research article mean-variance-cvar model of...
TRANSCRIPT
Research ArticleMean-Variance-CvaR Model of Multiportfolio Optimization viaLinear Weighted Sum Method
Younes Elahi and Mohd Ismail Abd Aziz
Faculty of Science Department of Mathematics Universiti Teknologi Malaysia 81310 Johor Bahru Johor Malaysia
Correspondence should be addressed to Younes Elahi elahimathgmailcom
Received 16 July 2013 Accepted 22 January 2014 Published 30 March 2014
Academic Editor Hao Shen
Copyright copy 2014 Y Elahi and M I Abd Aziz This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited
We propose a new approach to optimizing portfolios to mean-variance-CVaR (MVC) model Although of several researches havestudied the optimal MVCmodel of portfolio the linear weighted summethod (LWSM) was not implemented in the area The aimof this paper is to investigate the optimal portfolio model based on MVC via LWSM With this method the solution of the MVCmodel of portfolio as the multiobjective problem is presented In data analysis section this approach in investing on two assets isinvestigated AnMVCmodel of the multiportfolio was implemented inMATLAB and tested on the presented problem It is shownthat by using three objective functions it helps the investors to manage their portfolio better and thereby minimize the risk andmaximize the return of the portfolio The main goal of this study is to modify the current models and simplify it by using LWSMto obtain better results
1 Introduction
In financial activity a portfolio is a set of assets that allowssomebody to choose from several choices in investing Themain problem in this area is to find out the optimal combi-nation to distribute a given fund on a set of existing assetsMaximization of expected return and minimization of riskare two main aims of this problem [1]
Our goal is to develop the MVCmodel for portfolio withthe weighted coefficients in which this approach is adaptiveof the weighted sum method (WSM) One of the aims is tooffer a practical method for the construction and optimizethe proposed model in order to produce the Pareto optimalsolutionThere are several researches that have been done onthe optimal mean-variance-CVaR (MVC) model of portfoliobut nomuch studies have been carried out in the area by usingthe linear weighted summethod (LWSM)This approach willgive better insight into the multiportfolio process and theposition of investing via finding the feasible solution set
The paper is organized into five sections In the nextsection this paper presents the properties of multiobjectiveportfolio optimization based on mathematical approach andMVC model of portfolio optimization Then MVC model
of portfolio based on LWSM will be presented After thatthe empirical example and some challenges will be shown toclarify the discussion For this aim an MVC model of themultiportfolio was implemented in MATLAB and tested onthe presented problem Finally results and some future worksare presented in the conclusion
2 MVC Model of Portfolio
We face making many decisions in the real world Thereare numerous methods to optimize them The aim of thispaper is to find the Pareto optimal solutions However in thetheoretical problem cases if Pareto set normally cannot besolved by an algorithm thenwe try to approximate the Paretosets [2]
For example the multiobjective model for five objectivefunctions 119891
1 119891
5and some constraints can be shown by
following model [3]
Max [1198911(119909) 119891
5(119909)]119879
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014 Article ID 104064 7 pageshttpdxdoiorg1011552014104064
2 Mathematical Problems in Engineering
st119872
sum119894=1
119909119894= 1 119909
119894ge 0
For 119894 = 1 119872(1)
MVC uses the three parameters (the expected value (119864)the variance (1205902) and the CVaR (conditional value at risk)at a specified confidence level 120572 isin (0 1)) for better modelingThe aim of the proposedmodel is to give a developedmethodfor the solution [4]
We can obtain a better model of portfolio by replacingthe three indexes instead of two parameters as usual so thatit increases the efficiency of the model [5] Several studieson portfolio optimizations viamathematical approaches havebeen investigated in the literature To start the construction ofthis model we consider some variables terms and objectivefunctions as the following nomenclature [6]
Let
119873 be the number of assets that are available119877119909= return (as a random variable) depending on a
decision vector 119909 that belongs to a feasible set 119860119909119894= amount of investing in ith asset
Ω = the feasible set of solutions or search space119904 = a solution of problem120583119894the expected mean of the 119894th asset
1205902(119877119909) = the variance belong to 119877
119909
119870 the number of assets to invest (119870 le 119873)
We define the variable
119911119894= 1 if the 119894th (119894 = 1 119873) asset is chosen0 otherwise
(2)
119909119894= money ratio (0 le 119909
119894le 1) invested in the 119894th (119894 =
1 119873) assetWe consider the MVC model based on Aboulaich et al
method with following formula [4]
(119875)
min CVaR minus119864 var
st 119909 isin
(1199091 119909
119899) |119899
sum119895=1
119909119895= 1
119909119895ge 0 forall119895 isin 1 119899
(3)
We note that for random variable the value of varianceis calculated by the following formula
1205902 (119877119909) = 119864 [(119877
119909minus 119864(119877
119909))2
] (4)
To calculate the variance let 119877119909= 11987711199091+ sdot sdot sdot + 119877
119899119909119899 so we
have
1205902 (119877119909) =119899
sum119894=1
119899
sum119895=1
119909119894119909119895120575119894119895 (5)
where 120575119894119895= cov(119877
119894 119877119895) In addition in the model above
preferring the random variable 119877119909than random variable
119877119910is as the following conditions
119864 (119877119909) ge 119864 (119877
119910) 1205902 (119877
119909) le 1205902 (119877
119910)
CVaR (119877119909) le CVaR (119877
119910)
(6)
in which CVaR is calculated according to following theorem
Proposition 1 (CVaR calculation and optimization) Let 119877119909
be a random variable depending on a decision vector 119909 thatbelongs to a feasible set119860 and 120572 isin (0 1) Consider the function
119865120572(119909 V) =
1
120572119864 [minus119877
119909 +V]+ minus V (7)
where [119906]+ = 119906 for 119906 ge 0 and [119906]+ = 0 for 119906 lt 0 Thenaccording to a function of V 119865
120572is finite and continuous (hence
convex) and
119862119881119886119877 (119865119909) = min
visinR119865120572(119909 V) (8)
In addition the set consisting of the values of V for which theminimum is attained denoted by119860
120572(119909) is a nonempty closed
and bounded interval Minimizing 119862119881119886119877120572with respect to 119909 isin
119860 is equivalent tominimizing119865120572with respect to (119909 V) isin 119860119909119877
min119909isin119860
119862119881119886119877120572(119877119909) = min(119909V)isin119860119909119877
119865120572(119909 V) (9)
In addition a pair (119909lowast Vlowast) minimizes the right-handside if and only if 119909lowast minimize the left-hand side and Vlowast isin119860120572(119909lowast) 119862119881119886119877
120572(119877119909) is convex with respect to 119909 and 119865
120572(119909 V)
is convex with respect to (119909 V)Thus if the set 119860 of feasible decision vectors is convex
(which is the case for the basic version of the portfolio selectionproblem) and even if we impose a further lower limit on theexpected return minimizing CVaR is a convex optimizationproblem [7]
On the other expression MVC model denotes as thefollowing
min1205902 (119909)
subject to 119865120572(119909 V) le 119911
119864 (119909) ge 119889
119909 isin 119860 V isin 119877
(10)
where variables = (1199091 119909
119899) isin 119860 sub 119877119899 V isin 119877 and
set of feasible decision vectors denotes via 119860 that is a convexset In the model above there is one multiobjective system forportfolio optimization that consists of three objective functionsThe first objective function is to minimize the conditionalvalue at risk for portfolio optimization The second objectivefunction denotes minimizing the expected value of return of theportfolio And the last objective function considers minimizingthe variance of returns This procedure shows that we canmanage our capital to invest in several assets [5]
Mathematical Problems in Engineering 3
Table 1 Situation of two assets
AltCriteria
1198621
0101198622
0051198623
030A1 15 10 25A2 30 20 10
3 MVC Model of Portfolio Based on LWSM
Thismethod was introduced by Zadeh in 1963 and it is one ofthe main ways of solving theMO (see [8])Three types of thismethod are used for multiobjective portfolio optimization(MPO) in previous researchThey are weighted summethodweight quadratic method and weight quadratic variation
One of the main ways of weighting methods is theweighted sum method The aim of weighting method is theoptimization of the objective functions that they arrangedby linear combination (weighted sum) Different efficientsolutions can be found by changing the weights of theobjective functions [9]
The weighted sum method changes the MO problemwith a single model of mathematical optimization problemIn the model of this method sum weighting coefficient 119908
119894
multiply each objective function 119891119894to make the structure of
the objective function as follows (note that normalization ofthose coefficients is not necessary)
min119896
sum119894=1
120596119894119891119894(119909)
st 119909 isin Ω
where 120596119894ge 0 forall119894 = 1 119896
119896
sum119894=1
120596119894= 1
(11)
With convexity supposition if 120596119894gt 0 forall119894 = 1 119896 then
the solution of the above system is Pareto optimalThismeansthat if that system is convex then any Pareto optimal solutioncan be found [10] There are three criteria to measure theweightThey are subjective preference of the decisionmakersthe variance measure and the independence of criteriaUsually two methods can be used to achieve this aim theequal weights and the rank-order weights [11 12]
Example 2 (see [13]) Consider that a multiobjective problemincludes three criteria which are denoted by the sameunit and two alternatives According to Table 1 let 119882
1=
010 1198822= 005 and 119882
3= 030 The values of 120572
119894119895are
assumed to be as follows
119860 = [15 10 2530 20 10
] (12)
Therefore the decision matrix for this MCDM problem is asfollows
Criteria 010 005 030Alternatives A1 A2
With (10) we obtain data as follows
A1 (WSM Score) = 15 times 010 + 10
times 005 + 25 times 030 = 275(13)
In the same ways
A2 (WSM Score) = 30 times 010 + 20
times 005 + 10 times 030 = 7(14)
So the optimal case in the minimization situation is alterna-tiveA1 because it has the lowestWSM scoreWe can write thealternative ranking as follows A1 lt A1 (where ldquoltrdquo denotesldquobetter thanrdquo)
Now we formulate the portfolio revision problem withtransaction costs as a standard mathematical optimizationproblem with the mean-risk framework We will employ thefollowing notation
119873 number of risky assets119890 vector with all entries equal to ones
Consider the optimization problem with the normalizedsingle objective as follows
Maximize 119891 (119909) =119897
sum119896=1
1205961198961198760119896(119909)
Subject to 119909 isin 119883 = 119909 | 119891119894(119909) ge 0 119892
119894(119909)= 0 1 119897
(15)
where the weights are denoted by 120596119894 and
120596119894ge 0
119897
sum119894=1
120596119894= 1 119894 = 1 119897 (16)
And the objective function of 119896th objective function119876119896(119909) after normalization is denoted by 1198760
119896(119909) 119896 = 1 119897
Also we call 119891119894(119909) 119892
119894(119909)inequality and equality constraints
respectively With LWSM approach we have
119876119896(119909) =
119897
sum119894=1
119886119896119894119909119894 119886119896119894isin 119877 (17)
Therefore normalized objective functions have the followingform
1198760119896(119909) =
119876119896(119909)
119878119896
=1198861198961
119878119896
1199091+ sdot sdot sdot +
119886119896119899
119878119896
119909119899 (18)
in which case the floating-point values 119878119896are evaluated in the
following way
119878119896=119899
sum119895=1
1003816100381610038161003816100381611988611989611989510038161003816100381610038161003816 = 0 (19)
Obviously in many practical problems the objectivefunctions are represented by various measure units (eg if
4 Mathematical Problems in Engineering
1198761is measured in kilos 119876
2in seconds etc) For this reason
the objective function normalization is required It is nowobvious that the coefficients have values of the segment [0 1]Denote that we now have the linear programming problem
Maximize 119891 (119909) =119897
sum119896=1
120596119896
119876119896(119909)
119878119896
= 1205961
1198861198961
119878119896
1199091
+ sdot sdot sdot + 120596119899
119886119896119899
119878119896
119909119899
Subject to 119909 isin 119883
(20)
The following theorem gives the practical criteria forthe detection of some Pareto optimal solutions of problem(24) Yu et al [14] presented mean-CVaR model of portfoliooptimization based on LWSM inwhich they assumed that thereturn of the portfolio is based on multi-t-distribution andthey obtained the model for
max (119906 (119908)) = max (1205721119864 (119877) minus 120572
2CVaR) (21)
as follows
max05 (119908119879120583) + 05(120572 + 1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(22)
There are some conditions for coefficients that are usedin WSM For example the sum of them equal 1 and theyare strictly positive In addition to LWSM they must benormalized With this idea the multiportfolio optimizationproblem is as follows
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR)
Subject to 119909 isin 119860(23)
where 1205721 1205722 and 120572
3are strictly positive and 119860 is as the
above assumptions Since all objective functions on MVCmodel are convex we can define the model above to find thesolution as the Pareto optimal set [15] For instance if weconsider 120572
1= 1 120572
2= 0 and 120572
3= 0 the problem turns
to minimize the variance of the portfolio [5]Now according to (P2) as themultiobjective optimization
problem LWSM will be used to solve it as in the followingprocedure
Step 1 Construct the problem based on
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR) (24)
There are several positions of investing for investorwho can manage hisher budget which is related to theamount of 120572
1 1205722 and 120572
3 This note creates several cases
for investors to increase the return or decrease the riskof a portfolio The risk (or return) has an inverse (direct)relation with those coefficientsThe LWSM leads to followingprocedure for MVC model of portfolio
Step 2 Consider
min1205721(119908119879120590119908) minus 120572
2(119908119879120583)
+ 1205723(120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(25)
where 120572 120573 are the belief degree and specific parameterrespectively In this paper it is considered that 120573 = 05 Inaddition according to sum119899
119894=1120572119894= 1 we have 1minus(120572
1+ 1205722) = 1205723
we rewrite problem of (18) as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (120572 +1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894
0 le 119908119894le 1
(26)
where the belief degree of 120572 set is equal to 095 In additionin CVaR formula after linearization [16 17] we have
119865120573(119908 120572) = 120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
[119891 (119908 119910119896) + 120572]
+
(27)
and 119891(119908 119910119896) is the loss function which can be defined [16]
by 119891(119908 119910119896) = minus119908119879119910
119896 We can rewrite (19) according to the
assumptions above as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
119898 (1 minus 05)
119898
sum119896=1
[minus119908119879119910119896+ 095]
+
)
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(28)
where 120583119896is return of 119896th asset In the model above we
consider119898 = 2 and we have
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
2 (1 minus 05)2 ([minus119908119879119910
119896+ 095]
+
))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(29)
Mathematical Problems in Engineering 5
4 Empirical Research and Challenges
In order to evaluate the model above to find the portfoliooptimization based on LWSM we consider two assets whosemean return 120583 and covariance matrix 120590 are as the followingdata (data from [14 16])
120583 = (0067609870034191
)
120590 = (0058884101 00004640430000464043 00003643663
)
(30)
According to the data above and the model of (29) we have
min119885 = min1205721(1199111) minus 1205722(1199112) + (1 minus (120572
1+ 1205722)) (1199113) (31)
where
1199111= 1205752119901= 119908119879120590119908 =
= (0058884101 lowast 1199081+ 0000464043 lowast 119908
2) lowast 1199081
+ (0000464043 lowast 1199081+ 00003643663 lowast 119908
2) lowast 1199082
1199112= 120583119901= 119908119879120583 = 006760987119908
1+ 0034191119908
2
1199113= (095 +
1
2 (1 minus 05)2 ([minus119908119879120583 minus 095]
+
))
= 095 +1
2 (05)
times 2 ([minus (0067609871199081+ 0034191119908
2) + 095]
+
)
[119906]+ = max 119906 0
119899
sum119894=1
119908119894
= 1 0 le 119908119894le 1
(32)
Now we consider this proposed model for 1205721= 025 120572
2=
025 Therefore we have
minZ = min (001472102525 lowast 11990821
+ 000058005375 lowast 1199081lowast 1199082
+00003643663 lowast 11990822)
minus (001690246751199081+ 000854775119908
2)
+ (0475 + ([minus (0067609871199081+ 0034191119908
2)
+095]+))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(33)
01000
20003000
minus001
minus0005
0minus1
minus05
0
05
1
10
20
30
40
50
60
Figure 1 Mean-variance model of portfolio via LWSM
0
0
2000 4000 6000 8000 10000 120000
1
2
minus1
minus05
05
1
10
20
30
40
50
60
Data 1
Figure 2 MVC model of portfolio via LWSM
In order to compare the performance of mean-variancemodel and MVC model of multiportfolio optimization theminimum functions are demonstrated viaMATLABsoftwareas Figures 1 and 2 show the results of LWSMmethod for twomodels respectively
Comparative performance was measured using a MAT-LAB software and performance was shown in Figures 1 and2 which are called the mean-variance and MVC model ofmultiportfolio respectively
In summary we compare the performance of MVCmodel of multiportfolio with LWSM technique and othermodels like mean-variance approaches The mean-varianceminimization function and the MVC model are as describedin Figures 1 and 2 respectively
Table 2 is arranged to illustrate the results of MVCmodelof portfolio and refer to two assets byMatlab by this approach
As shown in the above results we tested two assets viaproposed model and we presented the results of that asTable 2 which included 20 results from our proposed modelbyMATLAB software It showed except for two obvious cases
6 Mathematical Problems in Engineering
Table 2 Results of MVC model of portfolio refer to two assets (seeSection 4) by MATLAB
1199081= 0000000 119908
2= 1000000 119911 = 1083819
1199081= 0010000 119908
2= 0990000 119911 = 1086554
1199081= 0020000 119908
2= 0980000 119911 = 1089293
1199081= 0030000 119908
2= 0970000 119911 = 1092035
1199081= 0040000 119908
2= 0960000 119911 = 1094779
1199081= 0050000 119908
2= 0950000 119911 = 1097527
1199081= 0060000 119908
2= 0940000 119911 = 1100278
1199081= 0070000 119908
2= 0930000 119911 = 1103031
1199081= 0080000 119908
2= 0920000 119911 = 1105788
1199081= 0090000 119908
2= 0910000 119911 = 1108547
1199081= 0100000 119908
2= 0900000 119911 = 1111310
1199081= 0110000 119908
2= 0890000 119911 = 1114075
1199081= 0120000 119908
2= 0880000 119911 = 1116844
1199081= 0130000 119908
2= 0870000 119911 = 1119615
1199081= 0140000 119908
2= 0860000 119911 = 1122389
1199081= 0150000 119908
2= 0850000 119911 = 1125167
1199081= 0160000 119908
2= 0840000 119911 = 1127947
1199081= 0170000 119908
2= 0830000 119911 = 1130730
1199081= 0180000 119908
2= 0820000 119911 = 1133517
1199081= 0190000 119908
2= 0810000 119911 = 1136306
1199081= 0200000 119908
2= 0800000 119911 = 1139098
1199081= 0210000 119908
2= 0790000 119911 = 1141893
1199081= 0220000 119908
2= 0780000 119911 = 1144691
1199081= 0230000 119908
2= 0770000 119911 = 1147493
1199081= 0240000 119908
2= 0760000 119911 = 1150297
1199081= 0250000 119908
2= 0750000 119911 = 1153104
1199081= 0260000 119908
2= 0740000 119911 = 1155914
1199081= 0270000 119908
2= 0730000 119911 = 1158727
1199081= 0280000 119908
2= 0720000 119911 = 1161543
1199081= 0290000 119908
2= 0710000 119911 = 1164362
1199081= 0300000 119908
2= 0700000 119911 = 1167184
1199081= 0310000 119908
2= 0690000 119911 = 1170009
1199081= 0320000 119908
2= 0680000 119911 = 1172837
1199081= 0330000 119908
2= 0670000 119911 = 1175668
1199081= 0340000 119908
2= 0660000 119911 = 1178501
1199081= 0350000 119908
2= 0650000 119911 = 1181338
1199081= 0360000 119908
2= 0640000 119911 = 1184178
1199081= 0370000 119908
2= 0630000 119911 = 1187021
1199081= 0380000 119908
2= 0620000 119911 = 1189867
1199081= 0390000 119908
2= 0610000 119911 = 1192715
1199081= 0400000 119908
2= 0600000 119911 = 1195567
1199081= 0410000 119908
2= 0590000 119911 = 1198422
1199081= 0420000 119908
2= 0580000 119911 = 1201279
1199081= 0430000 119908
2= 0570000 119911 = 1204140
1199081= 0440000 119908
2= 0560000 119911 = 1207004
1199081= 0450000 119908
2= 0550000 119911 = 1209870
1199081= 0460000 119908
2= 0540000 119911 = 1212740
1199081= 0470000 119908
2= 0530000 119911 = 1215612
1199081= 0480000 119908
2= 0520000 119911 = 1218488
1199081= 0490000 119908
2= 0510000 119911 = 1221366
Table 2 Continued
1199081= 0500000 119908
2= 0500000 119911 = 1224248
1199081= 0510000 119908
2= 0490000 119911 = 1227132
1199081= 0520000 119908
2= 0480000 119911 = 1230019
1199081= 0530000 119908
2= 0470000 119911 = 1232910
1199081= 0540000 119908
2= 0460000 119911 = 1235803
1199081= 0550000 119908
2= 0450000 119911 = 1238699
1199081= 0560000 119908
2= 0440000 119911 = 1241599
1199081= 0570000 119908
2= 0430000 119911 = 1244501
1199081= 0580000 119908
2= 0420000 119911 = 1247406
1199081= 0590000 119908
2= 0410000 119911 = 1250314
1199081= 0600000 119908
2= 0400000 119911 = 1253225
1199081= 0610000 119908
2= 0390000 119911 = 1256140
1199081= 0620000 119908
2= 0380000 119911 = 1259057
1199081= 0630000 119908
2= 0370000 119911 = 1261977
1199081= 0640000 119908
2= 0360000 119911 = 1264900
1199081= 0650000 119908
2= 0350000 119911 = 1267826
1199081= 0660000 119908
2= 0340000 119911 = 1270755
1199081= 0670000 119908
2= 0330000 119911 = 1273687
1199081= 0680000 119908
2= 0320000 119911 = 1276622
1199081= 0690000 119908
2= 0310000 119911 = 1279560
1199081= 0700000 119908
2= 0300000 119911 = 1282501
1199081= 0710000 119908
2= 0290000 119911 = 1285445
1199081= 0720000 119908
2= 0280000 119911 = 1288392
1199081= 0730000 119908
2= 0270000 119911 = 1291341
1199081= 0740000 119908
2= 0260000 119911 = 1294294
1199081= 0750000 119908
2= 0250000 119911 = 1297250
1199081= 0760000 119908
2= 0240000 119911 = 1300209
1199081= 0770000 119908
2= 0230000 119911 = 1303170
1199081= 0780000 119908
2= 0220000 119911 = 1306135
1199081= 0790000 119908
2= 0210000 119911 = 1309103
1199081= 0800000 119908
2= 0200000 119911 = 1312074
1199081= 0810000 119908
2= 0190000 119911 = 1315047
1199081= 0820000 119908
2= 0180000 119911 = 1318024
1199081= 0830000 119908
2= 0170000 119911 = 1321003
1199081= 0840000 119908
2= 0160000 119911 = 1323986
1199081= 0850000 119908
2= 0150000 119911 = 1326971
1199081= 0860000 119908
2= 0140000 119911 = 1329960
1199081= 0870000 119908
2= 0130000 119911 = 1332951
1199081= 0880000 119908
2= 0120000 119911 = 1335946
1199081= 0890000 119908
2= 0110000 119911 = 1338943
1199081= 0900000 119908
2= 0100000 119911 = 1341944
1199081= 0910000 119908
2= 0090000 119911 = 1344947
1199081= 0920000 119908
2= 0080000 119911 = 1347953
1199081= 0930000 119908
2= 0070000 119911 = 1350963
1199081= 0940000 119908
2= 0060000 119911 = 1353975
1199081= 0950000 119908
2= 0050000 119911 = 1356990
1199081= 0960000 119908
2= 0040000 119911 = 1360008
1199081= 0970000 119908
2= 0030000 119911 = 1363030
1199081= 0980000 119908
2= 0020000 119911 = 1366054
1199081= 0990000 119908
2= 0010000 119911 = 1369081
1199081= 1000000 119908
2= 0000000 119911 = 1372111
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Mathematical Problems in Engineering
st119872
sum119894=1
119909119894= 1 119909
119894ge 0
For 119894 = 1 119872(1)
MVC uses the three parameters (the expected value (119864)the variance (1205902) and the CVaR (conditional value at risk)at a specified confidence level 120572 isin (0 1)) for better modelingThe aim of the proposedmodel is to give a developedmethodfor the solution [4]
We can obtain a better model of portfolio by replacingthe three indexes instead of two parameters as usual so thatit increases the efficiency of the model [5] Several studieson portfolio optimizations viamathematical approaches havebeen investigated in the literature To start the construction ofthis model we consider some variables terms and objectivefunctions as the following nomenclature [6]
Let
119873 be the number of assets that are available119877119909= return (as a random variable) depending on a
decision vector 119909 that belongs to a feasible set 119860119909119894= amount of investing in ith asset
Ω = the feasible set of solutions or search space119904 = a solution of problem120583119894the expected mean of the 119894th asset
1205902(119877119909) = the variance belong to 119877
119909
119870 the number of assets to invest (119870 le 119873)
We define the variable
119911119894= 1 if the 119894th (119894 = 1 119873) asset is chosen0 otherwise
(2)
119909119894= money ratio (0 le 119909
119894le 1) invested in the 119894th (119894 =
1 119873) assetWe consider the MVC model based on Aboulaich et al
method with following formula [4]
(119875)
min CVaR minus119864 var
st 119909 isin
(1199091 119909
119899) |119899
sum119895=1
119909119895= 1
119909119895ge 0 forall119895 isin 1 119899
(3)
We note that for random variable the value of varianceis calculated by the following formula
1205902 (119877119909) = 119864 [(119877
119909minus 119864(119877
119909))2
] (4)
To calculate the variance let 119877119909= 11987711199091+ sdot sdot sdot + 119877
119899119909119899 so we
have
1205902 (119877119909) =119899
sum119894=1
119899
sum119895=1
119909119894119909119895120575119894119895 (5)
where 120575119894119895= cov(119877
119894 119877119895) In addition in the model above
preferring the random variable 119877119909than random variable
119877119910is as the following conditions
119864 (119877119909) ge 119864 (119877
119910) 1205902 (119877
119909) le 1205902 (119877
119910)
CVaR (119877119909) le CVaR (119877
119910)
(6)
in which CVaR is calculated according to following theorem
Proposition 1 (CVaR calculation and optimization) Let 119877119909
be a random variable depending on a decision vector 119909 thatbelongs to a feasible set119860 and 120572 isin (0 1) Consider the function
119865120572(119909 V) =
1
120572119864 [minus119877
119909 +V]+ minus V (7)
where [119906]+ = 119906 for 119906 ge 0 and [119906]+ = 0 for 119906 lt 0 Thenaccording to a function of V 119865
120572is finite and continuous (hence
convex) and
119862119881119886119877 (119865119909) = min
visinR119865120572(119909 V) (8)
In addition the set consisting of the values of V for which theminimum is attained denoted by119860
120572(119909) is a nonempty closed
and bounded interval Minimizing 119862119881119886119877120572with respect to 119909 isin
119860 is equivalent tominimizing119865120572with respect to (119909 V) isin 119860119909119877
min119909isin119860
119862119881119886119877120572(119877119909) = min(119909V)isin119860119909119877
119865120572(119909 V) (9)
In addition a pair (119909lowast Vlowast) minimizes the right-handside if and only if 119909lowast minimize the left-hand side and Vlowast isin119860120572(119909lowast) 119862119881119886119877
120572(119877119909) is convex with respect to 119909 and 119865
120572(119909 V)
is convex with respect to (119909 V)Thus if the set 119860 of feasible decision vectors is convex
(which is the case for the basic version of the portfolio selectionproblem) and even if we impose a further lower limit on theexpected return minimizing CVaR is a convex optimizationproblem [7]
On the other expression MVC model denotes as thefollowing
min1205902 (119909)
subject to 119865120572(119909 V) le 119911
119864 (119909) ge 119889
119909 isin 119860 V isin 119877
(10)
where variables = (1199091 119909
119899) isin 119860 sub 119877119899 V isin 119877 and
set of feasible decision vectors denotes via 119860 that is a convexset In the model above there is one multiobjective system forportfolio optimization that consists of three objective functionsThe first objective function is to minimize the conditionalvalue at risk for portfolio optimization The second objectivefunction denotes minimizing the expected value of return of theportfolio And the last objective function considers minimizingthe variance of returns This procedure shows that we canmanage our capital to invest in several assets [5]
Mathematical Problems in Engineering 3
Table 1 Situation of two assets
AltCriteria
1198621
0101198622
0051198623
030A1 15 10 25A2 30 20 10
3 MVC Model of Portfolio Based on LWSM
Thismethod was introduced by Zadeh in 1963 and it is one ofthe main ways of solving theMO (see [8])Three types of thismethod are used for multiobjective portfolio optimization(MPO) in previous researchThey are weighted summethodweight quadratic method and weight quadratic variation
One of the main ways of weighting methods is theweighted sum method The aim of weighting method is theoptimization of the objective functions that they arrangedby linear combination (weighted sum) Different efficientsolutions can be found by changing the weights of theobjective functions [9]
The weighted sum method changes the MO problemwith a single model of mathematical optimization problemIn the model of this method sum weighting coefficient 119908
119894
multiply each objective function 119891119894to make the structure of
the objective function as follows (note that normalization ofthose coefficients is not necessary)
min119896
sum119894=1
120596119894119891119894(119909)
st 119909 isin Ω
where 120596119894ge 0 forall119894 = 1 119896
119896
sum119894=1
120596119894= 1
(11)
With convexity supposition if 120596119894gt 0 forall119894 = 1 119896 then
the solution of the above system is Pareto optimalThismeansthat if that system is convex then any Pareto optimal solutioncan be found [10] There are three criteria to measure theweightThey are subjective preference of the decisionmakersthe variance measure and the independence of criteriaUsually two methods can be used to achieve this aim theequal weights and the rank-order weights [11 12]
Example 2 (see [13]) Consider that a multiobjective problemincludes three criteria which are denoted by the sameunit and two alternatives According to Table 1 let 119882
1=
010 1198822= 005 and 119882
3= 030 The values of 120572
119894119895are
assumed to be as follows
119860 = [15 10 2530 20 10
] (12)
Therefore the decision matrix for this MCDM problem is asfollows
Criteria 010 005 030Alternatives A1 A2
With (10) we obtain data as follows
A1 (WSM Score) = 15 times 010 + 10
times 005 + 25 times 030 = 275(13)
In the same ways
A2 (WSM Score) = 30 times 010 + 20
times 005 + 10 times 030 = 7(14)
So the optimal case in the minimization situation is alterna-tiveA1 because it has the lowestWSM scoreWe can write thealternative ranking as follows A1 lt A1 (where ldquoltrdquo denotesldquobetter thanrdquo)
Now we formulate the portfolio revision problem withtransaction costs as a standard mathematical optimizationproblem with the mean-risk framework We will employ thefollowing notation
119873 number of risky assets119890 vector with all entries equal to ones
Consider the optimization problem with the normalizedsingle objective as follows
Maximize 119891 (119909) =119897
sum119896=1
1205961198961198760119896(119909)
Subject to 119909 isin 119883 = 119909 | 119891119894(119909) ge 0 119892
119894(119909)= 0 1 119897
(15)
where the weights are denoted by 120596119894 and
120596119894ge 0
119897
sum119894=1
120596119894= 1 119894 = 1 119897 (16)
And the objective function of 119896th objective function119876119896(119909) after normalization is denoted by 1198760
119896(119909) 119896 = 1 119897
Also we call 119891119894(119909) 119892
119894(119909)inequality and equality constraints
respectively With LWSM approach we have
119876119896(119909) =
119897
sum119894=1
119886119896119894119909119894 119886119896119894isin 119877 (17)
Therefore normalized objective functions have the followingform
1198760119896(119909) =
119876119896(119909)
119878119896
=1198861198961
119878119896
1199091+ sdot sdot sdot +
119886119896119899
119878119896
119909119899 (18)
in which case the floating-point values 119878119896are evaluated in the
following way
119878119896=119899
sum119895=1
1003816100381610038161003816100381611988611989611989510038161003816100381610038161003816 = 0 (19)
Obviously in many practical problems the objectivefunctions are represented by various measure units (eg if
4 Mathematical Problems in Engineering
1198761is measured in kilos 119876
2in seconds etc) For this reason
the objective function normalization is required It is nowobvious that the coefficients have values of the segment [0 1]Denote that we now have the linear programming problem
Maximize 119891 (119909) =119897
sum119896=1
120596119896
119876119896(119909)
119878119896
= 1205961
1198861198961
119878119896
1199091
+ sdot sdot sdot + 120596119899
119886119896119899
119878119896
119909119899
Subject to 119909 isin 119883
(20)
The following theorem gives the practical criteria forthe detection of some Pareto optimal solutions of problem(24) Yu et al [14] presented mean-CVaR model of portfoliooptimization based on LWSM inwhich they assumed that thereturn of the portfolio is based on multi-t-distribution andthey obtained the model for
max (119906 (119908)) = max (1205721119864 (119877) minus 120572
2CVaR) (21)
as follows
max05 (119908119879120583) + 05(120572 + 1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(22)
There are some conditions for coefficients that are usedin WSM For example the sum of them equal 1 and theyare strictly positive In addition to LWSM they must benormalized With this idea the multiportfolio optimizationproblem is as follows
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR)
Subject to 119909 isin 119860(23)
where 1205721 1205722 and 120572
3are strictly positive and 119860 is as the
above assumptions Since all objective functions on MVCmodel are convex we can define the model above to find thesolution as the Pareto optimal set [15] For instance if weconsider 120572
1= 1 120572
2= 0 and 120572
3= 0 the problem turns
to minimize the variance of the portfolio [5]Now according to (P2) as themultiobjective optimization
problem LWSM will be used to solve it as in the followingprocedure
Step 1 Construct the problem based on
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR) (24)
There are several positions of investing for investorwho can manage hisher budget which is related to theamount of 120572
1 1205722 and 120572
3 This note creates several cases
for investors to increase the return or decrease the riskof a portfolio The risk (or return) has an inverse (direct)relation with those coefficientsThe LWSM leads to followingprocedure for MVC model of portfolio
Step 2 Consider
min1205721(119908119879120590119908) minus 120572
2(119908119879120583)
+ 1205723(120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(25)
where 120572 120573 are the belief degree and specific parameterrespectively In this paper it is considered that 120573 = 05 Inaddition according to sum119899
119894=1120572119894= 1 we have 1minus(120572
1+ 1205722) = 1205723
we rewrite problem of (18) as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (120572 +1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894
0 le 119908119894le 1
(26)
where the belief degree of 120572 set is equal to 095 In additionin CVaR formula after linearization [16 17] we have
119865120573(119908 120572) = 120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
[119891 (119908 119910119896) + 120572]
+
(27)
and 119891(119908 119910119896) is the loss function which can be defined [16]
by 119891(119908 119910119896) = minus119908119879119910
119896 We can rewrite (19) according to the
assumptions above as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
119898 (1 minus 05)
119898
sum119896=1
[minus119908119879119910119896+ 095]
+
)
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(28)
where 120583119896is return of 119896th asset In the model above we
consider119898 = 2 and we have
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
2 (1 minus 05)2 ([minus119908119879119910
119896+ 095]
+
))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(29)
Mathematical Problems in Engineering 5
4 Empirical Research and Challenges
In order to evaluate the model above to find the portfoliooptimization based on LWSM we consider two assets whosemean return 120583 and covariance matrix 120590 are as the followingdata (data from [14 16])
120583 = (0067609870034191
)
120590 = (0058884101 00004640430000464043 00003643663
)
(30)
According to the data above and the model of (29) we have
min119885 = min1205721(1199111) minus 1205722(1199112) + (1 minus (120572
1+ 1205722)) (1199113) (31)
where
1199111= 1205752119901= 119908119879120590119908 =
= (0058884101 lowast 1199081+ 0000464043 lowast 119908
2) lowast 1199081
+ (0000464043 lowast 1199081+ 00003643663 lowast 119908
2) lowast 1199082
1199112= 120583119901= 119908119879120583 = 006760987119908
1+ 0034191119908
2
1199113= (095 +
1
2 (1 minus 05)2 ([minus119908119879120583 minus 095]
+
))
= 095 +1
2 (05)
times 2 ([minus (0067609871199081+ 0034191119908
2) + 095]
+
)
[119906]+ = max 119906 0
119899
sum119894=1
119908119894
= 1 0 le 119908119894le 1
(32)
Now we consider this proposed model for 1205721= 025 120572
2=
025 Therefore we have
minZ = min (001472102525 lowast 11990821
+ 000058005375 lowast 1199081lowast 1199082
+00003643663 lowast 11990822)
minus (001690246751199081+ 000854775119908
2)
+ (0475 + ([minus (0067609871199081+ 0034191119908
2)
+095]+))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(33)
01000
20003000
minus001
minus0005
0minus1
minus05
0
05
1
10
20
30
40
50
60
Figure 1 Mean-variance model of portfolio via LWSM
0
0
2000 4000 6000 8000 10000 120000
1
2
minus1
minus05
05
1
10
20
30
40
50
60
Data 1
Figure 2 MVC model of portfolio via LWSM
In order to compare the performance of mean-variancemodel and MVC model of multiportfolio optimization theminimum functions are demonstrated viaMATLABsoftwareas Figures 1 and 2 show the results of LWSMmethod for twomodels respectively
Comparative performance was measured using a MAT-LAB software and performance was shown in Figures 1 and2 which are called the mean-variance and MVC model ofmultiportfolio respectively
In summary we compare the performance of MVCmodel of multiportfolio with LWSM technique and othermodels like mean-variance approaches The mean-varianceminimization function and the MVC model are as describedin Figures 1 and 2 respectively
Table 2 is arranged to illustrate the results of MVCmodelof portfolio and refer to two assets byMatlab by this approach
As shown in the above results we tested two assets viaproposed model and we presented the results of that asTable 2 which included 20 results from our proposed modelbyMATLAB software It showed except for two obvious cases
6 Mathematical Problems in Engineering
Table 2 Results of MVC model of portfolio refer to two assets (seeSection 4) by MATLAB
1199081= 0000000 119908
2= 1000000 119911 = 1083819
1199081= 0010000 119908
2= 0990000 119911 = 1086554
1199081= 0020000 119908
2= 0980000 119911 = 1089293
1199081= 0030000 119908
2= 0970000 119911 = 1092035
1199081= 0040000 119908
2= 0960000 119911 = 1094779
1199081= 0050000 119908
2= 0950000 119911 = 1097527
1199081= 0060000 119908
2= 0940000 119911 = 1100278
1199081= 0070000 119908
2= 0930000 119911 = 1103031
1199081= 0080000 119908
2= 0920000 119911 = 1105788
1199081= 0090000 119908
2= 0910000 119911 = 1108547
1199081= 0100000 119908
2= 0900000 119911 = 1111310
1199081= 0110000 119908
2= 0890000 119911 = 1114075
1199081= 0120000 119908
2= 0880000 119911 = 1116844
1199081= 0130000 119908
2= 0870000 119911 = 1119615
1199081= 0140000 119908
2= 0860000 119911 = 1122389
1199081= 0150000 119908
2= 0850000 119911 = 1125167
1199081= 0160000 119908
2= 0840000 119911 = 1127947
1199081= 0170000 119908
2= 0830000 119911 = 1130730
1199081= 0180000 119908
2= 0820000 119911 = 1133517
1199081= 0190000 119908
2= 0810000 119911 = 1136306
1199081= 0200000 119908
2= 0800000 119911 = 1139098
1199081= 0210000 119908
2= 0790000 119911 = 1141893
1199081= 0220000 119908
2= 0780000 119911 = 1144691
1199081= 0230000 119908
2= 0770000 119911 = 1147493
1199081= 0240000 119908
2= 0760000 119911 = 1150297
1199081= 0250000 119908
2= 0750000 119911 = 1153104
1199081= 0260000 119908
2= 0740000 119911 = 1155914
1199081= 0270000 119908
2= 0730000 119911 = 1158727
1199081= 0280000 119908
2= 0720000 119911 = 1161543
1199081= 0290000 119908
2= 0710000 119911 = 1164362
1199081= 0300000 119908
2= 0700000 119911 = 1167184
1199081= 0310000 119908
2= 0690000 119911 = 1170009
1199081= 0320000 119908
2= 0680000 119911 = 1172837
1199081= 0330000 119908
2= 0670000 119911 = 1175668
1199081= 0340000 119908
2= 0660000 119911 = 1178501
1199081= 0350000 119908
2= 0650000 119911 = 1181338
1199081= 0360000 119908
2= 0640000 119911 = 1184178
1199081= 0370000 119908
2= 0630000 119911 = 1187021
1199081= 0380000 119908
2= 0620000 119911 = 1189867
1199081= 0390000 119908
2= 0610000 119911 = 1192715
1199081= 0400000 119908
2= 0600000 119911 = 1195567
1199081= 0410000 119908
2= 0590000 119911 = 1198422
1199081= 0420000 119908
2= 0580000 119911 = 1201279
1199081= 0430000 119908
2= 0570000 119911 = 1204140
1199081= 0440000 119908
2= 0560000 119911 = 1207004
1199081= 0450000 119908
2= 0550000 119911 = 1209870
1199081= 0460000 119908
2= 0540000 119911 = 1212740
1199081= 0470000 119908
2= 0530000 119911 = 1215612
1199081= 0480000 119908
2= 0520000 119911 = 1218488
1199081= 0490000 119908
2= 0510000 119911 = 1221366
Table 2 Continued
1199081= 0500000 119908
2= 0500000 119911 = 1224248
1199081= 0510000 119908
2= 0490000 119911 = 1227132
1199081= 0520000 119908
2= 0480000 119911 = 1230019
1199081= 0530000 119908
2= 0470000 119911 = 1232910
1199081= 0540000 119908
2= 0460000 119911 = 1235803
1199081= 0550000 119908
2= 0450000 119911 = 1238699
1199081= 0560000 119908
2= 0440000 119911 = 1241599
1199081= 0570000 119908
2= 0430000 119911 = 1244501
1199081= 0580000 119908
2= 0420000 119911 = 1247406
1199081= 0590000 119908
2= 0410000 119911 = 1250314
1199081= 0600000 119908
2= 0400000 119911 = 1253225
1199081= 0610000 119908
2= 0390000 119911 = 1256140
1199081= 0620000 119908
2= 0380000 119911 = 1259057
1199081= 0630000 119908
2= 0370000 119911 = 1261977
1199081= 0640000 119908
2= 0360000 119911 = 1264900
1199081= 0650000 119908
2= 0350000 119911 = 1267826
1199081= 0660000 119908
2= 0340000 119911 = 1270755
1199081= 0670000 119908
2= 0330000 119911 = 1273687
1199081= 0680000 119908
2= 0320000 119911 = 1276622
1199081= 0690000 119908
2= 0310000 119911 = 1279560
1199081= 0700000 119908
2= 0300000 119911 = 1282501
1199081= 0710000 119908
2= 0290000 119911 = 1285445
1199081= 0720000 119908
2= 0280000 119911 = 1288392
1199081= 0730000 119908
2= 0270000 119911 = 1291341
1199081= 0740000 119908
2= 0260000 119911 = 1294294
1199081= 0750000 119908
2= 0250000 119911 = 1297250
1199081= 0760000 119908
2= 0240000 119911 = 1300209
1199081= 0770000 119908
2= 0230000 119911 = 1303170
1199081= 0780000 119908
2= 0220000 119911 = 1306135
1199081= 0790000 119908
2= 0210000 119911 = 1309103
1199081= 0800000 119908
2= 0200000 119911 = 1312074
1199081= 0810000 119908
2= 0190000 119911 = 1315047
1199081= 0820000 119908
2= 0180000 119911 = 1318024
1199081= 0830000 119908
2= 0170000 119911 = 1321003
1199081= 0840000 119908
2= 0160000 119911 = 1323986
1199081= 0850000 119908
2= 0150000 119911 = 1326971
1199081= 0860000 119908
2= 0140000 119911 = 1329960
1199081= 0870000 119908
2= 0130000 119911 = 1332951
1199081= 0880000 119908
2= 0120000 119911 = 1335946
1199081= 0890000 119908
2= 0110000 119911 = 1338943
1199081= 0900000 119908
2= 0100000 119911 = 1341944
1199081= 0910000 119908
2= 0090000 119911 = 1344947
1199081= 0920000 119908
2= 0080000 119911 = 1347953
1199081= 0930000 119908
2= 0070000 119911 = 1350963
1199081= 0940000 119908
2= 0060000 119911 = 1353975
1199081= 0950000 119908
2= 0050000 119911 = 1356990
1199081= 0960000 119908
2= 0040000 119911 = 1360008
1199081= 0970000 119908
2= 0030000 119911 = 1363030
1199081= 0980000 119908
2= 0020000 119911 = 1366054
1199081= 0990000 119908
2= 0010000 119911 = 1369081
1199081= 1000000 119908
2= 0000000 119911 = 1372111
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 3
Table 1 Situation of two assets
AltCriteria
1198621
0101198622
0051198623
030A1 15 10 25A2 30 20 10
3 MVC Model of Portfolio Based on LWSM
Thismethod was introduced by Zadeh in 1963 and it is one ofthe main ways of solving theMO (see [8])Three types of thismethod are used for multiobjective portfolio optimization(MPO) in previous researchThey are weighted summethodweight quadratic method and weight quadratic variation
One of the main ways of weighting methods is theweighted sum method The aim of weighting method is theoptimization of the objective functions that they arrangedby linear combination (weighted sum) Different efficientsolutions can be found by changing the weights of theobjective functions [9]
The weighted sum method changes the MO problemwith a single model of mathematical optimization problemIn the model of this method sum weighting coefficient 119908
119894
multiply each objective function 119891119894to make the structure of
the objective function as follows (note that normalization ofthose coefficients is not necessary)
min119896
sum119894=1
120596119894119891119894(119909)
st 119909 isin Ω
where 120596119894ge 0 forall119894 = 1 119896
119896
sum119894=1
120596119894= 1
(11)
With convexity supposition if 120596119894gt 0 forall119894 = 1 119896 then
the solution of the above system is Pareto optimalThismeansthat if that system is convex then any Pareto optimal solutioncan be found [10] There are three criteria to measure theweightThey are subjective preference of the decisionmakersthe variance measure and the independence of criteriaUsually two methods can be used to achieve this aim theequal weights and the rank-order weights [11 12]
Example 2 (see [13]) Consider that a multiobjective problemincludes three criteria which are denoted by the sameunit and two alternatives According to Table 1 let 119882
1=
010 1198822= 005 and 119882
3= 030 The values of 120572
119894119895are
assumed to be as follows
119860 = [15 10 2530 20 10
] (12)
Therefore the decision matrix for this MCDM problem is asfollows
Criteria 010 005 030Alternatives A1 A2
With (10) we obtain data as follows
A1 (WSM Score) = 15 times 010 + 10
times 005 + 25 times 030 = 275(13)
In the same ways
A2 (WSM Score) = 30 times 010 + 20
times 005 + 10 times 030 = 7(14)
So the optimal case in the minimization situation is alterna-tiveA1 because it has the lowestWSM scoreWe can write thealternative ranking as follows A1 lt A1 (where ldquoltrdquo denotesldquobetter thanrdquo)
Now we formulate the portfolio revision problem withtransaction costs as a standard mathematical optimizationproblem with the mean-risk framework We will employ thefollowing notation
119873 number of risky assets119890 vector with all entries equal to ones
Consider the optimization problem with the normalizedsingle objective as follows
Maximize 119891 (119909) =119897
sum119896=1
1205961198961198760119896(119909)
Subject to 119909 isin 119883 = 119909 | 119891119894(119909) ge 0 119892
119894(119909)= 0 1 119897
(15)
where the weights are denoted by 120596119894 and
120596119894ge 0
119897
sum119894=1
120596119894= 1 119894 = 1 119897 (16)
And the objective function of 119896th objective function119876119896(119909) after normalization is denoted by 1198760
119896(119909) 119896 = 1 119897
Also we call 119891119894(119909) 119892
119894(119909)inequality and equality constraints
respectively With LWSM approach we have
119876119896(119909) =
119897
sum119894=1
119886119896119894119909119894 119886119896119894isin 119877 (17)
Therefore normalized objective functions have the followingform
1198760119896(119909) =
119876119896(119909)
119878119896
=1198861198961
119878119896
1199091+ sdot sdot sdot +
119886119896119899
119878119896
119909119899 (18)
in which case the floating-point values 119878119896are evaluated in the
following way
119878119896=119899
sum119895=1
1003816100381610038161003816100381611988611989611989510038161003816100381610038161003816 = 0 (19)
Obviously in many practical problems the objectivefunctions are represented by various measure units (eg if
4 Mathematical Problems in Engineering
1198761is measured in kilos 119876
2in seconds etc) For this reason
the objective function normalization is required It is nowobvious that the coefficients have values of the segment [0 1]Denote that we now have the linear programming problem
Maximize 119891 (119909) =119897
sum119896=1
120596119896
119876119896(119909)
119878119896
= 1205961
1198861198961
119878119896
1199091
+ sdot sdot sdot + 120596119899
119886119896119899
119878119896
119909119899
Subject to 119909 isin 119883
(20)
The following theorem gives the practical criteria forthe detection of some Pareto optimal solutions of problem(24) Yu et al [14] presented mean-CVaR model of portfoliooptimization based on LWSM inwhich they assumed that thereturn of the portfolio is based on multi-t-distribution andthey obtained the model for
max (119906 (119908)) = max (1205721119864 (119877) minus 120572
2CVaR) (21)
as follows
max05 (119908119879120583) + 05(120572 + 1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(22)
There are some conditions for coefficients that are usedin WSM For example the sum of them equal 1 and theyare strictly positive In addition to LWSM they must benormalized With this idea the multiportfolio optimizationproblem is as follows
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR)
Subject to 119909 isin 119860(23)
where 1205721 1205722 and 120572
3are strictly positive and 119860 is as the
above assumptions Since all objective functions on MVCmodel are convex we can define the model above to find thesolution as the Pareto optimal set [15] For instance if weconsider 120572
1= 1 120572
2= 0 and 120572
3= 0 the problem turns
to minimize the variance of the portfolio [5]Now according to (P2) as themultiobjective optimization
problem LWSM will be used to solve it as in the followingprocedure
Step 1 Construct the problem based on
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR) (24)
There are several positions of investing for investorwho can manage hisher budget which is related to theamount of 120572
1 1205722 and 120572
3 This note creates several cases
for investors to increase the return or decrease the riskof a portfolio The risk (or return) has an inverse (direct)relation with those coefficientsThe LWSM leads to followingprocedure for MVC model of portfolio
Step 2 Consider
min1205721(119908119879120590119908) minus 120572
2(119908119879120583)
+ 1205723(120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(25)
where 120572 120573 are the belief degree and specific parameterrespectively In this paper it is considered that 120573 = 05 Inaddition according to sum119899
119894=1120572119894= 1 we have 1minus(120572
1+ 1205722) = 1205723
we rewrite problem of (18) as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (120572 +1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894
0 le 119908119894le 1
(26)
where the belief degree of 120572 set is equal to 095 In additionin CVaR formula after linearization [16 17] we have
119865120573(119908 120572) = 120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
[119891 (119908 119910119896) + 120572]
+
(27)
and 119891(119908 119910119896) is the loss function which can be defined [16]
by 119891(119908 119910119896) = minus119908119879119910
119896 We can rewrite (19) according to the
assumptions above as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
119898 (1 minus 05)
119898
sum119896=1
[minus119908119879119910119896+ 095]
+
)
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(28)
where 120583119896is return of 119896th asset In the model above we
consider119898 = 2 and we have
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
2 (1 minus 05)2 ([minus119908119879119910
119896+ 095]
+
))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(29)
Mathematical Problems in Engineering 5
4 Empirical Research and Challenges
In order to evaluate the model above to find the portfoliooptimization based on LWSM we consider two assets whosemean return 120583 and covariance matrix 120590 are as the followingdata (data from [14 16])
120583 = (0067609870034191
)
120590 = (0058884101 00004640430000464043 00003643663
)
(30)
According to the data above and the model of (29) we have
min119885 = min1205721(1199111) minus 1205722(1199112) + (1 minus (120572
1+ 1205722)) (1199113) (31)
where
1199111= 1205752119901= 119908119879120590119908 =
= (0058884101 lowast 1199081+ 0000464043 lowast 119908
2) lowast 1199081
+ (0000464043 lowast 1199081+ 00003643663 lowast 119908
2) lowast 1199082
1199112= 120583119901= 119908119879120583 = 006760987119908
1+ 0034191119908
2
1199113= (095 +
1
2 (1 minus 05)2 ([minus119908119879120583 minus 095]
+
))
= 095 +1
2 (05)
times 2 ([minus (0067609871199081+ 0034191119908
2) + 095]
+
)
[119906]+ = max 119906 0
119899
sum119894=1
119908119894
= 1 0 le 119908119894le 1
(32)
Now we consider this proposed model for 1205721= 025 120572
2=
025 Therefore we have
minZ = min (001472102525 lowast 11990821
+ 000058005375 lowast 1199081lowast 1199082
+00003643663 lowast 11990822)
minus (001690246751199081+ 000854775119908
2)
+ (0475 + ([minus (0067609871199081+ 0034191119908
2)
+095]+))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(33)
01000
20003000
minus001
minus0005
0minus1
minus05
0
05
1
10
20
30
40
50
60
Figure 1 Mean-variance model of portfolio via LWSM
0
0
2000 4000 6000 8000 10000 120000
1
2
minus1
minus05
05
1
10
20
30
40
50
60
Data 1
Figure 2 MVC model of portfolio via LWSM
In order to compare the performance of mean-variancemodel and MVC model of multiportfolio optimization theminimum functions are demonstrated viaMATLABsoftwareas Figures 1 and 2 show the results of LWSMmethod for twomodels respectively
Comparative performance was measured using a MAT-LAB software and performance was shown in Figures 1 and2 which are called the mean-variance and MVC model ofmultiportfolio respectively
In summary we compare the performance of MVCmodel of multiportfolio with LWSM technique and othermodels like mean-variance approaches The mean-varianceminimization function and the MVC model are as describedin Figures 1 and 2 respectively
Table 2 is arranged to illustrate the results of MVCmodelof portfolio and refer to two assets byMatlab by this approach
As shown in the above results we tested two assets viaproposed model and we presented the results of that asTable 2 which included 20 results from our proposed modelbyMATLAB software It showed except for two obvious cases
6 Mathematical Problems in Engineering
Table 2 Results of MVC model of portfolio refer to two assets (seeSection 4) by MATLAB
1199081= 0000000 119908
2= 1000000 119911 = 1083819
1199081= 0010000 119908
2= 0990000 119911 = 1086554
1199081= 0020000 119908
2= 0980000 119911 = 1089293
1199081= 0030000 119908
2= 0970000 119911 = 1092035
1199081= 0040000 119908
2= 0960000 119911 = 1094779
1199081= 0050000 119908
2= 0950000 119911 = 1097527
1199081= 0060000 119908
2= 0940000 119911 = 1100278
1199081= 0070000 119908
2= 0930000 119911 = 1103031
1199081= 0080000 119908
2= 0920000 119911 = 1105788
1199081= 0090000 119908
2= 0910000 119911 = 1108547
1199081= 0100000 119908
2= 0900000 119911 = 1111310
1199081= 0110000 119908
2= 0890000 119911 = 1114075
1199081= 0120000 119908
2= 0880000 119911 = 1116844
1199081= 0130000 119908
2= 0870000 119911 = 1119615
1199081= 0140000 119908
2= 0860000 119911 = 1122389
1199081= 0150000 119908
2= 0850000 119911 = 1125167
1199081= 0160000 119908
2= 0840000 119911 = 1127947
1199081= 0170000 119908
2= 0830000 119911 = 1130730
1199081= 0180000 119908
2= 0820000 119911 = 1133517
1199081= 0190000 119908
2= 0810000 119911 = 1136306
1199081= 0200000 119908
2= 0800000 119911 = 1139098
1199081= 0210000 119908
2= 0790000 119911 = 1141893
1199081= 0220000 119908
2= 0780000 119911 = 1144691
1199081= 0230000 119908
2= 0770000 119911 = 1147493
1199081= 0240000 119908
2= 0760000 119911 = 1150297
1199081= 0250000 119908
2= 0750000 119911 = 1153104
1199081= 0260000 119908
2= 0740000 119911 = 1155914
1199081= 0270000 119908
2= 0730000 119911 = 1158727
1199081= 0280000 119908
2= 0720000 119911 = 1161543
1199081= 0290000 119908
2= 0710000 119911 = 1164362
1199081= 0300000 119908
2= 0700000 119911 = 1167184
1199081= 0310000 119908
2= 0690000 119911 = 1170009
1199081= 0320000 119908
2= 0680000 119911 = 1172837
1199081= 0330000 119908
2= 0670000 119911 = 1175668
1199081= 0340000 119908
2= 0660000 119911 = 1178501
1199081= 0350000 119908
2= 0650000 119911 = 1181338
1199081= 0360000 119908
2= 0640000 119911 = 1184178
1199081= 0370000 119908
2= 0630000 119911 = 1187021
1199081= 0380000 119908
2= 0620000 119911 = 1189867
1199081= 0390000 119908
2= 0610000 119911 = 1192715
1199081= 0400000 119908
2= 0600000 119911 = 1195567
1199081= 0410000 119908
2= 0590000 119911 = 1198422
1199081= 0420000 119908
2= 0580000 119911 = 1201279
1199081= 0430000 119908
2= 0570000 119911 = 1204140
1199081= 0440000 119908
2= 0560000 119911 = 1207004
1199081= 0450000 119908
2= 0550000 119911 = 1209870
1199081= 0460000 119908
2= 0540000 119911 = 1212740
1199081= 0470000 119908
2= 0530000 119911 = 1215612
1199081= 0480000 119908
2= 0520000 119911 = 1218488
1199081= 0490000 119908
2= 0510000 119911 = 1221366
Table 2 Continued
1199081= 0500000 119908
2= 0500000 119911 = 1224248
1199081= 0510000 119908
2= 0490000 119911 = 1227132
1199081= 0520000 119908
2= 0480000 119911 = 1230019
1199081= 0530000 119908
2= 0470000 119911 = 1232910
1199081= 0540000 119908
2= 0460000 119911 = 1235803
1199081= 0550000 119908
2= 0450000 119911 = 1238699
1199081= 0560000 119908
2= 0440000 119911 = 1241599
1199081= 0570000 119908
2= 0430000 119911 = 1244501
1199081= 0580000 119908
2= 0420000 119911 = 1247406
1199081= 0590000 119908
2= 0410000 119911 = 1250314
1199081= 0600000 119908
2= 0400000 119911 = 1253225
1199081= 0610000 119908
2= 0390000 119911 = 1256140
1199081= 0620000 119908
2= 0380000 119911 = 1259057
1199081= 0630000 119908
2= 0370000 119911 = 1261977
1199081= 0640000 119908
2= 0360000 119911 = 1264900
1199081= 0650000 119908
2= 0350000 119911 = 1267826
1199081= 0660000 119908
2= 0340000 119911 = 1270755
1199081= 0670000 119908
2= 0330000 119911 = 1273687
1199081= 0680000 119908
2= 0320000 119911 = 1276622
1199081= 0690000 119908
2= 0310000 119911 = 1279560
1199081= 0700000 119908
2= 0300000 119911 = 1282501
1199081= 0710000 119908
2= 0290000 119911 = 1285445
1199081= 0720000 119908
2= 0280000 119911 = 1288392
1199081= 0730000 119908
2= 0270000 119911 = 1291341
1199081= 0740000 119908
2= 0260000 119911 = 1294294
1199081= 0750000 119908
2= 0250000 119911 = 1297250
1199081= 0760000 119908
2= 0240000 119911 = 1300209
1199081= 0770000 119908
2= 0230000 119911 = 1303170
1199081= 0780000 119908
2= 0220000 119911 = 1306135
1199081= 0790000 119908
2= 0210000 119911 = 1309103
1199081= 0800000 119908
2= 0200000 119911 = 1312074
1199081= 0810000 119908
2= 0190000 119911 = 1315047
1199081= 0820000 119908
2= 0180000 119911 = 1318024
1199081= 0830000 119908
2= 0170000 119911 = 1321003
1199081= 0840000 119908
2= 0160000 119911 = 1323986
1199081= 0850000 119908
2= 0150000 119911 = 1326971
1199081= 0860000 119908
2= 0140000 119911 = 1329960
1199081= 0870000 119908
2= 0130000 119911 = 1332951
1199081= 0880000 119908
2= 0120000 119911 = 1335946
1199081= 0890000 119908
2= 0110000 119911 = 1338943
1199081= 0900000 119908
2= 0100000 119911 = 1341944
1199081= 0910000 119908
2= 0090000 119911 = 1344947
1199081= 0920000 119908
2= 0080000 119911 = 1347953
1199081= 0930000 119908
2= 0070000 119911 = 1350963
1199081= 0940000 119908
2= 0060000 119911 = 1353975
1199081= 0950000 119908
2= 0050000 119911 = 1356990
1199081= 0960000 119908
2= 0040000 119911 = 1360008
1199081= 0970000 119908
2= 0030000 119911 = 1363030
1199081= 0980000 119908
2= 0020000 119911 = 1366054
1199081= 0990000 119908
2= 0010000 119911 = 1369081
1199081= 1000000 119908
2= 0000000 119911 = 1372111
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Mathematical Problems in Engineering
1198761is measured in kilos 119876
2in seconds etc) For this reason
the objective function normalization is required It is nowobvious that the coefficients have values of the segment [0 1]Denote that we now have the linear programming problem
Maximize 119891 (119909) =119897
sum119896=1
120596119896
119876119896(119909)
119878119896
= 1205961
1198861198961
119878119896
1199091
+ sdot sdot sdot + 120596119899
119886119896119899
119878119896
119909119899
Subject to 119909 isin 119883
(20)
The following theorem gives the practical criteria forthe detection of some Pareto optimal solutions of problem(24) Yu et al [14] presented mean-CVaR model of portfoliooptimization based on LWSM inwhich they assumed that thereturn of the portfolio is based on multi-t-distribution andthey obtained the model for
max (119906 (119908)) = max (1205721119864 (119877) minus 120572
2CVaR) (21)
as follows
max05 (119908119879120583) + 05(120572 + 1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(22)
There are some conditions for coefficients that are usedin WSM For example the sum of them equal 1 and theyare strictly positive In addition to LWSM they must benormalized With this idea the multiportfolio optimizationproblem is as follows
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR)
Subject to 119909 isin 119860(23)
where 1205721 1205722 and 120572
3are strictly positive and 119860 is as the
above assumptions Since all objective functions on MVCmodel are convex we can define the model above to find thesolution as the Pareto optimal set [15] For instance if weconsider 120572
1= 1 120572
2= 0 and 120572
3= 0 the problem turns
to minimize the variance of the portfolio [5]Now according to (P2) as themultiobjective optimization
problem LWSM will be used to solve it as in the followingprocedure
Step 1 Construct the problem based on
min (119906 (119908)) = min (12057211205902 (119877) minus 120572
2119864 (119877) + 120572
3CVaR) (24)
There are several positions of investing for investorwho can manage hisher budget which is related to theamount of 120572
1 1205722 and 120572
3 This note creates several cases
for investors to increase the return or decrease the riskof a portfolio The risk (or return) has an inverse (direct)relation with those coefficientsThe LWSM leads to followingprocedure for MVC model of portfolio
Step 2 Consider
min1205721(119908119879120590119908) minus 120572
2(119908119879120583)
+ 1205723(120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(25)
where 120572 120573 are the belief degree and specific parameterrespectively In this paper it is considered that 120573 = 05 Inaddition according to sum119899
119894=1120572119894= 1 we have 1minus(120572
1+ 1205722) = 1205723
we rewrite problem of (18) as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (120572 +1
119898 (1 minus 05)
119898
sum119896=1
119906119896)
st
119908119879120583 + 120572 + 119906119896ge 0 119906
119896ge 0
119899
sum119894=1
119908119894
0 le 119908119894le 1
(26)
where the belief degree of 120572 set is equal to 095 In additionin CVaR formula after linearization [16 17] we have
119865120573(119908 120572) = 120572 +
1
119898 (1 minus 120573)
119898
sum119896=1
[119891 (119908 119910119896) + 120572]
+
(27)
and 119891(119908 119910119896) is the loss function which can be defined [16]
by 119891(119908 119910119896) = minus119908119879119910
119896 We can rewrite (19) according to the
assumptions above as follows
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
119898 (1 minus 05)
119898
sum119896=1
[minus119908119879119910119896+ 095]
+
)
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(28)
where 120583119896is return of 119896th asset In the model above we
consider119898 = 2 and we have
min1205721(119908119879120590119908) minus 120572
2(119908119879120583) + (1 minus (120572
1+ 1205722))
times (095 +1
2 (1 minus 05)2 ([minus119908119879119910
119896+ 095]
+
))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(29)
Mathematical Problems in Engineering 5
4 Empirical Research and Challenges
In order to evaluate the model above to find the portfoliooptimization based on LWSM we consider two assets whosemean return 120583 and covariance matrix 120590 are as the followingdata (data from [14 16])
120583 = (0067609870034191
)
120590 = (0058884101 00004640430000464043 00003643663
)
(30)
According to the data above and the model of (29) we have
min119885 = min1205721(1199111) minus 1205722(1199112) + (1 minus (120572
1+ 1205722)) (1199113) (31)
where
1199111= 1205752119901= 119908119879120590119908 =
= (0058884101 lowast 1199081+ 0000464043 lowast 119908
2) lowast 1199081
+ (0000464043 lowast 1199081+ 00003643663 lowast 119908
2) lowast 1199082
1199112= 120583119901= 119908119879120583 = 006760987119908
1+ 0034191119908
2
1199113= (095 +
1
2 (1 minus 05)2 ([minus119908119879120583 minus 095]
+
))
= 095 +1
2 (05)
times 2 ([minus (0067609871199081+ 0034191119908
2) + 095]
+
)
[119906]+ = max 119906 0
119899
sum119894=1
119908119894
= 1 0 le 119908119894le 1
(32)
Now we consider this proposed model for 1205721= 025 120572
2=
025 Therefore we have
minZ = min (001472102525 lowast 11990821
+ 000058005375 lowast 1199081lowast 1199082
+00003643663 lowast 11990822)
minus (001690246751199081+ 000854775119908
2)
+ (0475 + ([minus (0067609871199081+ 0034191119908
2)
+095]+))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(33)
01000
20003000
minus001
minus0005
0minus1
minus05
0
05
1
10
20
30
40
50
60
Figure 1 Mean-variance model of portfolio via LWSM
0
0
2000 4000 6000 8000 10000 120000
1
2
minus1
minus05
05
1
10
20
30
40
50
60
Data 1
Figure 2 MVC model of portfolio via LWSM
In order to compare the performance of mean-variancemodel and MVC model of multiportfolio optimization theminimum functions are demonstrated viaMATLABsoftwareas Figures 1 and 2 show the results of LWSMmethod for twomodels respectively
Comparative performance was measured using a MAT-LAB software and performance was shown in Figures 1 and2 which are called the mean-variance and MVC model ofmultiportfolio respectively
In summary we compare the performance of MVCmodel of multiportfolio with LWSM technique and othermodels like mean-variance approaches The mean-varianceminimization function and the MVC model are as describedin Figures 1 and 2 respectively
Table 2 is arranged to illustrate the results of MVCmodelof portfolio and refer to two assets byMatlab by this approach
As shown in the above results we tested two assets viaproposed model and we presented the results of that asTable 2 which included 20 results from our proposed modelbyMATLAB software It showed except for two obvious cases
6 Mathematical Problems in Engineering
Table 2 Results of MVC model of portfolio refer to two assets (seeSection 4) by MATLAB
1199081= 0000000 119908
2= 1000000 119911 = 1083819
1199081= 0010000 119908
2= 0990000 119911 = 1086554
1199081= 0020000 119908
2= 0980000 119911 = 1089293
1199081= 0030000 119908
2= 0970000 119911 = 1092035
1199081= 0040000 119908
2= 0960000 119911 = 1094779
1199081= 0050000 119908
2= 0950000 119911 = 1097527
1199081= 0060000 119908
2= 0940000 119911 = 1100278
1199081= 0070000 119908
2= 0930000 119911 = 1103031
1199081= 0080000 119908
2= 0920000 119911 = 1105788
1199081= 0090000 119908
2= 0910000 119911 = 1108547
1199081= 0100000 119908
2= 0900000 119911 = 1111310
1199081= 0110000 119908
2= 0890000 119911 = 1114075
1199081= 0120000 119908
2= 0880000 119911 = 1116844
1199081= 0130000 119908
2= 0870000 119911 = 1119615
1199081= 0140000 119908
2= 0860000 119911 = 1122389
1199081= 0150000 119908
2= 0850000 119911 = 1125167
1199081= 0160000 119908
2= 0840000 119911 = 1127947
1199081= 0170000 119908
2= 0830000 119911 = 1130730
1199081= 0180000 119908
2= 0820000 119911 = 1133517
1199081= 0190000 119908
2= 0810000 119911 = 1136306
1199081= 0200000 119908
2= 0800000 119911 = 1139098
1199081= 0210000 119908
2= 0790000 119911 = 1141893
1199081= 0220000 119908
2= 0780000 119911 = 1144691
1199081= 0230000 119908
2= 0770000 119911 = 1147493
1199081= 0240000 119908
2= 0760000 119911 = 1150297
1199081= 0250000 119908
2= 0750000 119911 = 1153104
1199081= 0260000 119908
2= 0740000 119911 = 1155914
1199081= 0270000 119908
2= 0730000 119911 = 1158727
1199081= 0280000 119908
2= 0720000 119911 = 1161543
1199081= 0290000 119908
2= 0710000 119911 = 1164362
1199081= 0300000 119908
2= 0700000 119911 = 1167184
1199081= 0310000 119908
2= 0690000 119911 = 1170009
1199081= 0320000 119908
2= 0680000 119911 = 1172837
1199081= 0330000 119908
2= 0670000 119911 = 1175668
1199081= 0340000 119908
2= 0660000 119911 = 1178501
1199081= 0350000 119908
2= 0650000 119911 = 1181338
1199081= 0360000 119908
2= 0640000 119911 = 1184178
1199081= 0370000 119908
2= 0630000 119911 = 1187021
1199081= 0380000 119908
2= 0620000 119911 = 1189867
1199081= 0390000 119908
2= 0610000 119911 = 1192715
1199081= 0400000 119908
2= 0600000 119911 = 1195567
1199081= 0410000 119908
2= 0590000 119911 = 1198422
1199081= 0420000 119908
2= 0580000 119911 = 1201279
1199081= 0430000 119908
2= 0570000 119911 = 1204140
1199081= 0440000 119908
2= 0560000 119911 = 1207004
1199081= 0450000 119908
2= 0550000 119911 = 1209870
1199081= 0460000 119908
2= 0540000 119911 = 1212740
1199081= 0470000 119908
2= 0530000 119911 = 1215612
1199081= 0480000 119908
2= 0520000 119911 = 1218488
1199081= 0490000 119908
2= 0510000 119911 = 1221366
Table 2 Continued
1199081= 0500000 119908
2= 0500000 119911 = 1224248
1199081= 0510000 119908
2= 0490000 119911 = 1227132
1199081= 0520000 119908
2= 0480000 119911 = 1230019
1199081= 0530000 119908
2= 0470000 119911 = 1232910
1199081= 0540000 119908
2= 0460000 119911 = 1235803
1199081= 0550000 119908
2= 0450000 119911 = 1238699
1199081= 0560000 119908
2= 0440000 119911 = 1241599
1199081= 0570000 119908
2= 0430000 119911 = 1244501
1199081= 0580000 119908
2= 0420000 119911 = 1247406
1199081= 0590000 119908
2= 0410000 119911 = 1250314
1199081= 0600000 119908
2= 0400000 119911 = 1253225
1199081= 0610000 119908
2= 0390000 119911 = 1256140
1199081= 0620000 119908
2= 0380000 119911 = 1259057
1199081= 0630000 119908
2= 0370000 119911 = 1261977
1199081= 0640000 119908
2= 0360000 119911 = 1264900
1199081= 0650000 119908
2= 0350000 119911 = 1267826
1199081= 0660000 119908
2= 0340000 119911 = 1270755
1199081= 0670000 119908
2= 0330000 119911 = 1273687
1199081= 0680000 119908
2= 0320000 119911 = 1276622
1199081= 0690000 119908
2= 0310000 119911 = 1279560
1199081= 0700000 119908
2= 0300000 119911 = 1282501
1199081= 0710000 119908
2= 0290000 119911 = 1285445
1199081= 0720000 119908
2= 0280000 119911 = 1288392
1199081= 0730000 119908
2= 0270000 119911 = 1291341
1199081= 0740000 119908
2= 0260000 119911 = 1294294
1199081= 0750000 119908
2= 0250000 119911 = 1297250
1199081= 0760000 119908
2= 0240000 119911 = 1300209
1199081= 0770000 119908
2= 0230000 119911 = 1303170
1199081= 0780000 119908
2= 0220000 119911 = 1306135
1199081= 0790000 119908
2= 0210000 119911 = 1309103
1199081= 0800000 119908
2= 0200000 119911 = 1312074
1199081= 0810000 119908
2= 0190000 119911 = 1315047
1199081= 0820000 119908
2= 0180000 119911 = 1318024
1199081= 0830000 119908
2= 0170000 119911 = 1321003
1199081= 0840000 119908
2= 0160000 119911 = 1323986
1199081= 0850000 119908
2= 0150000 119911 = 1326971
1199081= 0860000 119908
2= 0140000 119911 = 1329960
1199081= 0870000 119908
2= 0130000 119911 = 1332951
1199081= 0880000 119908
2= 0120000 119911 = 1335946
1199081= 0890000 119908
2= 0110000 119911 = 1338943
1199081= 0900000 119908
2= 0100000 119911 = 1341944
1199081= 0910000 119908
2= 0090000 119911 = 1344947
1199081= 0920000 119908
2= 0080000 119911 = 1347953
1199081= 0930000 119908
2= 0070000 119911 = 1350963
1199081= 0940000 119908
2= 0060000 119911 = 1353975
1199081= 0950000 119908
2= 0050000 119911 = 1356990
1199081= 0960000 119908
2= 0040000 119911 = 1360008
1199081= 0970000 119908
2= 0030000 119911 = 1363030
1199081= 0980000 119908
2= 0020000 119911 = 1366054
1199081= 0990000 119908
2= 0010000 119911 = 1369081
1199081= 1000000 119908
2= 0000000 119911 = 1372111
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 5
4 Empirical Research and Challenges
In order to evaluate the model above to find the portfoliooptimization based on LWSM we consider two assets whosemean return 120583 and covariance matrix 120590 are as the followingdata (data from [14 16])
120583 = (0067609870034191
)
120590 = (0058884101 00004640430000464043 00003643663
)
(30)
According to the data above and the model of (29) we have
min119885 = min1205721(1199111) minus 1205722(1199112) + (1 minus (120572
1+ 1205722)) (1199113) (31)
where
1199111= 1205752119901= 119908119879120590119908 =
= (0058884101 lowast 1199081+ 0000464043 lowast 119908
2) lowast 1199081
+ (0000464043 lowast 1199081+ 00003643663 lowast 119908
2) lowast 1199082
1199112= 120583119901= 119908119879120583 = 006760987119908
1+ 0034191119908
2
1199113= (095 +
1
2 (1 minus 05)2 ([minus119908119879120583 minus 095]
+
))
= 095 +1
2 (05)
times 2 ([minus (0067609871199081+ 0034191119908
2) + 095]
+
)
[119906]+ = max 119906 0
119899
sum119894=1
119908119894
= 1 0 le 119908119894le 1
(32)
Now we consider this proposed model for 1205721= 025 120572
2=
025 Therefore we have
minZ = min (001472102525 lowast 11990821
+ 000058005375 lowast 1199081lowast 1199082
+00003643663 lowast 11990822)
minus (001690246751199081+ 000854775119908
2)
+ (0475 + ([minus (0067609871199081+ 0034191119908
2)
+095]+))
st
[119906]+ = max 119906 0119899
sum119894=1
119908119894= 1 0 le 119908
119894le 1
(33)
01000
20003000
minus001
minus0005
0minus1
minus05
0
05
1
10
20
30
40
50
60
Figure 1 Mean-variance model of portfolio via LWSM
0
0
2000 4000 6000 8000 10000 120000
1
2
minus1
minus05
05
1
10
20
30
40
50
60
Data 1
Figure 2 MVC model of portfolio via LWSM
In order to compare the performance of mean-variancemodel and MVC model of multiportfolio optimization theminimum functions are demonstrated viaMATLABsoftwareas Figures 1 and 2 show the results of LWSMmethod for twomodels respectively
Comparative performance was measured using a MAT-LAB software and performance was shown in Figures 1 and2 which are called the mean-variance and MVC model ofmultiportfolio respectively
In summary we compare the performance of MVCmodel of multiportfolio with LWSM technique and othermodels like mean-variance approaches The mean-varianceminimization function and the MVC model are as describedin Figures 1 and 2 respectively
Table 2 is arranged to illustrate the results of MVCmodelof portfolio and refer to two assets byMatlab by this approach
As shown in the above results we tested two assets viaproposed model and we presented the results of that asTable 2 which included 20 results from our proposed modelbyMATLAB software It showed except for two obvious cases
6 Mathematical Problems in Engineering
Table 2 Results of MVC model of portfolio refer to two assets (seeSection 4) by MATLAB
1199081= 0000000 119908
2= 1000000 119911 = 1083819
1199081= 0010000 119908
2= 0990000 119911 = 1086554
1199081= 0020000 119908
2= 0980000 119911 = 1089293
1199081= 0030000 119908
2= 0970000 119911 = 1092035
1199081= 0040000 119908
2= 0960000 119911 = 1094779
1199081= 0050000 119908
2= 0950000 119911 = 1097527
1199081= 0060000 119908
2= 0940000 119911 = 1100278
1199081= 0070000 119908
2= 0930000 119911 = 1103031
1199081= 0080000 119908
2= 0920000 119911 = 1105788
1199081= 0090000 119908
2= 0910000 119911 = 1108547
1199081= 0100000 119908
2= 0900000 119911 = 1111310
1199081= 0110000 119908
2= 0890000 119911 = 1114075
1199081= 0120000 119908
2= 0880000 119911 = 1116844
1199081= 0130000 119908
2= 0870000 119911 = 1119615
1199081= 0140000 119908
2= 0860000 119911 = 1122389
1199081= 0150000 119908
2= 0850000 119911 = 1125167
1199081= 0160000 119908
2= 0840000 119911 = 1127947
1199081= 0170000 119908
2= 0830000 119911 = 1130730
1199081= 0180000 119908
2= 0820000 119911 = 1133517
1199081= 0190000 119908
2= 0810000 119911 = 1136306
1199081= 0200000 119908
2= 0800000 119911 = 1139098
1199081= 0210000 119908
2= 0790000 119911 = 1141893
1199081= 0220000 119908
2= 0780000 119911 = 1144691
1199081= 0230000 119908
2= 0770000 119911 = 1147493
1199081= 0240000 119908
2= 0760000 119911 = 1150297
1199081= 0250000 119908
2= 0750000 119911 = 1153104
1199081= 0260000 119908
2= 0740000 119911 = 1155914
1199081= 0270000 119908
2= 0730000 119911 = 1158727
1199081= 0280000 119908
2= 0720000 119911 = 1161543
1199081= 0290000 119908
2= 0710000 119911 = 1164362
1199081= 0300000 119908
2= 0700000 119911 = 1167184
1199081= 0310000 119908
2= 0690000 119911 = 1170009
1199081= 0320000 119908
2= 0680000 119911 = 1172837
1199081= 0330000 119908
2= 0670000 119911 = 1175668
1199081= 0340000 119908
2= 0660000 119911 = 1178501
1199081= 0350000 119908
2= 0650000 119911 = 1181338
1199081= 0360000 119908
2= 0640000 119911 = 1184178
1199081= 0370000 119908
2= 0630000 119911 = 1187021
1199081= 0380000 119908
2= 0620000 119911 = 1189867
1199081= 0390000 119908
2= 0610000 119911 = 1192715
1199081= 0400000 119908
2= 0600000 119911 = 1195567
1199081= 0410000 119908
2= 0590000 119911 = 1198422
1199081= 0420000 119908
2= 0580000 119911 = 1201279
1199081= 0430000 119908
2= 0570000 119911 = 1204140
1199081= 0440000 119908
2= 0560000 119911 = 1207004
1199081= 0450000 119908
2= 0550000 119911 = 1209870
1199081= 0460000 119908
2= 0540000 119911 = 1212740
1199081= 0470000 119908
2= 0530000 119911 = 1215612
1199081= 0480000 119908
2= 0520000 119911 = 1218488
1199081= 0490000 119908
2= 0510000 119911 = 1221366
Table 2 Continued
1199081= 0500000 119908
2= 0500000 119911 = 1224248
1199081= 0510000 119908
2= 0490000 119911 = 1227132
1199081= 0520000 119908
2= 0480000 119911 = 1230019
1199081= 0530000 119908
2= 0470000 119911 = 1232910
1199081= 0540000 119908
2= 0460000 119911 = 1235803
1199081= 0550000 119908
2= 0450000 119911 = 1238699
1199081= 0560000 119908
2= 0440000 119911 = 1241599
1199081= 0570000 119908
2= 0430000 119911 = 1244501
1199081= 0580000 119908
2= 0420000 119911 = 1247406
1199081= 0590000 119908
2= 0410000 119911 = 1250314
1199081= 0600000 119908
2= 0400000 119911 = 1253225
1199081= 0610000 119908
2= 0390000 119911 = 1256140
1199081= 0620000 119908
2= 0380000 119911 = 1259057
1199081= 0630000 119908
2= 0370000 119911 = 1261977
1199081= 0640000 119908
2= 0360000 119911 = 1264900
1199081= 0650000 119908
2= 0350000 119911 = 1267826
1199081= 0660000 119908
2= 0340000 119911 = 1270755
1199081= 0670000 119908
2= 0330000 119911 = 1273687
1199081= 0680000 119908
2= 0320000 119911 = 1276622
1199081= 0690000 119908
2= 0310000 119911 = 1279560
1199081= 0700000 119908
2= 0300000 119911 = 1282501
1199081= 0710000 119908
2= 0290000 119911 = 1285445
1199081= 0720000 119908
2= 0280000 119911 = 1288392
1199081= 0730000 119908
2= 0270000 119911 = 1291341
1199081= 0740000 119908
2= 0260000 119911 = 1294294
1199081= 0750000 119908
2= 0250000 119911 = 1297250
1199081= 0760000 119908
2= 0240000 119911 = 1300209
1199081= 0770000 119908
2= 0230000 119911 = 1303170
1199081= 0780000 119908
2= 0220000 119911 = 1306135
1199081= 0790000 119908
2= 0210000 119911 = 1309103
1199081= 0800000 119908
2= 0200000 119911 = 1312074
1199081= 0810000 119908
2= 0190000 119911 = 1315047
1199081= 0820000 119908
2= 0180000 119911 = 1318024
1199081= 0830000 119908
2= 0170000 119911 = 1321003
1199081= 0840000 119908
2= 0160000 119911 = 1323986
1199081= 0850000 119908
2= 0150000 119911 = 1326971
1199081= 0860000 119908
2= 0140000 119911 = 1329960
1199081= 0870000 119908
2= 0130000 119911 = 1332951
1199081= 0880000 119908
2= 0120000 119911 = 1335946
1199081= 0890000 119908
2= 0110000 119911 = 1338943
1199081= 0900000 119908
2= 0100000 119911 = 1341944
1199081= 0910000 119908
2= 0090000 119911 = 1344947
1199081= 0920000 119908
2= 0080000 119911 = 1347953
1199081= 0930000 119908
2= 0070000 119911 = 1350963
1199081= 0940000 119908
2= 0060000 119911 = 1353975
1199081= 0950000 119908
2= 0050000 119911 = 1356990
1199081= 0960000 119908
2= 0040000 119911 = 1360008
1199081= 0970000 119908
2= 0030000 119911 = 1363030
1199081= 0980000 119908
2= 0020000 119911 = 1366054
1199081= 0990000 119908
2= 0010000 119911 = 1369081
1199081= 1000000 119908
2= 0000000 119911 = 1372111
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Mathematical Problems in Engineering
Table 2 Results of MVC model of portfolio refer to two assets (seeSection 4) by MATLAB
1199081= 0000000 119908
2= 1000000 119911 = 1083819
1199081= 0010000 119908
2= 0990000 119911 = 1086554
1199081= 0020000 119908
2= 0980000 119911 = 1089293
1199081= 0030000 119908
2= 0970000 119911 = 1092035
1199081= 0040000 119908
2= 0960000 119911 = 1094779
1199081= 0050000 119908
2= 0950000 119911 = 1097527
1199081= 0060000 119908
2= 0940000 119911 = 1100278
1199081= 0070000 119908
2= 0930000 119911 = 1103031
1199081= 0080000 119908
2= 0920000 119911 = 1105788
1199081= 0090000 119908
2= 0910000 119911 = 1108547
1199081= 0100000 119908
2= 0900000 119911 = 1111310
1199081= 0110000 119908
2= 0890000 119911 = 1114075
1199081= 0120000 119908
2= 0880000 119911 = 1116844
1199081= 0130000 119908
2= 0870000 119911 = 1119615
1199081= 0140000 119908
2= 0860000 119911 = 1122389
1199081= 0150000 119908
2= 0850000 119911 = 1125167
1199081= 0160000 119908
2= 0840000 119911 = 1127947
1199081= 0170000 119908
2= 0830000 119911 = 1130730
1199081= 0180000 119908
2= 0820000 119911 = 1133517
1199081= 0190000 119908
2= 0810000 119911 = 1136306
1199081= 0200000 119908
2= 0800000 119911 = 1139098
1199081= 0210000 119908
2= 0790000 119911 = 1141893
1199081= 0220000 119908
2= 0780000 119911 = 1144691
1199081= 0230000 119908
2= 0770000 119911 = 1147493
1199081= 0240000 119908
2= 0760000 119911 = 1150297
1199081= 0250000 119908
2= 0750000 119911 = 1153104
1199081= 0260000 119908
2= 0740000 119911 = 1155914
1199081= 0270000 119908
2= 0730000 119911 = 1158727
1199081= 0280000 119908
2= 0720000 119911 = 1161543
1199081= 0290000 119908
2= 0710000 119911 = 1164362
1199081= 0300000 119908
2= 0700000 119911 = 1167184
1199081= 0310000 119908
2= 0690000 119911 = 1170009
1199081= 0320000 119908
2= 0680000 119911 = 1172837
1199081= 0330000 119908
2= 0670000 119911 = 1175668
1199081= 0340000 119908
2= 0660000 119911 = 1178501
1199081= 0350000 119908
2= 0650000 119911 = 1181338
1199081= 0360000 119908
2= 0640000 119911 = 1184178
1199081= 0370000 119908
2= 0630000 119911 = 1187021
1199081= 0380000 119908
2= 0620000 119911 = 1189867
1199081= 0390000 119908
2= 0610000 119911 = 1192715
1199081= 0400000 119908
2= 0600000 119911 = 1195567
1199081= 0410000 119908
2= 0590000 119911 = 1198422
1199081= 0420000 119908
2= 0580000 119911 = 1201279
1199081= 0430000 119908
2= 0570000 119911 = 1204140
1199081= 0440000 119908
2= 0560000 119911 = 1207004
1199081= 0450000 119908
2= 0550000 119911 = 1209870
1199081= 0460000 119908
2= 0540000 119911 = 1212740
1199081= 0470000 119908
2= 0530000 119911 = 1215612
1199081= 0480000 119908
2= 0520000 119911 = 1218488
1199081= 0490000 119908
2= 0510000 119911 = 1221366
Table 2 Continued
1199081= 0500000 119908
2= 0500000 119911 = 1224248
1199081= 0510000 119908
2= 0490000 119911 = 1227132
1199081= 0520000 119908
2= 0480000 119911 = 1230019
1199081= 0530000 119908
2= 0470000 119911 = 1232910
1199081= 0540000 119908
2= 0460000 119911 = 1235803
1199081= 0550000 119908
2= 0450000 119911 = 1238699
1199081= 0560000 119908
2= 0440000 119911 = 1241599
1199081= 0570000 119908
2= 0430000 119911 = 1244501
1199081= 0580000 119908
2= 0420000 119911 = 1247406
1199081= 0590000 119908
2= 0410000 119911 = 1250314
1199081= 0600000 119908
2= 0400000 119911 = 1253225
1199081= 0610000 119908
2= 0390000 119911 = 1256140
1199081= 0620000 119908
2= 0380000 119911 = 1259057
1199081= 0630000 119908
2= 0370000 119911 = 1261977
1199081= 0640000 119908
2= 0360000 119911 = 1264900
1199081= 0650000 119908
2= 0350000 119911 = 1267826
1199081= 0660000 119908
2= 0340000 119911 = 1270755
1199081= 0670000 119908
2= 0330000 119911 = 1273687
1199081= 0680000 119908
2= 0320000 119911 = 1276622
1199081= 0690000 119908
2= 0310000 119911 = 1279560
1199081= 0700000 119908
2= 0300000 119911 = 1282501
1199081= 0710000 119908
2= 0290000 119911 = 1285445
1199081= 0720000 119908
2= 0280000 119911 = 1288392
1199081= 0730000 119908
2= 0270000 119911 = 1291341
1199081= 0740000 119908
2= 0260000 119911 = 1294294
1199081= 0750000 119908
2= 0250000 119911 = 1297250
1199081= 0760000 119908
2= 0240000 119911 = 1300209
1199081= 0770000 119908
2= 0230000 119911 = 1303170
1199081= 0780000 119908
2= 0220000 119911 = 1306135
1199081= 0790000 119908
2= 0210000 119911 = 1309103
1199081= 0800000 119908
2= 0200000 119911 = 1312074
1199081= 0810000 119908
2= 0190000 119911 = 1315047
1199081= 0820000 119908
2= 0180000 119911 = 1318024
1199081= 0830000 119908
2= 0170000 119911 = 1321003
1199081= 0840000 119908
2= 0160000 119911 = 1323986
1199081= 0850000 119908
2= 0150000 119911 = 1326971
1199081= 0860000 119908
2= 0140000 119911 = 1329960
1199081= 0870000 119908
2= 0130000 119911 = 1332951
1199081= 0880000 119908
2= 0120000 119911 = 1335946
1199081= 0890000 119908
2= 0110000 119911 = 1338943
1199081= 0900000 119908
2= 0100000 119911 = 1341944
1199081= 0910000 119908
2= 0090000 119911 = 1344947
1199081= 0920000 119908
2= 0080000 119911 = 1347953
1199081= 0930000 119908
2= 0070000 119911 = 1350963
1199081= 0940000 119908
2= 0060000 119911 = 1353975
1199081= 0950000 119908
2= 0050000 119911 = 1356990
1199081= 0960000 119908
2= 0040000 119911 = 1360008
1199081= 0970000 119908
2= 0030000 119911 = 1363030
1199081= 0980000 119908
2= 0020000 119911 = 1366054
1199081= 0990000 119908
2= 0010000 119911 = 1369081
1199081= 1000000 119908
2= 0000000 119911 = 1372111
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 7
(1199081= 0 and 119908
2= 1 119908
1= 1 and 119908
2= 0) that the minimum
situation ofMVCmodel ofmultiportfoliowill occurr on1199081=
001 and 1199082= 099 The results illustrate that the proposed
method is better for investing a small amount of assets suchas the above two assets
Despite of the several advantages this method has somedisadvantages which one of them is the weights 120572
1 1205722
and 1205723that are hard to understand [5] The other one is
how to use the Simplex method to solve the weightedsum problem The solution of this problem is using othertechniques (eg interior-point methods) [10]
5 Conclusion
In this investigation we consider three objective functions tomodel portfolio optimization with LWSM In addition wepropose a linear weighted sum method to solve the MCVmodel of portfolio optimization Also with empirical exam-ple and MATLAB software we evaluated proposed modelOur example includes two assets for investing However itis flexible to choose more assets and find the Pareto optimalaccording to procedures that are used in this paper
The contribution of this study is a presentation of MVCmodel of portfolio optimization based on LWSMThe empir-ical examples have shown that using the three objectivefunctions tomake a powerfulmulti-portfoliomodelWe havedescribed the MVC model of multiportfolio optimizationthat it is extended of the model presented by Yu et al [14]In addition we showed that theMVCmodel generally coversboth mean-variance and mean-CVaR models of portfoliooptimization (with changing the coefficients of 120572
119894 119894 = 1 3)
Furthermore this paper has given new insight into themultiobjective decision making process via WSM A hybridof this method and other techniques such as interior-pointmethods can be a good idea for future researchers
To sum up with using three objective functions it helpsthe investors to manage their portfolio better and therebyminimize the risk and maximize the return of the portfolioAlso using LWSM dues to modify the current models andsimplify it to obtain better results The same approach canalso be used for other multiobjective optimizationmodel andinvesting applications
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
Research reported in this publication was supported byUniversiti Teknologi Malaysia under grant number 4B112(Research Management Centre Universiti TeknologiMalaysia)
References
[1] Y Elahi and A A Mohd-Ismail ldquoMulti-objectives portfo-lio optimization challenges and opportunities for islamicapproachrdquoAustralian Journal of Basic and Applaied Sicence vol6 no 10 pp 297ndash302 2012
[2] I Radziukyniene and A Zilinskas ldquoEvolutionary methods formulti-objective portfolio optimizationrdquo in Proceedings of theWorld Congress on Engineering 2008
[3] C Chiranjeevi and V N Sastry ldquoMulti objective portfoliooptimization models and its solution using genetic algorithmsrdquoin Proceedings of the International Conference on ComputationalIntelligence andMultimediaApplications (ICCIMA rsquo07) Decem-ber 2007
[4] R Aboulaich R Ellaia and S El Moumen ldquoThe mean-variance-CVaR model for portfolio optimization modelingusing a multi-objective approach based on a hybrid methodrdquoMathematical Modelling of Natural Phenomena vol 5 no 7 pp103ndash108 2010
[5] D Roman K Darby-Dowman and G Mitra ldquoMean-riskmodels using two risk measures a multi-objective approachrdquoQuantitative Finance vol 7 no 4 pp 443ndash458 2007
[6] R Armananzas and J A Lozano ldquoAmultiobjective approach tothe portfolio optimization problemrdquo in Proceedings of the IEEECongress on Evolutionary Computation (CEC rsquo05) September2005
[7] D Roman K Darby-Dowman and G Mitra ldquoPortfolio con-struction based on stochastic dominance and target returndistributionsrdquo Mathematical Programming vol 108 no 2 pp541ndash569 2006
[8] L Zadeh ldquoOptimality and non-scalar-valued performancecriteriardquo IEEE Transactions on Automatic Control vol 8 no 1pp 59ndash60 1963
[9] K Kirytopoulos V Leopoulos G Mavrotas and D Voulgar-idou ldquoMultiple sourcing strategies and order allocation anANP-AUGMECON meta-modelrdquo Supply Chain Managementvol 15 no 4 pp 263ndash276 2010
[10] O Grodzevich and O Romanko ldquoNormalization and othertopics in multi-objective optimizationrdquo 2006
[11] J Jia GW Fischer and J S Dyer ldquoAttribute weightingmethodsand decision quality in the presence of response error asimulation studyrdquo Journal of Behavioral Decision Making vol11 no 2 pp 85ndash105 1998
[12] B Sawik ldquoA reference point approach to bi-objective dynamicportfolio optimizationrdquo Decision Making in Manufacturing andServices vol 3 no 1-2 pp 73ndash85 2009
[13] O Romanko A Ghaffari-Hadigheh and T Terlaky ldquoMulti-objective optimization via parametric optimization modelsalgorithms and applicationsrdquo Modeling and Optimization pp77ndash119 2012
[14] X Yu Y Tan L Liu and W Huang ldquoThe optimal portfoliomodel based onmean-CvaRwith linear weighted summethodrdquoin Fifth International Joint Conference on Computational Sci-ences and Optimization (CSO rsquo12) pp 82ndash84 June 2012
[15] J Jahn ldquoSome characterizations of the optimal solutions of avector optimization problemrdquo Operations-Research-Spektrumvol 7 pp 7ndash17 1985
[16] X YuH Sun andG Chen ldquoThe optimal portfoliomodel basedon mean-CVaRrdquo Journal of Mathematical Finance vol 1 no 3pp 132ndash134 2011
[17] S Uryasev ldquoConditional value-at-risk optimization algorithmsand applicationsrdquo in Proceedings of the IEEEIAFEINFORNS6th Conference on Computational Intelligence for FinancialEngineering (CIFEr rsquo00) pp 49ndash57 March 2000
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of