming problem, while the cardinality constrained Markowitz model is a mixed integer quadratic programming problem that is a more difficult NP-hard problem.

    The contribution of this work is as follows. We pro-pose a new credibilitic multiperiod mean semivariance portfolio selection model with borrowing constraints, transaction costs, threshold constraints and cardinality constraints. Using the fuzzy decision-making approach, the proposed model is transformed into a crisp mix integer dynamic optimization problem with path dependence. The proposed model is approximated to dynamic programming model. A novel discrete iteration method is designed to obtain the optimal portfolio strategy, and is proved linearly convergent.

    The rest of this paper is organized as follows. In Section 2, the definitions of the credibilitic mean, the credibilitic semivariance, and some properties are introduced. In Section 3, transaction costs, the threshold constraints and cardinality constraints are formulated into the multiperiod portfolio. A new credibilitic multiperiod portfolio selection model is constructed. The multiperiod portfolio selection is a mix integer dynamic optimization problem with path dependence. The proposed model is approximated to dynamic programming model. A discrete iteration method is proposed to solve it in Section 4. In Section 5, the comparison analysis of differently desired number of risky assets in the portfolio and different preference coefficients are given to illustrate the modeling idea and the effectiveness of the designed algorithm. Finally, some conclusions are given in Section 6.

    2. PRELIMINARIES

    Let ξ be a fuzzy variable with membership function μ, and u and x be real numbers. μξ(x) represents the possibility distribution of ξ. For any given set B, the possibility and necessity measures of ξ ∈ B are, respectively, defined as (see Zadeh, 1999).(1)(2)

    Pos{ ξ B }= sup x B μ ( x )
    (1)

    Nec{ ξ B }=1 sup x B c μ ( x )
    (2)

    where Pos and Nec are the possibility and necessity measures, respectively.

    Note that both possibility measure and necessity measure satisfy the properties of normality, nonnegativity and monotonicity. However, both of them are not self-dual. Based on the possibility measure and the necessity measure, Liu and Liu (2002) defined the credibility measure as follow, which is self-dual measure.

    C r { ξ B }= 1 2 ( sup x B μ ( x ) +1 sup x B c μ ( x ) )
    (3)

    where Cr is self-dual measure and satisfies Cr{ξ ≤ r} + Cr{ξ ≥ r} = 1.

    Definition 1 (Liu and Liu, 2002). Let x be a fuzzy variable, the expected value of x is defined as(4)

    E [ ξ ] = 0 C r { ξ r } d r 0 C r { ξ r } d r
    (4)

    Theorem 1 (Liu and Liu, 2002) Let ξ be a fuzzy variable with finite expected value, and let λ and μ be two real numbers, then(5)

    E ( λ ξ + μ ) = λ E ( ξ ) + μ
    (5)

    Theorem 2 (Liu and Liu, 2002) Let ξ and η be independent fuzzy variables with finite expected values, and let λ and μ be two real numbers, then(6)

    E ( λ ξ + μ η ) = λ E ( ξ ) + μ E ( η )
    (6)

    Definition 2 (Huang, 2008) Let ξ be a fuzzy variable with finite expected value e, then the semivariance and credibilitic semivariance of ξ are defined as(7)(8)

    S V [ x ] = E [ [ ( x e ) ] 2 ]
    (7)

    where

    ( ξ e ) = { ξ e  if  ξ e   0        if  ξ > e S V ( ξ ) = E [ ( ( ξ e ) ) 2 ] = 0 Cr{(( ξ e ) ) 2 r } d r
    (8)

    Theorem 3 Let ξ be a fuzzy variable with finite expected value e,and let λ be any given nonnegative real number, then(9)

    S V ( λ ξ ) = λ 2 S V ( ξ )
    (9)

    Proof. From Definition 2, it follows that

    S V ( λ ξ ) = E [ ( λ ξ λ e ) ] 2 = 0 Cr{ [ ( λ ξ λ e ) ] 2 r }d r = 0 Cr{ [ ( ξ e ) ] 2 r λ 2 }d r = 0 Cr{ [ ( ξ e ) ] 2 t } λ 2 d t = λ 2 0 Cr{ [ ( ξ e ) ] 2 t }d t = λ 2 S V ( ξ )

    Thus, the proof of the Theorem 3 is ended. Furthermore, we can get the following form.

    Cr{ [ ( ξ e ) ] 2 r }= Cr{ [ ( ξ e ) ] 2 r , ξ e } = Cr{ [ ( ξ e ) ] r } = Cr{ ξ e r }

    Thus, the credibilistic semivariance of ξ can be ex-pressed as follows:(10)

    S V ( ξ ) = 0 Cr{ ξ e r }d r
    (10)

    Theorem 4 Let ξ and η be independent fuzzy va-riables with finite expected values. Then, for any nonnegative real numbers λ and μ, it holds that

    S V ( λ ξ + μ η ) = λ 2 S V ( ξ ) + μ 2 S V ( η )
    (11)

    Proof.

    S V ( λ ξ + μ η ) = E [ ( λ ξ + μ η λ E ( ξ ) μ E ( η ) ) ] 2 = 0 Cr{ [ ( λ ξ + μ η λ E ( ξ ) μ E ( η ) ) ] 2 r }d r = 0 Cr{ [ λ ξ + μ η λ E ( ξ ) μ E ( η ) ] 2 r , λ ξ + μ η λ E ( ξ ) + μ E ( η ) }d r = 0 Cr{ [ λ 2 ( ξ E ( ξ ) ) 2 + μ 2 ( η E ( η ) ) 2   + 2 λ μ ( ξ E ( ξ ) ) ( η E ( η ) ) ] 2 r , λ ξ λ E ( ξ ) , μ η μ E ( η ) , λ ξ + μ η λ E ( ξ ) + μ E ( η ) } d r = λ 2 0 Cr{ [ ( ξ E ( ξ ) ) ] 2 r }d r + μ 2 0 Cr{ [ ( μ E ( η ) ) ] 2 r }d r + λ μ 0 Cr{ [ ( ξ E ( ξ ) ) ] [ ( μ E ( η ) ) r }d r     
    (12)

    If ξ and η is independent fuzzy variables,

    0 Cr{ [ ( ξ E ( ξ ) ) ] [ ( μ E ( η ) ) r }d r = 0

    Then the Eq. (12) can be turned into

    S V ( λ ξ + μ η ) = λ 2 S V ( ξ ) + μ 2 S V ( η )

    Thus, the proof of the Theorem 4 is ended.

    If ξ = (a, a, b) be a triangular fuzzy number, then its membership function μξ(x) can be described as:(13)

    μ ξ ( x ) = { 1 x a α , x [ a α , a ] 1 a x β , x [ a , a + β ] 0 ,             otherwise
    (13)

    From Eq. (3), the credibility of the event {ξr} can be represented by

    Cr { ξ r } = { 0 ,             if   r a α r a + α 2 α , i f a α r a r a + β 2 β , i f a r a + β 1 ,               otherwise
    (14)

    Theorem 5 Let ξ = (a, a, b) be a triangular fuzzy number, then the credibilistic expected value of ξ can be given by:(15)

    E ( ξ ) = a + ( b a ) / 4
    (15)

    Proof.

    From the Definition 1, it follows that

    E [ ξ ] = 0 C r { ξ r } d r 0 C r { ξ r } d r = 0 ( 1 C r { ξ r } ) d r = 0 a α 1 d r + a α a ( 1 r a + α 2 α ) d r + a a + β ( 1 r a + β 2 β ) d r + 0 = a + β α 4

    Thus, the proof of the theorem is ended.

    Theorem 6 Let ξ = (a, a, b) be a triangular fuzzy number. Then, semivariance of ξ can be obtained

    S V ( ξ ) = 1 6 β ( β α 4 ) 3 + 1 6 α ( β + 3 α 4 ) 3 1 6 α ( β α 4 ) 3
    (16)

    Proof.

    From the Definition 2, it follows that

    S V ( ξ ) = E [ ( ( ξ e ) ) 2 ]   = 0 Cr{(( ξ e ) ) 2 r } d r = 0 Cr{ ξ e r }d r = 0 Cr{ ξ a + β α 4 r }d r
    (17)

    From Eq. (14), it follows that

    Cr{ ξ a + β α 4 r } = { 0 ,    if   a + β α 4 r a α a + β α 4 r a + α 2 α , i f a α a + β α 4 r a a + β α 4 r a + β 2 β , i f a a + β α 4 r a + β 1 ,               otherwise = { 0 ,                     if   r ( β + 3 α 4 ) 2 β + 3 α 4 r 8 α , i f   ( β α 4 ) 2 r ( β + 3 α 4 ) 2 5 β α 4 r 8 β , i f   0 r ( β α 4 ) 2 1 ,                     , i f   r 0
    (18)

    From Eq. (17) and Eq. (18), it follows that

    Eqmml14.gif

    Thus, the proof of the theorem is ended.

    3. THE MULTIPERIOD PORTFOLIO SELECTION MODEL

    Assume that there are n risky assets and one risk-free asset in financial market for trading. An investor wants to allocate his initial wealth W1 among n+1 assets at the beginning of period 1, and obtains the final wealth at the end of period T. He can reallocate his wealth among the n +1 assets at the beginning of each of the following T consecutive investment periods. Suppose that the return rates of the n risky assets at each period are denoted as triangular fuzzy variables. For the sake of description, let us first introduce the following notations. Let xi0 be the initial investment proportion of risky asset i at period 0; xt be the portfolio at period t, where x t = ( x 1 t , x 2 t , , x n t ) ; x f t be the investment proportion of risk-free asset at period t, where x f t = 1 i = 1 n x i t ; x β f t be the lower bound of the investment proportion of risk-free asset at period t, where xft ³ xbft; Rit be the return of risky asset i at period t; rpt be the return rate of the portfolio xt at period t; rbt be the borrowing rate of the risk-free asset at period t; rlt be the lending rate of the risk-free asset at period t; lit be the preset lower bound value of xit; uit be the preset upper bound value of xit; rNt be the net return rate of the portfolio xt at period t; Wt be the crisp form of the holding wealth at the beginning of period t; cit be the unit transaction cost of risky asset i at period t; K be the desired number of risky assets in the portfolio at period t.

    3.1. Return, Risk and Cardinality Constraints

    In this section, we employ the credibilistic mean value of the return on the portfolio at each period to measure the return of portfolio. The risk on the return rate of portfolio at each period is quantified by the credibilistic semivariance. The cardinality constraints limit the number of risky assets to be held in an efficient portfolio. The return rate of security i at period t, Rit = (ait, a it, bit), is triangular fuzzy variable for all i = 1, …, n and t = 1, … T.

    Most of the brokerage houses provide the opportunity to make an acquisition on different assets by borrowing the money from the brokerage. Some researchers studied the borrowing constraints, for example, Sadjadi et al. (2011) proposed the fuzzy portfolio model with different rates for borrowing and lending. Deng and Li (2012) proposed a mean-variance fuzzy portfolio with borrowing constraints. Zhang and Zhang (2014), Zhang (2016), and Zhang (2017) proposed the fuzzy multiperiod portfolio model with different borrowing and lending rates. Considering the different rates for borrowing and lending, the credibilistic mean value of the portfolio xt = (x1t, x2t, …, xnt)′ at period t (t = 1, …, T) can be expressed as(19)

    r p t = i = 1 n E ( R i t ) x i t = i = 1 n ( a i t + β i t α i t 4 ) x i t + r f t ( 1 i = 1 n x i t )
    (19)

    when 1-(r1t+…+rnt)³ 0, rft = rlt; when 1-(r1t+…+rnt)£ 0, rft = rbt.

    The transaction costs which are entailed by buying or selling assets to adjust the existing portfolio are one of the main concerns for portfolio managers. As mentioned by Arnott and Wagner (1990), and Yoshimoto (1996) ignoring the transaction costs may fail to obtain the efficient portfolio. So Güpınar and Rustem (2007), and Bertsimas and Pachamanova (2008) incorporated transaction costs into consideration to study the multiperiod portfolio selection problem. In this paper, the transaction costs are assumed V-shaped functions of differences between the t th period portfolio xt = (x1t, x2t, …, xnt) and the t-1th period portfolio xt-1 = (x1t-1, x2t-1, …, xnt-1). That’s to say, the transaction cost for asset i at period t (t = 1, …, T) can be expressed as(20)

    C i t = c i t | x i t x i t 1 |
    (20)

    Hence, the total transaction costs of the portfolio x t = ( x 1 t , x 2 t , , x n t ) at period t (t = 1, … T) can be represented as(22)

    C t = i = 1 n c i t | x i t x i t 1 |
    (21)

    The net return rate of the portfolio xt at period t (t = 1, …, T) can be denoted as

    r N t = i = 1 n ( a i t + β i t α i t 4 ) x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 |
    (22)

    The crisp form of the holding wealth at the begin-ning of the period t (t = 1, …, T) can be written as(23)

    W t + 1 = W t ( 1 + r N t ) = W t ( 1 + i = 1 n ( a i t + β i t α i t 4 ) x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 | )
    (23)

    Derived from Eq. (11) and Eq. (16), the credibilistic semivariance of the portfolio xt can be expressed as(24)

    S V t ( x t ) = i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2
    (24)

    To formulate cardinality constraints into the multiperiod portfolio model, zero-one decision variables are added as:(25)

    z i t = { 1   if any of asset  i  of period  t   ( i = 1, , n t = 1, , T ) is held 0   otherwise
    (25)

    where i = 1 n z i t K .

    Let lit and uit be the preset lower and upper bounds values of xit, respectively, The threshold constraints of multiperiod portfolio selection can be expressed as(27)

    l i t x i t u i t
    (26)

    Let the preset value be xbft, where xbft ≤ 0, the bor-rowing constraint of risk-free asset at period t is

    x f t = 1 i = 1 n x i t x f t b
    (27)

    For a rational investor, he/she wishes not only to maximize expected return but also to minimize the risk which is measured by the variance of the rate of return on a portfolio. So he/she must make a tradeoff between the two objectives. Let (1-θ) and θ be the preference coefficients associated with criteria rpt and SVt (xt) respectively. Then the investor attempts to maximize

    F t ( r N t , S V t ( x t ) ) = ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 | ) θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 )
    (28)

    Here the parameter θ sthe risk aversion factor of the investor. The greater the factor θ is, the more risk aversion the investor has. In this paper, we assume that 0≤θ≤1.

    3.2. The Basic Multiperiod Portfolio Optimization Models

    Assume that the investor wants to maximize his/her utility over the whole T periods investment. The multiperiod credibilitic portfolio selection problem with transaction costs, borrowing constraints, threshold constraints and cardinality constraints can be formulated as the following problem:

    max t = 1 T ( ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 | ) θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 ) s . t { W t + 1 = ( 1 + ( i = 1 n ( a i t + β i t α i t 4 ) x i t i = 1 n c i t | x i t x i t 1 | ) ) W t ( a ) 1 i = 1 n x i t x f t b                                                                    (b ) i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n                                          ( c ) l i t z i t x i t u i t z i t , i = 1 , , n ,    t = 1 , , T                             ( d )
    (29)

    where constraint (a) denotes the wealth accumulation constraint; constraint (b) indicates that the investment proportion of risk-free asset at period t must exceed the given lower bound; constraint (c) represents the desired number of risky assets in the portfolio must not exceed the given value K; constraint (d) states threshold con-straints of xit.

    4. SOLUTION ALGORITHM

    In this section, the credibilitic multiperiod mean se-mivariance portfolio selection problem with real con-straints will be approximated into a mix integer dynamic programming problem. A novel discrete iteration method will be proposed to solve the dynamic programming problem. The linear convergence of the method will be proved.

    4.1. The Proposed Model Approximated to Dynamic Programming Problem

    The sub-problem of Model (29) at period t is as fol-lows:

    max ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 | )      θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 )      s . t . { 1 i = 1 n x i t x f t b i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n l i t z i t x i t u i t z i t , i = 1 , , n
    (30)

    Let x i t 1 = x i t 1 ¯ , where x i t 1 ¯ is preset value, i = 1 n x i t 1 ¯ = 1 x b f t , the Model (29) can be approximated into the following model:

    max t = 1 T ( ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 ¯ | )       θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 )       s . t { W t + 1 = ( 1 + ( i = 1 n ( a i t + β i t α i t 4 ) x i t i = 1 n c i t | x i t x i t 1 ¯ | ) ) W t 1 i = 1 n x i t x f t b i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n   l i t z i t x i t u i t z i t , i = 1 , , n , t = 1 , , T  
    (31)

    The sub-problem of Model (31) at period t is as fol-lows:

    max ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 ¯ | )      θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 )      s . t { 1 i = 1 n x i t x f t b i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n l i t z i t x i t u i t z i t , i = 1 , , n
    (32)

    Theorem 7. Let the optimal solution and objective function value of Model (30) and Model (32) respectively be x1* and G(x1*), x2* and F(x2*).

    Then G ( x 1 * ) G ( x 2 * ) 4 ( 1 θ ) max i = 1 n { c i t } ( 1 x f t b ) .

    Proof. Because the feasible solution sets of the Model (30) and Model (32) are same, x1* and x2* are the feasible solutions of Model (30) and Model (32), respec-tively. Then, G ( x 1 * ) G ( x 2 * ) and

    F ( x 2 * ) F ( x 1 * ) , that is G ( x 1 * ) + F ( x 2 * ) G ( x 2 * ) + F ( x 1 * )

    then

    G ( x 1 * ) G ( x 2 * ) + F ( x 2 * ) F ( x 1 * ) 0
    (33)

    The right-hand side of Eq. (33) is(34)

    G ( x 1 * ) G ( x 2 * ) + F ( x 2 * ) F ( x 1 * ) = ( 1 θ ) [ i = 1 n c i t | x i t 1 * x i t 1 | i = 1 n c i t | x i t 1 * x ¯ i t 1 | ] + ( 1 θ ) [ i = 1 n c i t | x i t 2 * x ¯ i t 1 | i = 1 n c i t | x i t 2 * x i t 1 | ] 2 ( 1 θ ) i = 1 n c i t | x i t 1 x ¯ i t 1 |
    (34)

    Because x i t 1 0 , x i t 0 ,

    2 i = 1 n c i t | x i t 1 x ¯ i t 1 | 2 i = 1 n c i t x i t 1 + 2 i = 1 n c i t x ¯ i t 1 = 2 max i = 1 n { c i t } i = 1 n x i t + 2 max i = 1 n i = 1 n x ¯ i t 1 2 ( 1 θ ) max i = 1 n { c i t } ( 1 x f t b ) + 2 ( 1 θ ) max i = 1 n { c i t } ( 1 x f t b ) = 4 ( 1 θ ) max i = 1 n { c i t } ( 1 x f t b ) .

    So G ( x 1 * ) G ( x 2 * ) 4 ( 1 θ ) max i = 1 n { c i t } ( 1 x f t b ) .

    Because max i = 1 n { c i t } < < r i t , where asset i∈ efficient as-set set of portfolio, 4 ( 1 θ ) max i = 1 n { c i t } ( 1 x f t b ) is small, that G ( x 1 * ) G ( x 2 * ) is also small.

    which ends the proof.

    4.2. The Smallest and Biggest Value of State Variable at Every Period

    In Model (31), investors can choose Wt between Wtmin and Wtmax. Wtmin and Wtmax can be respectively obtained as follows:

    The investor considers to maximize the expected return of the portfolio at period t.

    max i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x ¯ i t 1 |    s . t . { 1 i = 1 n x i t x f t b i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n l i t z i t x i t u i t z i t , i = 1 , , n
    (35)

    Let y i t = | x i t x ¯ i ( t 1 ) | . Then the Model (35) can be turned into as follows.

    max i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t y i t    s . t { 1 i = 1 n x i t x f t b y i t x i t x ¯ i ( t 1 ) y i t ( x i t x ¯ i ( t 1 ) ) i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n l i t z i t x i t u i t z i t , i = 1 , , n
    (36)

    xtmax*(the optimal solution x t = ( x 1 t , x 2 t , , x n t ) ) can be obtained solving Model (36) by the CPLEX. rNtmax (the biggest of i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t y i t ) can also be obtained. Then, Wt+1max can be obtained as follows:(37)

    W t + 1 max = W t max ( 1 + i = 1 n [ a i t + β i t α i t 4 ] x i t max + r f t ( 1 i = 1 n x i t max ) i = 1 n c i t y i t max ) , t = 1 , . .. , T
    (37)

    where W1 is initial wealth, which is preset value.

    The investor only considers to minimize the semivariance of the portfolio at period t, that is, the smallest value of the rNt can be obtained as follows:

    min i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2   s . t { 1 i = 1 n x i t x f t b i = 1 n z i t K , z i t { 0 , 1 } , i = 1 , , n l i t z i t x i t u i t z i t , i = 1 , , n
    (38)

    xtmin*(the optimal solution x t = ( x f t , x 1 t , x 2 t , , x n t ) ) can be obtained solving the Model (38) by the CPLEX. Simultaneously, rNtmin (the smallest of i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t | x i t x i t 1 ¯ | ) is also obtained. Then, Wt+1min Figure 1. The multiperiod weighted digraph. can be obtained as follows:(39)

    W t + 1 min = W t min ( 1 + i = 1 n [ a i t + β i t α i t 4 ] x i t min + r f t ( 1 i = 1 n x i t min ) i = 1 n c i t | x i t min x i t 1 min ¯ | ) , t = 1 , ... , T
    (39)

    where W1 is initial wealth, which is preset value.

    4.3. The Discrete Iteration Method

    In this section, the max-plus algebra, which is proposed by Heidergott et al. (2006), will be used. The method solving the longest path of the following multiperiod weighted digraph will be proposed, and some definitions will be introduced first. Then, the method of finding the longest path of the multiperiod weighted digraph can be obtained. Finally, k+1th iteration method will be presented.

    Definition 3. A semi-field (or semi-ring) is a triplet A = {Q, ⊕, ⊗} consisting of non-empty set Q and two binary operations ⊕ and ⊗, called modi-addition and modi-multiplication respectively, defined on Q, such that for all a1, a2, a3 in Q

    • (i) each of the operations ⊕ and ⊗ is commutative

      a 1 a 2 = a 2 a 1 , a 1 a 2 = a 2 a 1

    • (ii) each of the operations ⊕ and ⊗ is associative

      ( a 1 a 2 ) a 3 = a 1 ( a 2 a 3 ) , ( a 1 a 2 ) a 3 = a 1 ( a 2 a 3 )

    • (iii) the operation ⊗ is distributive with respect to the operation ⊕

      a 1 ( a 2 a 3 ) = a 1 a 2 a 1 a 3

    • (iv) there exists an element in Q which is the zero element, denoted by ξ, such that for all a in Q, we have

      ε a = a , ε a = ε

    If there exists an element denoted by e in a semi-field A = { Q , , } , such that for all a in Q, we have

    e a = a

    then e is called an identity element of the semi-field.

    Definition 4.>a1, a2R, in semi-field A = {R, max, +}, then ε = -∞ and e = 0.

    Definition 5. Let us consider matrices A n × n = ( a i j ) n × n , B n × n = ( b i j ) n × n , a i j , b i j R , in semi-field A = {R, max, +}. Then C = A B , C n × n = ( c i j ) n × n , where cij = max{aij, bij}.

    Definition 6. Let us consider matrices A n × m = ( a i j ) n × m , B m × k = ( b i j ) m × k , a i j , b i j R , in semi-field A = {R, max, +}. Then C = A B , C n × k = ( c i j ) n × k , where c i j = max l = 1 m { a i l + b l j } .

    The Model (31) is a mix integer dynamic program-ming problem with linear inequality constraints, the op-timal solution can’t be obtained by the dynamic pro-gramming recursive relationship. In this section, a novel discrete iteration method is proposed. The method is as follows: Firstly, according to the network approach, dis-cretizes the state variables and transforms the model into multiperiod weighted digraph. Secondly, uses the max-plus algebra to solve the largest path that is the admissible solution. Thirdly, based on the admissible solution, continues iterating until the two admissible solutions are real near. Finally, the method is proved linearly convergent.

    The state variable Wt of the period t is discretized into four intervals of same widths from the smallest value to the biggest one. It means that there are five discrete values for the state variable in every period. In this way, Model (31) is transformed into a multiperiod weighted digraph as shown in Figure 1. The investment period, the value of the objective function of the period t and a discrete value of the state variable are respectively represented by the stage, the weight of the period t and the point of the multiperiod weighted digraph.

    In this section, a discrete approximate iteration me-thod will be proposed to solve the Model (31).

    Step 1: The discrete state variables at period t (t = 2, …, T+1) can be obtained by discretizing the interval value of Wmax - Wmin into four equalities. That is

    W i t = W t min + ( W t max W t min ) 4 i , i = 0 , , 4

    Step 2: The weight of the arcs in Figure 1 can be obtained following three steps:

    Step 2.1: The net expected return of the portfolio r N 1 ( 1 , j ) , j = 1 , , 5 , can be obtained as follows:

    r N 1 ( 1 , j ) = W j 2 W 1 1

    Step 2.2: The net expected return of portfolio r N t ( j , k ) at period t, j = 1, …, 5, k = 1, …, 5, can be ob-tained as follows:

    r N t ( j , k ) = W k t + 1 W j t 1

    Step 2.3: The weights F1(1,j) and Ft(j, k) of the side at period t can be obtained as follows:

    When r N t ( k , l ) is known, the sub-problem at period t of the Model (31) can be turned into(40)

    max ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t i = 1 n c i t | x i t x i t 1 | )     θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 )     s . t { ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t ( | x i t x i t 1 | ) ) r N t ( k , l ) 1 i = 1 n x i t x f t b i = 1 n z i t K , z i t { 0 , 1 } l i t z i t x i t u i t z i t , i = 1 , , n
    (40)

    Let y i t = | x i t x ¯ i ( t 1 ) | . Then the Model (41) can be turned into as follows.

    max ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t i = 1 n c i t | x i t x i t 1 | )     θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] x i t 2 )    s . t { ( i = 1 n [ a i t + β i t α i t 4 ] x i t + r f t ( 1 i = 1 n x i t ) i = 1 n c i t ( | x i t x i t 1 | ) ) r N t ( k , l ) 1 i = 1 n x i t x f t b y i t x i t x i ( t 1 ) ¯ y i t ( x i t x i ( t 1 ) ¯ ) i = 1 n z i t K , z i t { 0 , 1 } l i t z i t x i t u i t z i t , i = 1 , , n
    (41)

    x t * (the optimal solution x t = ( x f t , x 1 t , x 2 t , , x n t ) ) can be obtained solving the Model (41) by the CPLEX. Simultaneously, the objective function value Ft(j, k) can be obtained as follows:

    F t ( j , k ) = ( 1 θ ) ( i = 1 n [ a i t + β i t α i t 4 ] x i t * i = 1 n c i t y i t * ) θ ( i = 1 n [ 1 6 β i t ( β i t α i t 4 ) 3 + 1 6 α i t ( β i t + 3 α i t 4 ) 3 1 6 α i t ( β i t α i t 4 ) 3 ] ( x i t * ) 2 )

    Step 3: Calculation of the longest path of the multiperiod weighted digraph

    According to Definition 6, the longest path F(1) of the multiperiod weighted digraph of the first iteration can be obtained as follows:(42)

    F ( 1 ) = F 1 ( 1 ) F 2 ( 1 ) F T ( 1 )
    (42)

    where F 1 ( 1 ) = ( F ( 1 ) ( 1 , j ) ) 1 × 5 , F 2 ( 1 ) = ( F 2 ( 1 ) ( i , j ) ) 5 × 5 , , F T ( 1 ) = ( F T ( 1 ) ( i , j ) ) 5 × 5 .

    Step 4: The discrete approximate iteration method k+1th iteration can be described as follows:

    Let the longest path of the kth iteration be W 1 W i 2 2 ( k ) W i 3 3 ( k ) W i T + 1 T + 1 ( k ) . The optimal solutions of the longest path of Figure 1 are also feasible solutions of the multiperiod mean- absolute deviation portfolio selection model. Based on ( W 1 , W i 2 2 ( k ) , W i 3 3 ( k ) , , W i T + 1 T + 1 ( k ) ) , the state variables from period 1 to period T are discretized into four intervals as following three steps.

    Step 4.1: W 2 min and W i 2 2 ( k ) , W i 2 2 ( k ) and W 2 max are discre-tized into two same internals, respectively. The five discrete points of S2, i.e., W 2 min , W 22 ( k + 1 ) , W i 1 2 ( k + 1 ) , W 32 ( k + 1 ) , W 2 max can be obtained.

    Step 4.2: Based on ( W i 3 3 ( k ) , , W i T + 1 T + 1 ( k ) ) , using the same method, the state variables from period 3 to period T + 1 are respectively discretized into the five points. The utility of period t can also be obtained.

    Step 4.3: The longest path of the k+1th iteration F(k+1) and another feasible solution can be obtained as follows:(43)

    F ( k + 1 ) = F 1 ( k + 1 ) F 2 ( k + 1 ) F T ( k + 1 )
    (43)

    where F 1 ( k + 1 ) = ( F ( k + 1 ) ( 1 , j ) ) 1 × 5 , F 2 ( k + 1 ) = ( F 2 ( k + 1 ) ( i , j ) ) 5 × 5 , , F T ( k + 1 ) = ( F T ( k + 1 ) ( i , j ) ) 5 × 5 .

    If | F ( k + 1 ) F ( k ) | 10 6 , then the optimal solution of the longest path F(k+1) also is the approximately optimal solution of the Model (31). Otherwise turn Step 2.

    4.4. Convergence and Complex Property of the Discrete Approximate Iteration Method

    Theorem 8. The discrete approximate iteration method is convergent.

    Proof. Let the longest path in period 1 be F 1 max ( 1 , j 2 ) , and the longest path in period t be F t max ( i t , j t + 1 ) t = 2, …, T. Then the upper bound of the solution of Model (31) is

    F 1 max ( 1 , j 2 ) × F 2 max ( i 2 , j 3 ) × × F T max ( i T , j T + 1 )

    The longest path of the multiperiod weighted di-graph of the kth iteration is obtained as follows:

    F ( k ) = F 1 ( k ) F 2 ( k ) F T ( k )
    (44)

    where F 1 ( k ) = ( F ( k ) ( 1 , j ) ) 1 × 5 , F 2 ( k ) = ( F 2 ( k ) ( i , j ) ) 5 × 5 , , ( F T ( k ) ( i , j ) ) 5 × 5 . .

    Let the longest path of the k th iteration be W 1 W i 2 2 ( k ) W i 3 3 ( k ) W i T + 1 T + 1 ( k ) , Using the Step 4, the multipe-riod weighted digraph of the k+1th iteration can be ob-tained. F(k+1), which is the value of the longest path of the multiperiod weighted digraph of the k+1th iteration, can be obtained by the Eq. (44). So F ( k + 1 ) F ( k ) . The solution is becoming bigger and bigger. Because the solutions of Model (31) are increasing sets and there is an upper bound of the solution of Model (31). Then, the discrete approximate iteration method is convergent.

    Thus, the proof of the Theorem 8 is ended.

    5. NUMERICAL EXAMPLE

    In this section, a numerical example is given to ex-press the idea of the proposed model. Assume that an investor chooses thirty stocks from Shanghai Stock Ex-change for his investment. The stocks codes are respec-tively S1, …, S30. He intends to make five periods of investment with initial wealth W1 = 1 and his wealth can be adjusted at the beginning of each period. He/she assumes that the returns, risk and turnover rates of the thirty stocks at each period are represented as trapezoidal fuzzy numbers. We collect historical data of them from April 2006 to March 2017 and set every three months as a period to handle the historical data. By using the simple estimation method in Vercher et al. (2007) to handle their historical data, the triangular possibility distributions of the return rates of assets at each period can be obtained as shown in Appendix.

    Suppose that the unit transaction costs of assets of the two periods investment take the same value cit = 0.003 (i = 1, …, 30; t = 1, …, 5), the desired number of risky assets in the portfolio K = 0, …, 9 at period t, t = 1, …, 5, the lower bound of the investment proportion of risk-free asset xbft = -0.5, preference coefficient θ = 0, 0.1, …, 1, the borrowing rate of the risk-free asset rbt = 0.017, the lending rate of the risk-free asset rlt = 0.009, t = 1, …, 5, the lower lit = 0.05 and upper bound constraints uit = 0.2 (i = 1, …, 30; t = 1, …, 5), xi0 = 0(i = 1, …, 30).

    As illustrations, the following numerical examples are given to show the effectiveness of the proposed model and the discrete approximate iteration method. The algorithm was programmed by C++ language and run on a personal computer with Pentium Dual CPU, 4GHz and 8GB RAM.

    By using the discrete iteration method to solve the Model (31), the corresponding results can be obtained as follows.

    If θ = 0.5, K = 6, the optimal solution will be ob-tained as the Table 1.

    When θ = 0.5, K = 6, the optimal investment strategy at period 1 is x11 = 0.2, x131 = 0.2, x171 = 0.2, x191 = 0.2, x241 = 0.1, x281 = 0.2, xf1 = -0.2 and being the rest of va-riables equal to zero, which means investor should allo-cate his initial wealth on asset 1, asset 13, asset 17, asset 19, asset 24 , asset 28, risk-free asset and otherwise asset by the proportions of 20%, 20%, 30%, 20%, 20% , 20%, -20% and being the rest of variables equal to zero among the thirty stocks, respectively. From Table 1, the optimal investment strategy at period 2, period 3, period 4 and period 5 can also be obtained. In this case, the available terminal wealth is 2.0396.

    If θ = 0.5, K = 8, the optimal solution will be ob-tained as the Table 2.

    When q = 0.5, K = 8, the available terminal wealth is 2.2268.

    To display the influence of K on the optimal solution of multiperiod, its value is set as 6 and 8, respectively, and the Model (28) for portfolio decision-making will be used afterwards. After using the discrete approximate iteration method, the corresponding optimal investment strategies can be obtained as shown in Table 1 and Table 2. From Table 1 and Table 2, it can be seen that most of assets of the optimal solutions of K = 6 and K = 8 are same. There are six assets in period 1, i.e. asset 1, asset 13, asset 17, asset 19, asset 24, asset 28; there are six assets in period 2, i.e. asset 1, asset 12, asset 13, asset 15, asset 17, asset 18; there are six assets in period 3, i.e. asset 1, asset 12, asset 13, asset 15, asset 17, asset 18; there are six assets in period 4, i.e. asset 1, asset 12, asset 13, asset 15, asset 17, asset 18; there are six assets in period 5, i.e. asset 1, asset 12, asset 13, asset 15, asset 17, asset 18.

    When θ = 0.5, K = 0, 1, …, 9, the optimal terminal wealth can be obtained as shown in Table 3.

    In the used data sets, the same optimal solutions with the K = 8 is suitable for K ≥ 9. The experiments in this paper correspond to the values of K in the interval [0, 9]. In Table 3, the terminal wealth also becomes larger, when the preset the desired number of assets in the portfolio become larger, which reflects the influence of the desired number of assets on portfolio selection.

    When K = 7, θ = 0, 0.1, …, 0.9, 1, the optimal ter-minal wealth can be obtained as shown in Table 4.

    In the used data sets, the experiments in this paper correspond to the values of θ in the interval [0, 1]. In Table 4, the terminal wealth is same, when 0 ≤ θ ≤ 0.6; the terminal wealth becomes smaller, when preference coefficient θ which 0.6 < θ ≤ 1, become larger, which reflects the influence of preference coefficient θ on portfolio selection.

    6. CONCLUSIONS

    In this paper, we consider the multi-period portfolio selection problem in fuzzy environment. We use the credibilitic mean value and the semivariance to measure the return and the risk of the multiperiod portfolio, respectively. A new multi-period portfolio optimization model with transaction cost, borrowing constraints, threshold constraints and cardinality constraints is proposed. In the proposed model, the cardinality constraints are added as a constraint to control the number of assets to be held in an efficient portfolio. Based on the credibilitic theories, the proposed model is transformed into a crisp mix integer nonlinear dynamic optimization. Because of the transaction cost and cardinality constraints, the multiperiod portfolio selection is a mix integer dynamic optimization problem with path dependence. The proposed model is approximated to dynamic programming model. A novel discrete iteration method is designed to obtain the optimal portfolio strategy, and is proved linearly convergent. Finally, an example is given to illustrate the behavior of the proposed model and the designed algorithm using real data from the Shanghai Stock Exchange.

    ACKNOWLEDGMENTS

    This research was supported by the National Natural Science Foundation of China (Nos.71271161).

    Figure

    IEMS-17-464_F1.gif

    The multiperiod weighted digraph.

    Table

    The optimal solution when θ = 0.5, K = 6

    The optimal solution when θ = 0.5, K = 8

    The optimal terminal wealth When θ = 0.5, K = 1, …, 9

    The optimal terminal wealth When K = 7, θ = 0, 0.1, …, 0.9, 1

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    The fuzzy return rates on assets of five periods investment

    REFERENCES

    1. Anagnostopoulos, K. P. and Mamanis, G. (2011), The mean-variance cardinality constrained portfolio optimization problem: an experimental evaluation of five multiobjective evolutionary algorithms , Expert Systems with Applications, 38(11), 14208-14217.
    2. Arnott, R. D. and Wagner, W. H. (1990), The measurement and control of trading costs , Financial Analysts Journal, 46(6) 73-80.
    3. Bertsimas, D. and Shioda, R. (2009), Algorithms for cardinality-constrained quadratic optimization , Computational Optimization and Applications, 43(1), 1-22.
    4. Bertsimas, D. and Pachamanova, D. (2008), Robust multiperiod portfolio management in the presence of transaction costs , Computers and Operations Research, 35(1), 3-17.
    5. Bielecki, T. R. , Jin, H. Q. , Pliska, S. R. , and Zhou, X. Y. (2005), Continuous-time mean-variance portfolio selection with bankruptcy prohibition , Mathematical Finance, 15(2), 213-244.
    6. Cesarone, F. , Scozzari, A. , and Tardella, F. (2013), A new method for mean-variance portfolio optimization with cardinality constraints , Annals of Operations Research, 205(1), 213- 234.
    7. Cui, X. T. , Zheng, X. J. , Zhu, S. S. , and Sun, X. L. (2013), Convex relaxations and MIQCQP reformulations for a class of cardinality-constrained portfolio selection problems , Journal of Global Optimization, 56(4), 1409-1423.
    8. Cui, X. Y. , Li, X. , and Li, D. (2014), Unified framework of mean-field formulations for optimal multi-period mean-variance portfolio selection , IEEE Transactions on Automatic Control, 59(7), 1833-1844.
    9. Cui, X. Y. , Li, D. , and Li, X. (2017), Mean variance policy for discrete time cone-constrained markets: time consistency in efficiency and the minimum-variance signed supermartingale measure , Mathematical Finance, 27(2), 471-504.
    10. Deng, X. and Li, R. J. (2012), A portfolio selection model with borrowing constraint based on possibility theory , Applied Soft Computing, 12(2) 754-758.
    11. Deng, G. F. , Lin, W. T. , and Lo, C. C. (2012), Markowitz-based portfolio selection with cardinality constraints using improved particle swarm optimization , Expert Systems with Applications, 39(4) 4558-4566.
    12. Fern ndez, A. and G mez, S. (2007), Portfolio selection using neural networks , Computers & Operations Research, 34(4), 1177-1191.
    13. Gao, J. J. , Li, D. , Cui, X. Y. , and Wang, S. Y. (2015), Time cardinality constrained mean-variance dynamic portfolio selection and market timing: A stochastic control approach , Automatica, 54, 91-99.
    14. Güpınar, N. and Rustem, B. (2007), Worst-case robust decisions for multi-period mean-variance portfolio optimization , European Journal of Operational Research, 183(3), 981-1000.
    15. Heidergott, B. , Olsder, G. J. , and Van der Woude, J. (2006), Max plus at work-modeling and analysis of synchronized systems: A course on max-plus algebra and its applications, Princeton University Press, 2006.
    16. Huang, X. (2008), Mean-semivariance models for fuzzy portfolio selection , Journal of Computational and Applied Mathematics, 217(1), 1-8.
    17. Li, D. and Ng, W. L. (2000), Optimal dynamic portfolio selection: multi-period mean-variance formulation , Mathematical Finance, 10(3), 387-406.
    18. Li, D. , Sun, X. , and Wang, J. (2006), Optimal lot solution to cardinality constrained mean-variance formulation for portfolio selection , Mathematical Finance, 16(1), 83-101.
    19. Li, C. J. and Li, Z. F. (2012), Multi-period portfolio optimization for asset-liability management with bankrupt control , Applied Mathematics and Computation, 218(2), 11196-11208.
    20. Li, X. , Qin, Z. , and Kar, S. (2010), Mean-variance-skewness model for portfolio selection with fuzzy returns , European Journal of operational Research, 202(1), 239-247.
    21. Liu, B. and Liu, Y. K. (2002), Expected value of fuzzy variable and fuzzy expected value models , IEEE Transactions on Fuzzy Systems, 10(4), 445-450.
    22. Liu, Y. J. , Zhang, W. G. , and Xu, W. J. (2012), Fuzzy multi-period portfolio selection optimization models using multiple criteria , Automatica, 48(12), 3042-3053.
    23. Liu, Y. J. , Zhang, W. G. , and Zhang, P. (2013), A multi-period portfolio selection optimization model by using interval analysis , Economic Modelling, 33, 113-119.
    24. Markowitz, H. M. (1952), Portfolio selection , Journal of Finance, 7(1), 77-91.
    25. Mehlawat, M. K. (2016), Credibilistic mean-entropy models for multi-period portfolio selection with multi-choice aspiration levels , Information Sciences, 345, 9-26.
    26. Murray, W. and Shek, H. (2012), A local relaxation method for the cardinality constrained portfolio optimization problem , Computational Optimization and Applications, 53(3), 681-709.
    27. Qin, Z. (2017), Random fuzzy mean-absolute deviation models for portfolio optimization problem with hybrid uncertainty , Applied Soft Computing, 56, 597-603.
    28. Ruiz-Torrubiano, R. and Suarez, A. (2010), Hybrid approaches and dimensionality reduction for portfolio selection with cardinality constrains , IEEE Computational Intelligence Magazine, 5(2), 92-107.
    29. Sadjadi, S. J. , Seyedhosseini, S. M. , and Hassanlou, K. H. (2011), Fuzzy multi period portfolio selection with different rates for borrowing and lending , Applied Soft Computing, 11(4), 3821-3826.
    30. Shaw, D. X. , Liu, S. , and Kopman, L. (2008), Lagrangian relaxation procedure for cardinality-constrained portfolio optimization , Optimization Methods & Software, 23(3), 411-420.
    31. Sun, X. L. , Zheng, X. J. , and Li, D. (2013), Recent advances in mathematical programming with semi-continuous variables and cardinality constraint , Journal of the Operations Research Society of China, 1(1), 55-77.
    32. Le Thi, H. A. , Moeini, M. , and Dinh, T. P. (2009), Portfolio selection under downside risk measures and cardinality constraints based on DC programming and DCA , Computational Management Science, 6(4), 459-475.
    33. Le Thi, H. A. and Moeini, M. (2014), Long-short portfolio optimization under cardinality constraints by difference of convex functions algorithm , Journal of Optimization Theory and Applications, 161(1), 199-224.
    34. Vercher, E. , Bermudez, J. , and Segura, J. (2007), Fuzzy portfolio optimization under downside risk measures , Fuzzy Sets and Systems, 158,769-782.
    35. Woodside-Oriakhi, M. , Lucas, C. , and Beasley, J. E. (2011), Heuristic algorithms for the cardinality constrained efficient frontier , European Journal of Operational Research, 213(3), 538-550.
    36. Wu, H. and Zeng, Y. (2015), Equilibrium investment strategy for defined-contribution pension schemes with generalized mean-variance criterion and mortality risk , Insurance: Mathematics and Economics, 64, 396-408.
    37. Yoshimoto, A. (1996), The mean-variance approach to portfolio optimization subject to transaction costs , Journal of the Operational Research Society of Japan, 39(1), 99-117.
    38. Zadeh, L. (1999), Fuzzy sets as a basis for a theory of possibility , Fuzzy Sets and Systems, 1, 9-34.
    39. Zhang, W. G. , Liu, Y. J. , and Xu, W. J. (2012), A possibilistic mean-semivariance-entropy model for multi-period portfolio selection with transaction costs , European Journal of Operational Research, 222(2), 41-349.
    40. Zhang, W. G. , Liu, Y. J. , and Xu, W. J. (2014), A new fuzzy programming approach for multi-period portfolio optimization with return demand and risk control , Fuzzy Sets and Systems, 246, 107-126.
    41. Zhang, P. and Zhang, W. G. (2014), Multiperiod mean absolute deviation fuzzy portfolio selection model with risk control and cardinality constraints , Fuzzy Sets and Systems, 255, 74-91.
    42. Zhang, P. (2016), An interval mean-average absolute deviation model for multiperiod portfolio selection with risk control and cardinality constraints , Soft Computing, 20(3), 1203-1212.
    43. Zhang, P. (2017), Multiperiod mean semi- absolute deviation interval portfolio selection with entropy constraints , Journal of Industrial and Management Optimization, 13(3), 1169-1187.
    44. Zhu, S. S. , Li, D. , and Wang, S. Y. (2004), Risk control over bankruptcy in dynamic portfolio selection: A generalized mean-variance formulation , IEEE Transactions on Automatic Control, 49(3), 447-457.
    오늘하루 팝업창 안보기 닫기