Constrained Portfolio Optimisation: The State-of-the-Art Markowitz
Models
Yan Jin, Rong Qu and Jason Atkin
ASAP Group, School of Computer Science, The University of Nottingham, Nottingham, U.K.
Keywords:
Constrained Portfolio Optimization, Mean-variance, Cardinality, Pre-assignment, Round-lot, Class.
Abstract:
This paper studies the state-of-art constrained portfolio optimization models, using exact solver to identify the
optimal solutions or lower bound for the benchmark instances at the OR-library with extended constraints. The
effects of pre-assignment, round-lot, and class constraints based on the quantity and cardinality constrained
Markowitz model are firstly investigated to gain insights of increased problem difficulty, followed by the
analysis of various constraint settings including those mostly studied in the literature. The study aims to
provide useful guidance for future investigations in computational algorithms.
1 INTRODUCTION
Portfolio optimization (PO) is an extensively studied
area in finance. The seminal work by (Markowitz,
1952) had a profound impact on the development of
PO in the last 60 years. The Mean-Variance (MV)
model introduced by Markowitz focusses on finding
the best trade-off between the return and risk of port-
folios, i.e. mean of return and covariance of return,
to minimise the risk given an expected return level or
vice versa.
Although the significance of the MV model is
unanimously recognized, the basic model has been
widely challenged for some underlying assumptions.
It neglects many realistic restrictions faced by in-
vestors like tax and transaction cost; personal or
strategic investment decisions, etc. It assumes as-
sets are traded at any fractions. It implicitly encour-
ages holdings of as many assets as possible to diver-
sify the overall risk. In reality, an investment man-
ager may face the restrictions on the minimum and/or
maximum capital allocated to an asset or industry. In-
vestors also prefer a limited number of assets (Jansen
and van Dijk, 2002).
The complexity of the PO problem to a large ex-
tent depends on the constraints (Maringer, 2008). The
basic MV model is a standard quadratic programming
problem. There has been numerous tools to solve it
optimally. However real-world financial constraints
significantly increase the level of complexity. For in-
stance, cardinality constraint requires only a limited
number of assets to be included in the portfolio, which
turns the problem into non-convex. It is no longer
always suitable to use exact methods to find optimal
solutions thus the majority of work in the current lit-
erature has focused on heuristics for the constrained
PO problem.
Nevertheless, due to the fast development some
constraints now can be handled by commercial
solvers such as CPLEX with limited computational
cost for many difficult optimisation problems. The
purpose of our study is to provide an insight into the
current state of the MV models with subsets of prac-
tical constraints using CPLEX, and provide useful
guidance of potential areas of future research on com-
putational algorithms for constrained PO problems.
This paper is organised as follows. Firstly, we
overview the basic MV model and the extended con-
straints in Section 2. Then a comprehensive overview
of related literature for various settings of practical
constraints is presented in Section 3. It summarises
the representative works in terms of different con-
straints applied. Section 4 presents the experimen-
tal study by reviewing the models in the literature
and provides the optimal solutions or lower bound for
those mostly studied models. Finally, in Section 5, we
identify some possible future research on algorithms
for the constrained MV models in portfolio optimisa-
tion.
388
Jin, Y., Qu, R. and Atkin, J.
Constrained Portfolio Optimisation: The State-of-the-Art Markowitz Models.
DOI: 10.5220/0005758303880395
In Proceedings of 5th the International Conference on Operations Research and Enterprise Systems (ICORES 2016), pages 388-395
ISBN: 978-989-758-171-7
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
2 PROBLEM FORMULATION
2.1 Mean-Variance Model
The MV model considers a single period of invest-
ment. The process of PO allocates among N different
assets the proportions (x
i
, i = 1, ...N) of the capital to
form a portfolio. Each asset i has a return rate r
i
and
is associated with a covariance σ
i j
of the return with
each other asset j. The total return r
P
of a portfolio is
given by the weighted combination of the constituent
assets’ returns
N
i=1
x
i
r
i
, and its risk v
P
is defined
by
N
i=1
N
j=1
σ
i j
x
i
x
j
. The aim is to minimize the
portfolio risk v
P
for a given level of expected return
R
exp
or vice versa, mathematical formulation of the
model as follows:
Minimise
v
P
=
N
i=1
N
j=1
σ
i j
x
i
x
j
(1)
Subject to
r
P
=
N
i=1
x
i
r
i
= R
exp
(2)
N
i=1
x
i
= 1 (3)
0 6 x
i
6 1, i = 1, ..., N (4)
Constraint (2) defines the expected return. The
budget constraint (3) requires the whole capital
should be invested. Investment to each asset is non-
negative, defined in constraint (4). A set of optimal
portfolios of the lowest risk for various values of R
exp
can be obtained by solving the above model repeat-
edly, which forms the efficient frontier (EF). For each
point on EF, there should be no portfolio with a higher
expected return at the same risk level, and no portfolio
with a lower risk at the same level of return.
2.2 Additional Practical Constraints
Several extensions have been proposed to enrich the
MV model with real world constraints in the liter-
ature. In this paper, we investigate the following
mostly studied additional practical constraints based
on the basic MV model.
Cardinality Constraint. The cardinality constraint
restricts the number of K assets in a portfolio. A bi-
nary variable z
i
is introduced to denote whether an
asset is selected or not. This constraint is relaxed to
i.e.
N
i=1
z
i
K in some work in the literature.
N
i=1
z
i
= K (5)
z
i
{0, 1}, i = 1, ..., N (6)
Quantity Constraint. The quantity constraint
specifies the lower (ε) and upper (δ) bounds allowed
for the allocated proportions to each asset in a
portfolio.
εz
i
6 x
i
6 δz
i
, i = 1, ..., N (7)
Pre-assignment Constraint. Pre-assignment con-
straint pre-selects investor’s preferred assets in the
portfolio. It was firstly discussed in (Chang et al.,
2000) and firstly applied in (Di Gaspero et al., 2011).
A binary variable s
i
is introduced to denote if asset i
belongs to the pre-assigned set P, 0 otherwise.
s
i
= 1, i P (8)
z
i
> s
i
, i = 1, ..., N (9)
Round Lot Constraint. Round lot constraint de-
fines that the investment of any asset in the portfolio
should be an exact multiple units of a minimum lot.
An integer variable y
i
and minimum lot l
i
for each
asset are introduced. The round lot constraint might
cause the budget constraint (3) not strictly satisfied,
as the capital cannot always be divided as an exact
multiple of trading lot for all the assets.
x
i
= y
i
l
i
, i = 1, ..., N (10)
Class Constraint. Introduced by (Chang et al.,
2000), class constraint is used to limit the total pro-
portion invested in those assets with common charac-
teristics, leading to a more diversified and safe port-
folio. Classes of assets are considered to be mutually
exclusive, i.e. C
i
C
j
= for all assets i 6= j. In this
study we require at least one asset from each of the
M classes to be selected, thus K M. Here the upper
bound is set as 1, and the lower bound of each class
is L
m
> 0 for every class C
m
, m = 1,..., M. Then the
lower bound is formulated as follows:
L
m
6
iC
m
w
i
6 1, m = 1, ...M (11)
3 STUDIES OF VARIOUS
EXTENDED MV MODELS
The basic MV model is a quadratic programming
problem and can be solved efficiently by some spe-
Constrained Portfolio Optimisation: The State-of-the-Art Markowitz Models
389
cialized exact methods such as simplex method and
branch and bound methods. These techniques can
also handle arbitrary linear constraints, like quantity
constraint, see (Borchers and Mitchell, 1997). Never-
theless the problem becomes increasingly much more
complex when the number of assets increases and
with additional constraints. For instance, with the
cardinality constraint, the problem turns into a mixed
integer nonlinear programming and NP-hard (Bien-
stock, 1995). (Bienstock, 1995) presented a branch
and cut algorithm for the cardinality constrained PO
problem of up to 3897 assets with various cardinality
values. At the time of publication results shown that
this type of problem with larger size may be impossi-
ble to solve to a proved optimality within a reasonable
time. Some works require strictly K assets to be in-
cluded in a portfolio ((Chang et al., 2000; Fernandez
and Gomez, 2007; Xu et al., 2010; Woodside-Oriakhi
et al., 2011; Jin et al., 2014)) while some others use
relaxed version ((Schaerf, 2002; Ruiz-Torrubiano and
Suarez, 2010)).
There are not many experimental studies on the
pre-assignment constraint so far, except some infor-
mal discussion (Di Tollo and Roli, 2008).(Di Gaspero
et al., 2011) examined the impact of the pre-
assignment constraint and reported that pre-assigning
one asset tended to worsen the performance except
when the asset is in the optimal solution for all the
expected return levels.
Recently, some scholars included the round-lot
constraint into PO problems, which makes it more
difficult to find a feasible solution. Some of these
were measured in units of money ((Speranza, 1996;
Mansini and Speranza, 1999; Kellerer et al., 2000;
Lin and Liu, 2008; Bonami and Lejeune, 2009; Gol-
makani and Fazel, 2011)), while others imposed that
the continuous weight variables should be an integer
multiple of a given fraction ((Jobst et al., 2001; Stre-
ichert et al., 2004; Skolpadungket et al., 2007)). Some
of them included transaction cost at the same time.
(Mansini and Speranza, 1999) showed that a PO prob-
lem with minimum lot and without any fixed transac-
tion cost is NP-complete.
Class constraint was first introduced by (Chang
et al., 2000), also mentioned in (Ruiz-Torrubiano
and Suarez, 2010), and applied in ((Anagnostopoulos
and Mamanis, 2010; Anagnostopoulos and Mamanis,
2011a; Anagnostopoulos and Mamanis, 2011b)). An-
other form of this constraint is splitting the universe
of assets into subsets with similar features. Optimi-
sation is performed on the best representative of each
class (Vijayalakshmi Pai and Michel, 2009).
Most of the research in the literature adopt two
constraints in the problem formulation. In particu-
lar, cardinality and quantity constraints were stud-
ied 44.93% and 30.43%, respectively, in the portfo-
lio management models (Metaxiotis and Liagkouras,
2012).
4 ANALYSIS ON DIFFERENT
CONSTRAINTS IN THE MV
MODEL
As most of the applications in the literature dealt
with quantity and cardinality constrained PO prob-
lems, in this section we first discuss the effect of
pre-assignment, round-lot and class constraint based
on the quantity and cardinality constrained MV
model in terms of solution quality and computational
cost. Then we present experimental framework us-
ing CPLEX 12.6 to identify solutions for the most
commonly applied constrained MV models in the lit-
erature, some models with different settings in con-
straints.
Five widely tested benchmark datasets in the OR-
library (Beasley, 1990) (http://people.brunel.ac.uk/
mastjjb/jeb/info.html) are chosen in our experi-
ments. They were extracted from the well-known
indices, the Hong Kong HangSeng, the German
DAX100, the UK FTSE100, the US S&P100 and
the Japan Nikkei, with dimension N = 31, 85, 89, 98
and 225 (N31-N225), respectively. The portfo-
lios obtained under different constraints and settings
form constrained efficient frontiers (CEF). The un-
constrained efficient frontier (UEF), for the uncon-
strained PO problem from the OR-library, is used as
the upper bounds for CEF. For each dataset, 50 points
across the UEF are chosen. For each point on the
CEF, the solver is run to minimise risk (QP) given
a respected return. The stopping time limit is set to
3600 seconds.
Several performance measures are adopted, for
which smaller values denote better solutions. The
Average Percentage Error (APE) (Di Gaspero et al.,
2011) measures the relative distance between the ob-
tained CEF from the UEF with the same expected re-
turn, calculated using Formula (11), where x
and x
denote the portfolios on the CEF and UEF, respec-
tively, f
r
i
denotes the value of the obtained risk, and p
is the number of portfolios on the frontier.
APE =
1
p
p
i=1
f
r
i
(x
) f
r
i
(x)
f
r
i
(x)
(11)
Generational distance (GD) (Cura, 2009) refers to
the average minimum distance of each portfolio on
the CEF from the UEF, calculated using Formula (12)
ICORES 2016 - 5th International Conference on Operations Research and Enterprise Systems
390
where d
x
x
is the Euclidean distance. f
v
and f
r
are the
risk (Eq.(1)) and return (Eq.(2)) of the portfolio. It
measures how close the obtained CEF is to the UEF.
GD =
1
p
x
X
CEF
min{d
x
x
|x X
UEF
} (12)
d
x
x
=
q
( f
r
(x
) f
r
(x))
2
+ ( f
v
(x
) f
v
(x))
2
(13)
Inverted generational distance(IGD) is a variant of
GD. It uses UEF as a reference to calculate the av-
erage minimum distance of its each portfolio to the
CEF, calculated using Formula (14). This measure
mainly shows the overall quality of the obtained so-
lution set (i.e. its diversity and convergence to the
UEF).
IGD =
1
p
x
X
UEF
min{d
x
x
|x X
CEF
} (14)
Algorithm Effort (AE) (Chen et al., 2012), AE =
Time
p
, measures the ratio of the total run time to the
number of feasible portfolios obtained on the CEF.
MAX and MIN denote the maximum or minimum
time spent for one portfolio. Optimal Rate is used
to denote the rate of optimal solutions obtained out of
all the points on the CEF.
The experiments are coded in C++ in Microsoft
Visual Studio 2012, and run on a PC with Windows
7 Operating System (64-bit), 6GB of RAM, and an
Intel Core i7 CPU (960@3.2GHz).
4.1 Pre-assignment, Round-lot and
Class Constraints
We first test all the extended constraints on the MV
model. The lower and upper bounds of the weights
are set as ε = 0.01, δ = 1, and cardinality K = 10.
These are widely used in the literature. Preliminary
computational results indicate that the pre-assigned
asset(s) is not a main performance discriminator of
the basic MV model, thus is randomly set as P =
{30}. The class constraints is set to: randomly de-
fine two classes with size of 20% proportionately to
problem dimension (N). These constraints are set
as those in (Lwin et al., 2014). Initial experiment
results showed that cardinality constraint contributes
the most to the problem difficulty.
We then assess the effect on the computational
cost from pre-assignment (C3), round-lot (C4) and
class (C5) constraints based on the MV model with
cardinality (C1) and quantity (C2) constraints, i.e.
ε = 0.01, K = 10.
4.1.1 Pre-assignment Constraint(C3)
To evaluate the impact of pre-assignment constraint
on the computational cost, for each benchmark in-
stance, we fix in turn one of the assets as pre-assigned
(imposing C1 and C2 with the settings as mentioned
above). For each instance we then obtain a group of N
CEFs, one for each pre-assigned asset. Intuitively, the
choice of the pre-assigned assets determines the so-
lution quality measured in APE. For example, if the
pre-assigned asset does not belong to any optimal so-
lution for most of the values of the required return
R
exp
on UEF, the pre-assignment constraint generally
deteriorates the solution quality. Moreover, the mag-
nitude of this worsening may depend on the features
of the asset, e.g. return rate, standard deviation(sd),
and the ratio of return/sd, etc.
As an example, in Figure 1, for instance N98, the
average time spent for computing the CEF for each
pre-assigned asset is plotted. It can be seen that the
time used varies with obvious difference among dif-
ferent assets, ranging from around 20s to 400s. Figure
2 reports the frequency of each asset appearing in the
optimal portfolios on the UEF. For instance, asset 1
which leads to the maximum time used (400s) never
appears in the optimal solution in the unconstrained
problem. Moreover, the rankings of its return rate,
sd, and the ratio of return/sd are quite low, ranked as
61, 64, and 71, respectively. The assets ranked top
in return rate, sd and ratio (namely assets 82, 62, and
89) appear to lead to around only 50s, which is at the
lower level of time used in Figure 1. Moreover, asset
89 appears 36 times out of 50 in the optimal portfo-
lios on the UEF shown in Figure 2. These imply that
there is some correlation between the feature of the
pre-assigned asset and the solution construction.
Intuitively, if we pre-assign more assets in ad-
vance, the problem becomes smaller with only (K
pre assigned) number of assets. We conduct an an-
other experiment on two groups of pre-assigned assets
on instance N98, each with the top five and bottom
five shown in Figure 2. As expected in Table 1, the
computational time with more pre-assigned assets is
much lower than that with fewer assets. Pre-assigning
the assets which intend to appear in the optimal port-
folio on UEF can reduce the computational cost and
lead to better solution quality.
Table 1: Results of pre-assigned assets with different fea-
tures.
Pre-assigned APE MAX AE Optimal Rate
Top 5 0.100718 19.005 4.609 0.98
Bottom 5 0.234882 122.136 22.716 0.96
Constrained Portfolio Optimisation: The State-of-the-Art Markowitz Models
391
96928884807672686460565248444036322824201612841
400
350
300
250
200
150
100
50
0
Index of asset
Time(s)
Figure 1: Average time used for N98, with each pre-
assigned asset.
9693898682767366656462525145444342373634332322201914112
40
30
20
10
0
index of asset
Count No.
23
8
36
28
23
18
11
1
10
15
19
22
25
3
4
34
7
41
38
3
27
4
28
14
6
26
34
Figure 2: Frequency of each in the optimal solution on UEF
in N98, y-axix is the occurrence number.
4.1.2 Round-lot Constraint(C4)
We test on all the instances with three different lot
unit sizes and compare the results against the model
without round-lot constraint. As seen in Table 2, there
are no significant differences in terms of the computa-
tional time and optimal rate. The chosen lot unit size
has a slight effect on the optimal rate.
4.1.3 Class Constraint(C5)
The class constraint is tested with different settings on
the class defined and the lower bound. From the theo-
retical perspective, class constraint is linear thus does
not increase the problem difficulty. The number of
classes should also reduce the problem difficulty as
the original problem is partitioned into a set of sub-
problems with lower dimension. Table 3 shows the
results obtained on instance N98 with different class
settings based on the C1 and C2 constrained models.
The class classification here is randomly assigned.
From Table 3, it can be seen that the higher lower
bound 0.1 could reduce the optimal rate compared
Table 2: Results on the C4 constrained problem based on
C1 and C2 constrained MV model.
Dataset Lot unit APE Max Time AE Optimal Rate
N31
NA 0.02107 0.983 0.138 0.92
0.005 0.02157 0.719 0.269 0.92
0.008 0.03321 9.612 2.675 0.88
0.01 0.02194 0.577 0.108 0.92
N85
NA 0.07539 61.463 11.500 0.94
0.005 0.07686 49.954 5.550 0.94
0.008 0.07686 49.954 5.550 0.94
0.01 0.07838 44.33 5.03887 0.94
N89
NA 0.02569 902.803 101.885 0.94
0.005 0.02625 1586.61 97.140 0.94
0.008 0.02936 400.068 39.773 0.92
0.01 0.02681 1177.83 91.809 0.94
N98
NA 0.06181 3600.59 1055.72 0.7
0.005 0.06265 3600.47 1042.23 0.72
0.008 0.06749 3600.64 979.171 0.76
0.01 0.06345 3600.4 1037.42 0.72
N225
NA 0.01034 16.333 7.229 0.98
0.005 0.01107 12.592 5.284 0.98
0.008 0.01129 23.865 8.139 0.96
0.01 0.01138 9.441 4.146 0.98
to 0.05. As expected, the setting with less classes
requires much more computational effort than those
with more classes. Intuitively expected, in terms of
computational cost, the class constraint does not gen-
erate much difficulty on the cardinality constrained
model, while it can affect the rate of optimality.
Table 3: Results with different class settings based on the
C1 and C2 constrained MV Model.
Class setting APE MAX AE Optimal Rate
NA 0.061814 3600.59 1055.72 0.7
size=6, LC=0.05 0.163718 519.235 101.793 0.84
size=6, LC=0.1 0.246108 999.286 156.248 0.68
size=2, LC=0.1 0.096405 3601.05 1123.34 0.62
size=2, LC=0.05 0.079334 3599.82 433.568 0.9
4.2 Performance of CPLEX for the
Complete Model
Table4 reports our results under different constraint
settings (Columns C1-C5) in terms of optimality, so-
lution quality (GD, IGD, APE), and computational
cost (MAX, MIN, AE) for the constrained models.
This feature is varied by the researchers. In the last
two models, for each instance, the asset which took
the longest time in Figure 1 is pre-assigned. Two
classes are defined in the class constraint with the
lower bound set as 0.05. The results illustrate that,
the last two models spent much less computational
cost compared to the C1 and C2 constrained models.
In other words, combination of all these constraints
seems to have neutralizing effect.
4.3 Performance on Existing MV
Models in the Literature
In addition, to provide the optimal or lower bound
for various subsets of the constrained MV models
ICORES 2016 - 5th International Conference on Operations Research and Enterprise Systems
392
Table 4: Results of the complete MV Model with all five extended constraints.
Setting C1 C2 C3 C4 C5 Data GD IGD APE MAX MIN AE Optimal Rate
r
P
R
exp
K = 10 ε = 0.01
N31 3.25e-05 9.64e-05 0.02107 0.98 0 0.13 0.92
N85 2.87e-05 7.4e-05 0.075388 61.5 0.04 11.49 0.94
N89 9.08e-05 4.45e-05 0.025686 902.8 0.06 101.85 0.94
N98 1.8e-05 5.46e-05 0.061814 3600.5 0.07 1055.72 0.7
N225 3.72e-06 2.48e-05 0.010336 16.3 2.57 7.22 0.98
r
P
= R
exp
K = 10 ε = 0.01
N31 3.24e-005 9.64e-005 0.020935 2.65 0.01 0.36 0.92
N85 2.87e-005 7.40e-005 0.075509 74.5 0.18 13.90 0.94
N89 8.99e-006 4.44e-005 0.025524 577.3 0.34 64.34 0.94
N98 1.80e-005 5.47e-005 0.061832 3600.5 0.06 893.54 0.78
N225 3.71e-006 2.48e-005 0.010293 20.3 2.75 10.55 0.98
r
P
R
exp
K 10 ε = 0.01
N31 8.66e-008 4.94e-005 0.000124 5.94 0.03 0.48 1
N85 3.93e-006 4.74e-005 0.024219 84.1 0.19 11.68 1
N89 4.41e-006 3.24e-005 0.018769 652.2 0.17 109.47 1
N98 7.97e-006 4.83e-005 0.047498 3600.2 0.03 1128.91 0.7
N225 7.35e-007 2.35e-005 0.002157 27.4 0.21 7.37 1
r
P
= R
exp
K 10 ε = 0.01
N31 8.86e-08 4.94e-05 0.000128 7.48 0.01 0.51 1
N85 3.92e-06 4.74e-05 0.024198 89.9 0.07 12.28 1
N89 4.39e-06 3.24e-05 0.018719 609.4 0.07 107.88 1
N98 7.97e-06 4.83e-05 0.0475 3600.5 0.04 1103.17 0.7
N225 7.25e-07 2.35e-05 0.00217 14.1 0.14 6.24 1
r
P
R
exp
K = 10 ε = 0.01
[2]
0.005
C
1
= 1..5
C
2
= 6..10
L
m
= 0.05
N31 4.61e-05 0.000106 0.032018 0.18 0.01 0.05 0.92
[83] N85 5.15e-05 0.000114 0.135537 3.9 0.04 0.92 0.9
[74] N89 1.81e-05 5.84e-05 0.049098 22.2 0.08 4.83 0.92
[1] N98 3.99e-05 8.46e-05 0.108055 248.3 0.05 41.12 0.94
[36] N225 2.16e-05 3.65e-05 0.053238 5.7 1.64 3.51 0.96
r
P
R
exp
K = 10 ε = 0.01
[2]
0.008
C
1
= 1..5
C
2
= 6..10
L
m
= 0.05
N31 5.72e-05 0.000153 0.041071 3.08 0.06 0.44 0.88
[83] N85 4.72e-05 0.000153 0.124248 14.6 0.45 3.55 0.86
[74] N89 2.18e-05 7.13e-05 0.057599 34.2 0.26 7.63 0.9
[1] N98 4.36e-05 0.0001 0.115075 314.9 0.10 54.49 0.92
[36] N225 2.60e-05 4.50e-05 0.063057 6.1 1.01 3.58 0.94
Constrained Portfolio Optimisation: The State-of-the-Art Markowitz Models
393
Table 5: Results for various models in literature.
Models in the literature Data GD IGD APE AE Optimal Rate
(Maringer and Kellerer, 2003) N31 5.82E-08 4.94E-05 8.20E-05 0.01176 1
N85 3.91E-06 4.74E-05 0.024185 0.66952 1
N89 4.40E-06 3.24E-05 0.018769 1.95914 1
N98 7.96E-06 4.83E-05 0.047486 10.6813 1
N225 6.76E-07 2.35E-05 0.002015 1.2937 1
(Chang et al., 2000) N31 3.25E-05 9.64E-05 0.02107 0.137674 0.92
(Fernandez and Gomez, 2007) N85 2.87E-05 7.40E-05 0.075388 11.4999 0.94
(Xu et al., 2010) N89 9.08E-06 4.45E-05 0.025686 101.885 0.94
K = 10 N98 1.80E-05 5.46E-05 0.061814 1055.72 0.7
ε = 0.01 N225 3.72E-06 2.48E-05 0.010336 7.22906 0.98
(Ruiz-Torrubiano and Suarez, 2010) N31 8.86E-08 4.94E-05 0.000128 0.51264 1
(Schaerf, 2002) N85 3.92E-06 4.74E-05 0.024198 12.2765 1
K <= 10 N89 4.39E-06 3.24E-05 0.018719 107.883 1
ε = 0.01 N98 7.97E-06 4.83E-05 0.0475 1103.17 0.7
N225 7.25E-07 2.35E-05 0.00217 6.2358 1
(Lwin et al., 2014) N31 5.38E-05 0.00015 0.036569 1.90675 0.88
K = 10 N85 4.14E-05 0.000108 0.104911 16.5399 0.9
ε = 0.01 N89 2.22E-05 5.93E-05 0.054954 12.4044 0.92
N98 2.68E-05 6.68E-05 0.084467 107.943 0.96
N225 1.74E-05 3.86E-05 0.04161 7.39513 0.94
(Woodside-Oriakhi et al., 2011) N31 2.66E-05 0.000159 NA 0.0766 1
K = 10 N85 1.08E-05 0.000128 NA 9.91606 1
ε = 0.01 N89 6.81E-06 9.35E-05 NA 94.7458 1
R
exp
in range:[0.9R
exp
,1.1R
exp
] N98 1.46E-05 0.00014 NA 1105.7 0.72
N225 2.06E-06 5.27E-05 NA 6.54972 1
in the literature, the most commonly applied settings
in different constraints in relevant works are tested.
Due to the space limited, a sketch of the performance
of the representative models with different combi-
nations of constraints is shown in Table 5. More
detailed data and updated models are published in
http://www.cs.nott.ac.uk/pszrq/benchmarks.htm.
5 CONCLUSION AND FUTURE
WORK
This paper studies the constrained Markowitz MV
model for the portfolio optimisation problem, where
quantity, cardinality, pre-assignment, round-lot and
class constraints are considered. We first discuss the
effect of the pre-assignment, round-lot and class con-
straints based on the cardinality and quantity con-
strained models in terms of computational cost. Then
we conduct experiments using CPLEX to obtain op-
timal or feasible solutions within a limited computa-
tional cost for various models with different constraint
settings. The results for the thoroughly studied con-
strained models are presented.
According to the results, the pre-assignment,
round-lot and class constraints do not make a big dif-
ference to the cost of solving the quantity and cardi-
nality constrained problem. As the mostly considered
constraint in the current literature, the cardinality con-
straint is proved to contribute to mainly the problem
difficulty. In addition, the more specific settings im-
posed on these constraints, the easier the problem is.
The complete constrained MV model with all the con-
straints can neutralize the effect of cardinality con-
straint, and the optimal solutions can be much easier
to obtain.
There are some exceptional points on instance
N98, which still requires a huge amount of compu-
tational effort for the cardinality constrained models.
Nevertheless, due to the fast development of commer-
cial solvers, most of the currently studied constrained
MV models can be efficiently solved to obtain most
of the optimal solutions within a limited time.
The above results motivate our future research to
more challenging PO problem. In the future, we plan
to further study the constrained PO problem based on
other risk measures such as VaR and CVaR which are
favoured by investors in reality. It is also interesting
to consider other constraints such as transaction cost
which occurs for problem with more than one invest-
ment period. We also aim to analyse the performance
on difficult real instances of larger problem size.
REFERENCES
Anagnostopoulos, K. and Mamanis, G. (2010). A portfo-
lio optimization model with three objectives and dis-
crete variables. Computers & Operations Research,
37(7):1285 1297. Algorithmic and Computational
Methods in Retrial Queues.
Anagnostopoulos, K. and Mamanis, G. (2011a). The mean-
variance cardinality constrained portfolio optimiza-
tion problem: An experimental evaluation of five mul-
tiobjective evolutionary algorithms. Expert Systems
with Applications, 38(11):14208 – 14217.
Anagnostopoulos, K. P. and Mamanis, G. (2011b). Multiob-
jective evolutionary algorithms for complex portfolio
optimization problems. Computational Management
Science, 8(3):259–279.
Beasley, J. E. (1990). Or-library: Distributing test prob-
ICORES 2016 - 5th International Conference on Operations Research and Enterprise Systems
394
lems by electronic mail. Journal of the Operational
Research Society, 41:1069–1072.
Bienstock, D. (1995). Computational study of a family of
mixed-integer quadratic programming problems. In
Integer Programming and Combinatorial Optimiza-
tion, volume 920 of Lecture Notes in Computer Sci-
ence, pages 80–94. Springer Berlin Heidelberg.
Bonami, P. and Lejeune, M. A. (2009). An exact solu-
tion approach for portfolio optimization problems un-
der stochastic and integer constraints. Operations Re-
search, 57(3):650–670.
Borchers, B. and Mitchell, J. E. (1997). A computational
comparison of branch and bound and outer approxi-
mation algorithms for 01 mixed integer nonlinear pro-
grams. Computers & Operations Research, 24(8):699
– 701.
Chang, T.-J., Meade, N., Beasley, J., and Sharaiha, Y.
(2000). Heuristics for cardinality constrained portfo-
lio optimisation. Computers & Operations Research,
27(13):1271 – 1302.
Chen, A., Liang, Y.-C., and Liu, C.-C. (2012). An
artificial bee colony algorithm for the cardinality-
constrained portfolio optimization problems. In 2012
IEEE Congress on Evolutionary Computation (CEC),
pages 1–8.
Cura, T. (2009). Particle swarm optimization approach
to portfolio optimization. Nonlinear Analysis: Real
World Applications, 10(4):2396 – 2406.
Di Gaspero, L., Di Tollo, G., Roli, A., and Schaerf,
A. (2011). Hybrid metaheuristics for constrained
portfolio selection problems. Quantitative Finance,
11(10):1473–1487.
Di Tollo, G. and Roli, A. (2008). Metaheuristics for the
portfolio selection problem. International Journal of
Operationas Research, 5(1):13–35.
Fernandez, A. and Gomez, S. (2007). Portfolio selection
using neural networks. Computers & Operations Re-
search, 34(4):1177 – 1191.
Golmakani, H. R. and Fazel, M. (2011). Constrained port-
folio selection using particle swarm optimization. Ex-
pert Systems with Applications, 38(7):8327 – 8335.
Jansen, R. and van Dijk, R. (2002). Optimal benchmark
tracking with small portfolios. The Journal of Portfo-
lio Management, 28(2):33–39.
Jin, Y., Qu, R., and Atkin, J. (2014). A population-based
incremental learning method for constrained portfolio
optimisation. In Symbolic and Numeric Algorithms
for Scientific Computing (SYNASC), 2014 16th Inter-
national Symposium on, pages 212–219.
Jobst, N., Horniman, M., Lucas, C., and Mitra, G. (2001).
Computational aspects of alternative portfolio selec-
tion models in the presence of discrete asset choice
constraints. Quantitative Finance, 1(5):489–501.
Kellerer, H., Mansini, R., and Speranza, M. (2000). Se-
lecting portfolios with fixed costs and minimum trans-
action lots. Annals of Operations Research, 99(1-
4):287–304.
Lin, C.-C. and Liu, Y.-T. (2008). Genetic algorithms for
portfolio selection problems with minimum transac-
tion lots. European Journal of Operational Research,
185(1):393 – 404.
Lwin, K., Qu, R., and Kendall, G. (2014). A learning-
guided multi-objective evolutionary algorithm for
constrained portfolio optimization. Applied Soft Com-
puting, 24:757 – 772.
Mansini, R. and Speranza, M. G. (1999). Heuristic algo-
rithms for the portfolio selection problem with min-
imum transaction lots. European Journal of Opera-
tional Research, 114(2):219 – 233.
Maringer, D. (2008). Heuristic optimization for portfolio
management [application notes]. Computational In-
telligence Magazine, IEEE, 3(4):31 –34.
Maringer, D. and Kellerer, H. (2003). Optimization of
cardinality constrained portfolios with a hybrid local
search algorithm. OR Spectrum, 25(4):481–495.
Markowitz, H. (1952). Portfolio selection. The Journal of
Finance, 7(1):pp. 77–91.
Metaxiotis, K. and Liagkouras, K. (2012). Multiobjective
evolutionary algorithms for portfolio management: A
comprehensive literature review. Expert Systems with
Applications, 39(14):11685 – 11698.
Ruiz-Torrubiano, R. and Suarez, A. (2010). Hybrid ap-
proaches and dimensionality reduction for portfolio
selection with cardinality constraints. Computational
Intelligence Magazine, IEEE, 5(2):92–107.
Schaerf, A. (2002). Local search techniques for constrained
portfolio selection problems. Computational Eco-
nomics, 20:177–190.
Skolpadungket, P., Dahal, K., and Harnpornchai, N. (2007).
Portfolio optimization using multi-objective genetic
algorithms. In Evolutionary Computation, 2007. CEC
2007. IEEE Congress on, pages 516–523.
Speranza, M. G. (1996). A heuristic algorithm for a portfo-
lio optimization model applied to the milan stock mar-
ket. Computers & Operations Research, 23(5):433
441.
Streichert, F., Ulmer, H., and Zell, A. (2004). Evaluating a
hybrid encoding and three crossover operators on the
constrained portfolio selection problem. In Evolution-
ary Computation, 2004. CEC2004. Congress on, vol-
ume 1, pages 932–939.
Vijayalakshmi Pai, G. and Michel, T. (2009). Evolution-
ary optimization of constrained k -means clustered as-
sets for diversification in small portfolios. Evolution-
ary Computation, IEEE Transactions on, 13(5):1030–
1053.
Woodside-Oriakhi, M., Lucas, C., and Beasley, J. (2011).
Heuristic algorithms for the cardinality constrained ef-
ficient frontier. European Journal of Operational Re-
search, 213(3):538 – 550.
Xu, R.-t., Zhang, J., Liu, O., and Huang, R.-Z. (2010).
An estimation of distribution algorithm based portfo-
lio selection approach. In Technologies and Applica-
tions of Artificial Intelligence (TAAI), 2010 Interna-
tional Conference on, pages 305–313.
Constrained Portfolio Optimisation: The State-of-the-Art Markowitz Models
395