Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
3
,
No.
3
,
Ma
rch
201
9
, p
p.
108
7
~
1094
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
3
.pp
108
7
-
109
4
1087
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Modellin
g volatili
ty of
Ku
ala L
um
pu
r comp
osite ind
ex
(KL
CI)
usin
g
SV
and ga
rch mod
els
Ez
at
ul Akm
a
Ab
d
ull
ah
,
Siti
M
eri
am
Z
ahari
,
S
.S
arif
ah
Radiah
Shari
f
f,
Muham
ma
d
Asmu’i
Abdul
Rahim
Cent
re
for
St
at
is
ti
cs
and
De
ci
sio
n
Scie
n
ce Studi
e
s,
Facult
y
of
Co
m
pute
r
&
M
at
he
m
at
ic
a
l
Sci
ences,
Univer
sit
i
T
ekn
ologi
MA
RA Shah
Alam,
Mal
a
ysia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Sep
1,
2018
Re
vised N
ov 28
, 2
018
Accepte
d Dec
8
, 2
018
It
is
wel
l
-
known
th
at
fina
n
cial
tim
e
serie
s
exh
ibi
t
s
cha
ng
ing
v
aria
nce
and
th
is
ca
n
have
impo
rta
nt
conse
qu
en
ce
s
in
form
ulat
ing
e
conomic
or
fin
anc
i
al
dec
isions.
In
m
uch
re
ce
nt
ev
ide
n
ce
show
s
that
vo
la
tilit
y
of
f
ina
nc
i
al
assets
is
not
consta
n
t,
bu
t
rat
h
er
th
at
re
l
at
iv
ely
vol
at
i
le
per
iods
alter
n
at
e
with
m
ore
tra
nquil
on
es.
T
hus,
th
ere
are
m
an
y
oppor
tuni
t
ies
to
obt
ai
n
fore
c
asts
of
th
is
ti
m
e
-
var
y
ing
r
isk.
The
pape
r
pr
ese
nts
the
m
ode
ll
ing
vol
atilit
y
o
f
the
Kual
a
Lumpur
Com
po
site
Inde
x
(KLC
I)
using
SV
and
GA
RCH
m
odel
s.
Thus,
th
e
ai
m
of
thi
s
stud
y
is
to
m
odel
t
he
KLCI
stock
m
ark
et
using
t
wo
m
odel
s;
Stocha
stic
Vol
a
ti
lit
y
(SV
)
and
Gene
r
alize
d
A
uto
-
Regre
ss
ive
Condit
ional
Hete
rosce
d
astici
t
y
(GA
RCH).
Th
is
stud
y
emplo
y
s
an
SV
m
odel
wit
h
Ba
y
esia
n
appr
oac
h
and
Markov
Ch
ai
n
Mo
n
te
Carl
o
(MC
MC)
sam
ple
r;
a
nd
GA
RCH
m
odel
with M
L
E esti
m
at
o
r.
The
best m
odel
wi
ll
be us
ed to
fore
cas
t the
fu
tur
e
vola
tilit
y
of s
to
c
k
ret
urns.
Th
e s
t
ud
y
invol
v
es
97
1
dail
y
observ
at
i
ons
of K
LCI
Closing
pri
ce
in
dex,
from
2
Jan
uar
y
2008
to
10
Novem
ber
201
6,
exc
lud
ing
publi
c
holi
d
a
y
s
.
SV
m
odel
is
fou
nd
to
b
e
th
e
b
est
base
d
on
th
e
lo
west
RMS
E
and
MA
E
v
al
ues
.
Ke
yw
or
d
s
:
Fo
r
ecast
GA
RC
H
KLCI
Ma
rkov chai
n m
on
te
carlo
Stoch
a
sti
v vo
la
ti
lity
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Sit
i M
eria
m
Zah
ari
,
Cent
re
for
St
at
is
ti
c
s
and
De
ci
sio
n
Scie
n
ce Studi
e
s,
Facul
t
y
of
Com
pute
r & Ma
the
m
a
ti
c
al
Sc
ie
n
ce
s,
Univer
siti
Te
kno
logi
MA
RA Sha
h
Alam,
Ma
lay
si
a
Em
a
il
:
m
ari
am@
tms
k.
uit
m
.
edu
.
m
y
1.
INTROD
U
CTION
Ma
ny
fina
ncial
tim
e
series
exh
ibit
s
cha
ngin
g
va
riance
a
nd
this
can
ha
ve
i
m
po
rtant
c
on
s
equ
e
nces
i
n
form
ulati
ng
econom
ic
or
fina
ncial
decisi
ons
.
In
m
uch
re
ce
nt
evide
nce
s
hows
t
hat
vola
ti
li
ty
of
finan
ci
a
l
asset
s
is
not
co
ns
ta
nt,
bu
t
rathe
r
t
hat
r
el
at
ively
vo
la
ti
le
per
i
ods
al
te
r
nate
wit
h
m
or
e
tran
qu
il
on
e
s.
Th
us
,
t
her
e
are
m
any
opport
un
it
ie
s
to
ob
ta
in
f
or
ec
ast
s
of
this
ti
m
e
-
va
ryi
ng
ris
k.
Mod
el
i
ng
of
f
inancial
ti
m
e
s
eries
ha
s
f
oc
use
d
on
est
i
m
ating
the
tim
e
-
var
yi
ng
vo
la
ti
li
ty
.
The
la
tt
er
key
is
for
m
easur
in
g
risk,
pri
ci
ng
asset
der
i
vativ
es,
an
d
hedgin
g
strat
e
gies
[1]
.
Wh
e
n
we
ta
lk
a
bout
vola
ti
li
ty
,
it
is
act
ually
a
sta
t
ist
ic
al
m
easur
e
of
t
he
disp
e
rsi
on
of
returns
f
or
a
g
iven
sec
ur
it
y or
m
ark
et
in
dex. V
olati
li
ty
can eit
her
b
e
m
eas
ur
e
d
by
us
i
ng
t
he
sta
ndar
d
de
viati
on
or
var
ia
nce
bet
ween
retu
rn
s
f
ro
m
the
sam
e
secur
it
y
or
m
ark
et
in
dex.
Co
m
m
on
ly
,
ris
kier
se
cu
rity
has
higher
vo
la
ti
li
ty
.
There
are
m
any
m
od
el
s
that
ca
n
be
us
e
d
in
fina
nc
ia
l
tim
e
series
data.
T
hose
m
od
el
s
ca
n
be
si
m
pl
e
m
od
el
su
c
h
a
s
ra
ndom
wal
k,
sm
oo
thin
g
m
od
el
s,
sim
pl
e
re
gr
e
ssio
n,
m
ov
ing
a
ve
ra
ge
a
nd
e
xpon
entia
l
sm
oo
thing,
or
com
plex
m
od
el
su
c
h
as
A
utoreg
ressive
Co
ndit
ion
al
Heter
osc
edasti
ci
ty
(AR
CH),
Ge
ner
al
iz
ed
Au
t
or
e
gr
e
ssive
Condit
ion
al
H
et
ero
sce
dastic
it
y
(G
ARC
H)
a
nd
Stoc
hastic
Vo
la
ti
li
ty
(S
V)
m
od
el
s
.
O
ne
of
t
he
obvious
intere
sts
in
em
pirical
fina
nce
is
f
oreca
sti
ng
f
utur
e
retu
r
ns
of
as
set
s
su
c
h
a
s
st
ocks
a
nd
cu
rr
e
ncies’
exch
a
nge
rates
.
If
on
e
wer
e
a
ble
to
f
or
ecast
tom
or
row’
s
return
with
s
om
e
degree
of
preci
sion,
one
co
ul
d
use
this
inf
or
m
at
ion
in
a
n
inv
e
stm
ent
dec
isi
on
tod
ay
.
It
is
tr
ue
that
it
is
no
t
easy
to
gen
e
r
at
e
a
ver
y
accurate
pr
e
dicti
on
f
or
asset
retu
rn
s
,
s
ince
we
a
re
a
war
e
t
hat
with
al
l
fo
reca
sti
ng
m
et
ho
ds,
s
ucc
ess
is
not
guar
anteed
,
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
108
7
–
109
4
1088
bu
t
we
ca
n
at
le
ast
forecast
an
d
get
a
sig
n
of
to
m
or
ro
w
’s
return.
Howe
ver
,
on
e
is
no
t
on
ly
int
erested
i
n
obta
ini
ng
accurate
foreca
sts
of
retu
rn
s
on
fi
nan
ci
al
ass
et
s,
but
al
so
in
forecast
s
in
as
so
ci
at
ed
vo
la
ti
li
ty
.
This
is
bec
ause;
vo
la
ti
li
ty
is o
ft
en rega
rd
e
d
a
s
a m
easur
e of th
e risk o
f ret
urn
.
A
pro
blem
in
t
he
a
naly
sis
of
fina
ncial
tim
e
series
data
is
t
o
forecast
the
vo
la
ti
li
ty
of
fut
ur
e
retu
rn
s
since
it
can
re
f
le
ct
the
risk
of
the
retu
rns.
T
he
first
at
tem
pt
to
so
l
ve
this
pr
ob
le
m
is
us
ing
the
Bl
ack
-
Sc
ho
le
s
m
od
el
.
This
m
od
el
wa
s
first
intr
oduce
d
by
Schole
s
a
nd
Bl
ack
i
n
19
73.
T
he
a
dva
ntages
of
the
Bl
ack
-
S
cho
le
s
m
od
el
hav
e
be
en
well
known
for
ye
a
rs
a
nd
t
his
m
od
el
is
sim
ple
to
i
m
plem
ent.
Acc
ordi
ng
to
the
Bl
ack
Sc
ho
le
s
form
ula,
the
m
od
el
ass
um
es
t
hat
c
on
ti
nu
ously
com
po
un
ded
stock
ret
urns
ar
e
norm
al
ly
distrib
uted
with
co
ns
ta
nt
m
ean
an
d
var
i
ance.
This
m
ea
ns
t
hat
m
easur
e
of
how
m
uch
a
stoc
k
can
be
ex
pected
to
m
ov
e
is
c
onsta
nt
ove
r
tim
e.
Howev
e
r
,
in
the
real
fin
ancial
w
or
l
d,
vola
ti
li
ty
of
fi
na
ncial
data
is
dynam
ic
.
The
ass
um
ption
of
co
nst
ant
vo
la
ti
li
ty
is
very
restrict
ive
si
nce
a
si
gn
i
f
ic
ant
num
ber
of
e
m
pirical
stud
ie
s
show
that
vo
l
at
il
ity
in
asset
pr
ic
es
is
tim
e
-
var
yi
ng.
T
her
e
fore,
Bl
ack
-
Sc
ho
le
s
is
not
s
uitabl
e
to
us
e.
T
o
overc
om
e
this
pro
blem
,
tim
e
va
ryi
ng
vo
la
ti
li
ty
m
od
el
nam
ed
Au
t
oreg
ressive
Co
ndit
ion
al
ly
Hete
ro
sce
dastic
(
A
RC
H)
m
od
el
was
e
xpresse
d.
The
n,
the
m
od
el
was
exte
nded
to
G
ener
al
iz
ed
Au
t
o
-
Re
gr
es
sive
Condit
ion
al
H
et
ero
sce
dastic
it
y
(GARC
H
)
m
od
el
s.
GA
RC
H
m
od
el
is
the
m
os
t
appr
opriat
e
m
od
el
to
us
e
in
m
easur
ing
t
he
vola
ti
li
ty
[
2].
This
m
od
e
l
can
be
est
i
m
at
ed
us
ing
m
axi
m
u
m
lik
el
ihoo
d
est
im
at
or
(ML
E)
or
oth
e
r
r
obus
t
es
tim
a
tors.
How
ever,
a
basic
G
ARCH
m
od
el
can
on
l
y
al
low
for
a
sing
le
e
rror
pr
ocess
w
hile
Stoch
a
sti
c
V
olati
li
ty
(S
V)
m
odel
assum
es
two
e
rro
r
processes
.
Thi
s
sh
ows
t
hat,
SV
will
pro
vide
a
bette
r
in
-
sa
m
ple
fit
and
it
is
belie
ved
to
be
a
bette
r
m
od
el
t
o
us
e
[3
]
.
Fina
nc
ia
l
econom
et
ri
cs
li
te
ratur
es
gi
ve
lots
of
at
te
ntion
in
t
he
St
och
a
sti
c
vola
ti
li
ty
(S
V)
m
odel
s
[4
]
rev
ie
wed
t
he
univa
riat
e
and
m
ulti
var
ia
te
of
St
och
a
sti
c
Vo
la
ti
li
ty
(S
V)
m
od
el
and
sta
te
d
t
ha
t
SV
m
od
el
pro
vid
es
m
or
e
flexibili
ty
in
desc
ribi
ng
sty
li
zed
fact
s.
O
ne
of
the
m
od
el
s
that
can
be
use
d
t
o
de
scribe
t
he
sto
chasti
c
dynam
ic
is
def
init
el
y
the
Hest
on
Stoc
hastic
Vo
la
ti
li
ty
m
od
el
.
Sim
il
arly
,
[5
]
al
so
a
dopte
d
the
Hesto
n
St
oc
hastic
Vo
la
ti
li
ty
m
od
el
to
m
od
el
the
i
m
plied
vo
la
ti
lity
ind
ex
in
con
ti
nu
ous
tim
e
s
et
ti
ng
an
d
pri
ce
opti
ons
on
it
[6
]
sh
owe
d t
hat t
he
b
a
sic
S
V
m
od
el
giv
es
a
m
or
e acc
ur
at
e
r
es
ul
t of
pr
ic
es
fo
r Eur
opean
call
op
ti
ons
on c
urr
encies
com
p
ared
t
o
th
e
Bl
ack
–
Sc
hol
es
m
od
el
[7]
,
a
pp
li
ed
stoc
hast
ic
vo
la
ti
li
ty
facto
rs
i
n
the
m
odel
and
f
ound
th
at
it
pro
v
ides
good
resu
lt
f
or
short
te
rm
[8]
,
pr
op
ose
d
a
sim
ple
extensio
n
of
a
sta
nd
a
r
d
G
ARCH
(1,1)
m
od
el
,
w
hich
can
ca
ptu
re
S
V
-
li
ke
pro
per
ti
es
of
data
that
cal
le
d
Stoc
ha
sti
c
GA
RC
H
(
SGARC
H).
It
sh
ows
t
hat
SGARC
H
m
od
el
is
bette
r
to
us
e
si
nce
it
c
aptu
res
t
he
ty
pi
cal
SV
m
od
el
pro
per
ty
w
hich
has
t
wo
er
r
o
rs
,
bu
t
with
on
ly
si
ng
le
par
am
et
ers
tha
t
m
akes
it
eas
y
to
est
im
a
te
[9
]
,
f
ound
t
hat
SV
m
od
el
outper
form
ed
the
GA
RC
H
m
od
e
l.
This
fin
dings
is
al
s
o
suppo
rted
by
[
10
]
.
Stoc
hastic
V
olati
li
t
y
(S
V
)
m
od
el
is
f
ound
to
be
t
he
best
m
od
el
to
predi
ct
the
NZ
st
ock m
ark
et
, [11
]
. Othe
r a
dv
a
ntage
s
of
Stoch
a
sti
c Vola
ti
li
t
y (SV
)
m
od
el
can
b
e
fo
und
in
[1
2].
Nev
e
rtheless
,
t
he
est
im
ation
of
SV
m
od
el
s
is
the
m
ai
n
chall
eng
e
in
ap
pl
yi
ng
the
m
od
el
.
De
pe
nd
i
ng
on
t
he
m
od
el
,
s
om
e
m
o
m
ents
m
ay
or
m
ay
be
unknow
n
as
in
cl
os
ed
f
or
m
resu
lt
ing
t
he
un
known
of
t
he
tra
nsi
ti
on
densi
ty
of
the
s
ta
te
vector
[
13]
.
H
oweve
r,
this
issue
can
be
s
ol
ved
by
perf
orm
ing
the
ef
fici
ent
a
nd
fast
Ba
ye
sia
n
Ma
rkov
c
hain
Mon
te
Ca
rlo
(
MC
MC
)
est
im
at
ion
al
gorith
m
[1
4].
The
re
f
or
e
,
t
he
ai
m
of
this
stu
dy
is
t
o
m
od
el
and
c
om
par
e
be
tween
both
th
e
S
V
a
nd
GAR
CH
m
od
el
s
by
a
pp
ly
in
g
t
he
MC
MC
m
e
t
hod
in
t
he
e
sti
m
at
ion
process
for SV
m
od
el
. Th
e
b
e
st m
od
el
is u
se
d for
forecast
in
g pur
po
se
.
2.
RESEA
R
CH MET
HO
D
The
data
of
thi
s
stu
dy
c
onsist
s
of
97
1
daily
cl
os
in
g
i
nd
e
xe
s
f
r
om
Ku
al
a
Lum
pu
r
Com
po
sit
e
I
nde
x
(K
LC
I),
from
2nd
Ja
nu
a
ry
2008
to
10th
N
ov
em
ber
2016.
The
data
is
e
xt
racted
f
ro
m
Tho
m
so
n
DataS
tream
.
Ku
al
a
L
um
pu
r
com
po
sit
e
in
de
x
or
KLCI
is
t
he
m
ai
n i
nd
e
x a
nd
m
ark
et
in
dicat
or
i
n M
al
ay
sia
. T
his
in
de
x
is
the
represe
ntati
ve
of
Ma
la
ysi
a’s
s
toc
k
m
ark
et
.
K
LCI
is
now
en
han
ce
d
an
d
kn
own
a
s
F
TSE
Burs
a
Ma
la
ysi
a
KLC
I.
The
FTSE
B
ursa
Ma
la
ysi
a
K
LCI
was
i
ntr
oduce
d
on
4t
h
April
19
86
as
the
Ku
al
a
Lu
m
pu
r
Com
po
si
te
Inde
x
(K
LC
I).
T
his
ind
e
x
is
der
i
ved
from
10
0
com
pan
ie
s
.
T
he
co
m
pan
ie
s
hav
e
been
ch
os
e
n
by
Burs
a
Ma
la
ysi
a
from
a cr
os
s secti
on
of the t
otal l
ist
ed
c
om
pan
ie
s in
Ma
la
ysi
a.
2.1.
St
oc
hastic
V
ola
tili
ty (SV
) M
od
el
Fo
r
the
SV
m
od
el
,
Ba
ye
sia
n
par
am
et
er
est
im
at
ion
via
Ma
rko
v
Chai
n
M
on
te
Ca
rlo
(M
CM
C)
m
e
tho
d
is
us
e
d.
Let
=
(
1
,
2
,
…
,
)
be
a
vect
or
of
KL
CI
returns
.
The
feat
ur
e
of
S
V
m
od
el
is
that
e
ach
obser
vatio
n
of
is
ass
um
e
d
t
o
ha
ve
it
s
“own”
co
ntem
pora
neous
va
ri
ance
of
ℎ
.
T
he
r
efore,
it
will
r
el
ax
the
usual
assum
ption
of
ho
m
os
cedasti
c
it
y.
The
va
rian
ce
is
not
al
lo
w
ed t
o
va
ry
unr
e
stric
te
dly
over
tim
e
in
orde
r
to
m
ake
the
est
im
a
ti
on
of
a
m
od
el
is
fe
asi
ble.
T
he
lo
ga
rithm
of
the
m
od
el
is
a
ssu
m
ed
to
f
ollow
an
a
utoreg
ressive
proces
s
(A
R)
of
orde
r on
e
.
The SV
m
od
el
can be e
xpress
ed
in
the
f
or
m
(1)
.
=
exp
(
ℎ
2
)
(1)
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Mo
delli
ng
v
olat
il
i
ty
o
f ku
ala l
ump
ur
c
omp
osi
te
ind
ex (
KLC
I)
…
(
Siti
Meri
am Z
ahar
i
)
1089
ℎ
=
+
ℎ
−
1
+
(2)
The
ab
ove
e
qu
at
ion
is
the
e
quat
ion
for
S
V
m
od
el
.
S
V
m
od
el
is
sim
ply
interest
ed
in
m
ark
et
ne
ws
an
d
in
m
od
el
ing
th
ei
r
ef
fects
on
vola
ti
li
ty
.
Fr
om
the
eq
uatio
n,
it
sh
ows
that
the
vo
la
ti
li
ty
in
SV
m
od
el
is
not
af
fecte
d
by
past
retu
rn
s
.
S
V
i
s
s
uppos
ed
t
o
desc
ribe
fina
ncial
tim
e
series
bette
r
since
it
e
ssentia
l
ly
involves
tw
o
e
rro
r
processes
,
w
hich
a
re
a
nd
.
Th
e
ad
de
d
noise
process
in
the
equ
at
io
n
m
akes
the
m
od
el
m
or
e
fle
xib
le
.
ℎ
in
(1)
a
nd
(
2)
is
e
qu
al
t
o
l
og
vol
at
il
ity
of
KLCI
at
tim
e
t.
The
par
am
et
er
vect
or
is
=
(
,
,
)
.
are
the
er
ror
proce
sses.
T
he
pa
ra
m
et
er
involve
d
le
vel
of
l
og
-
va
riance,
,
,
t
he
pe
rsiste
nce
of
lo
g
-
va
riance
ϕ,
an
d
the
vo
la
ti
li
ty
of
lo
g
-
var
ia
nce
.
T
he
value
of
ϕ,
(
-
1<ϕ<
1)
,
m
ea
su
res
the
a
utoc
orrelat
ion
pre
sent
in
the
lo
gge
d
sq
ua
re
d
data.
Th
us
ϕ
ca
n
be
inter
pr
et
e
d
a
s
the
per
sist
en
ce
in
the
vola
ti
lity
.
High
ϕ
ind
ic
at
in
g
vola
ti
li
t
y
cl
us
te
rin
g.
ϕ
is
assum
ed
to
f
ol
low
a
sta
ti
onar
y
process
(|ϕ|<
1).
Pa
ram
et
er
in
the
eq
uatio
n
is
the
vola
ti
li
t
y
of
the lo
g
-
vola
ti
lity
. Log
-
var
ia
nc
e proces
s in
the e
qu
at
io
n
is
ℎ
=
(
ℎ
0
,
ℎ
1
,
…
,
ℎ
)
.
. I
t i
s
unobse
rved.
2.2.
Bay
esi
an
App
roa
c
h: Pri
or Distri
bu
tio
n
A
pri
or
distri
buti
on
nee
ds
to
be
sp
e
ci
fied
fir
st
to
com
plete
the
est
i
m
at
ion
set
up
for
S
V
m
od
el
.
Each
par
am
et
er
is
as
su
m
ed
to
be
pr
ior
i
ndepe
nden
t.
Th
e
Ba
ye
sia
n
sta
ti
sti
cs
is
a
m
at
he
m
a
ti
cal
process
that
a
ppli
es
pro
bab
il
it
ie
s
to
sta
ti
sti
ca
l
pr
oble
m
s
by
pr
ovidin
g
the
to
ols
to
update
the
be
li
efs
in
the
ev
idence
of
new
data
.
The
m
os
t
i
m
portant
of
Ba
ye
si
an
is
t
he
est
a
bl
ishm
ent
of
para
m
et
ers
and
m
od
el
s
.
T
he
pr
i
or
belie
f
distri
buti
on
is
us
e
d
to
represe
nt
the
str
en
gths
on
belie
fs
a
bout
t
he
pa
ram
e
te
rs
base
d
on
previ
ou
s
ex
per
i
ence.
H
ow
e
ver,
ther
e
m
igh
t
be
no
previ
ou
s
e
xperi
ence.
T
her
e
for
e,
m
at
he
m
at
ici
ans
ha
ve
f
or
m
ulate
d
m
et
ho
ds
to
ov
e
rc
om
e
this
pro
blem
k
now
n
as
unin
form
a
ti
ve
pri
ors. T
hi
s stu
dy u
s
ed
pr
ior param
et
ers fr
om
the p
ast
s
tud
y.
2.3.
Ma
r
kov
Chai
n Mon
te
C
arlo Met
hods
(MCMC)
Stoch
a
sti
c
V
ol
at
il
ity
(S
V
)
m
od
el
can
be
de
scribe
d
i
n
t
hr
e
e
co
ndit
ion
al
di
stribu
ti
ons
t
ha
t
is
f
(
θ)
,y|
h,
and h
|θ. Th
e
f
unct
ion of
f
(θ
)
i
s
the
p
rio
r
distr
ibu
ti
on of
θ
. T
he
joint distri
buti
on o
f
(
θ,
h)
is
de
rive
d
f
ro
m
Ba
ye
s
theo
rem
.
Therefo
re,
f(
θ
,h
|y
)
i
s
pro
portio
nal
to
the
pro
duct
of
th
e
f(
y│
h)f(h│θ)
f(
θ
).
T
o
c
on
st
ru
ct
the
M
arko
v
chain
with
in
va
riant
di
stri
bu
t
ion
,
f(θ,
h|y),
a
lgorit
hm
called
Me
tro
poli
s
-
Hash
ti
ng
s
al
gorithm
is
pr
opose
d
as
il
us
trat
ed
i
n
Fi
gure
1.
Figure
1
.
Proce
ss f
lo
w dia
gr
a
m
f
or
Met
rop
ol
is
-
Hasti
ng M
CM
C
The
process
of
this
m
et
ho
d
ge
ner
at
es
t
he
pro
po
s
al
point
by
us
in
g
t
he
la
st
s
a
m
ple
an
d
a
dd
i
ng
a
rand
om
no
ise
w
hich
ge
ner
at
e
d
f
r
om
a
rand
om
no
ise
di
stribu
ti
on.
Ne
xt,
the
hei
gh
t
of
the
poste
ri
or
par
am
et
er
of
th
e
ne
w
pro
po
sal
is
the
n
com
par
e
d
wi
th
the
hei
gh
t
of
pa
ram
et
ers
of
the
rece
nt
pr
opos
al
a
nd
cal
culat
ed
as
R,
wh
ic
h
represe
nt
the ra
ti
o.
I
f
the
ne
w
pro
posal
has
hi
gh
e
r
poste
rior than
t
he
new
prop
os
al
,
the
n
t
he
new
pro
pos
al
will
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
108
7
–
109
4
1090
be
a
ccepte
d.
O
ther
wise,
MC
MC
will
ra
ndom
ly
acce
pt
or
rej
ect
e
d
t
he
ne
w
pro
posal
.
I
f
the
new
pro
posal
i
s
acce
pted
,
the
n
it
beco
m
es
the
nex
t
sam
ple
of
MC
MC
chain
,
+
1
.
If
the
new
pr
opos
al
is
re
j
ect
ed,
the
n
it
will
ta
ke
the
m
os
t
r
ecent
pro
posal
as
the
ne
w
sam
ple.
T
his
ste
p
will
con
ti
nue
a
nd
sta
rt
agai
n
a
s
the
fir
st
ste
p.
Wh
e
n
the sam
ple,
is
enou
gh, th
e
it
erati
on w
il
l st
op
.
MC
MC
is
on
e
good
ap
proac
h
to
est
i
m
at
e
the
Ba
ye
sia
n.
MC
MC
sh
ows
ho
w
the
poste
rior
par
am
et
er
is
aff
ect
ed
by
t
he
distrib
utio
n.
MC
MC
is
no
t
r
epr
ese
ntin
g t
he
certai
nty
but
just
the
pro
bab
i
li
ty
. Note
that,
hav
i
ng
m
or
e
inf
or
m
at
i
on
a
bout
the
sa
m
ple
will
help
to
m
ake
a
b
et
te
r
plan
f
or
the
f
uture.
A
naly
sis
is
ca
rr
ie
d
out usi
ng
Stochv
ol p
ac
ka
ge.
2.4.
Ga
rch
Model
Gen
e
rali
zed
A
utoreg
ressive
Condit
ion
al
H
et
ero
sce
dastic
it
y
(GARC
H
)
m
od
el
is
the
e
xtensi
on
of
th
e
ARCH m
od
el
. T
he
m
od
el
f
or
GA
RC
H
is:
=
(3)
2
=
+
−
1
2
+
−
1
2
(4)
Fr
om
(3),
or
t
he
retu
rn
s
of
K
L
CI
is
infl
uen
ce
d
by
sta
nd
a
r
d
de
viati
on
an
d
er
ror
.
(
4)
s
hows
tha
t
the
vola
ti
li
ty
,
2
is
determ
ined
by
s
qu
a
red
pas
t
retu
rn
s
,
−
1
2
and
s
qu
a
re
d
past
vola
ti
lity
,
−
1
2
.
The
GA
RC
H
m
od
el
captur
es
vola
ti
li
t
y
bette
r t
ha
n t
he
ARC
H m
od
el
. T
he
ste
ps
in
m
od
el
i
ng
the
vo
la
ti
li
ty
us
in
g
GA
RC
H are
as foll
ows:
3.
RES
ULTS
A
N
D A
NA
L
Y
SIS
Fr
om
the
gr
a
ph,
it
is
cl
ear
tha
t
the
grow
t
h
of
the
eco
no
m
y
has
slow
e
d
do
w
n
after
2008.
D
rop
in
st
ock
pr
ic
e
durin
g
20
08
-
2009 not on
ly
aff
ect
o
the
r c
ountries,
but al
so
Mal
ay
sia
n
stock
m
ark
et
b
ecau
se of the
global
fina
ncial
crisi
s.
The
gl
ob
al
finan
ci
al
c
risis
giv
es
a
n
im
pact
towards
fina
ncial
,
tra
de
an
d
real
ec
onom
y
[15]
.
Ther
e
f
or
e,
it
si
gn
i
ficantl
y
af
f
ect
s
the
Ma
la
ysi
a’s
st
ock
m
ark
et
,
KLC
I.
Fig
ur
e
2
s
how
s
t
he
cl
os
i
ng
pr
ic
e
in
dex
that i
s not stat
ion
a
ry.
Figure
2. Dail
y Pl
ot
of Kual
a
Lum
pu
r
C
om
po
sit
e I
nd
e
x (
K
LCI)
Since
the
in
de
x
grap
h
as
s
ho
wn
i
n
Fig
ur
e
2
is
no
t
sta
ti
on
a
r
y,
therefo
re
the
log
ret
urn
is
use
d.
Fi
gure
3
pr
ese
nts
t
he
K
LCI
ret
urns
fro
m
2
nd
Janu
ary
2008
un
ti
l
10
th
Novem
ber
2016.
The
x
-
axis
r
epr
ese
nted
ti
m
e
w
hile
the y
-
a
xis is lo
g retur
ns.
Figure
3
s
how
s
that
the
gra
ph
is
sta
ti
on
a
ry.
L
og
ret
urns
re
pr
ese
nt
t
he
vol
at
il
ity.
Ba
sed
on
t
he
grap
h,
vo
la
ti
li
ty
chan
ges
ove
r
tim
e,
i
nd
ee
d
the
re
exi
sts
vo
la
ti
li
ty
clu
ste
rin
g
because
there
are
se
ve
ral
la
rg
e
po
i
nts.
Th
e
plo
t
al
s
o
s
ho
ws
t
hat
ne
gative
returns
occ
ur
d
ur
i
ng
the
ob
s
er
ved
pe
rio
d.
The
la
r
gest
ne
gative
retu
rn
s
ar
e
ob
s
er
ved in t
he
yea
r 200
8.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Mo
delli
ng
v
olat
il
i
ty
o
f ku
ala l
ump
ur
c
omp
osi
te
ind
ex (
KLC
I)
…
(
Siti
Meri
am Z
ahar
i
)
1091
Figure
3
.
D
em
eaned L
og Ret
urns of K
LCI
3.1.
Simul
ati
on usin
g MCM
C
Si
m
ulati
on
us
i
ng
Ma
rko
v
C
ha
in
M
on
te
Ca
r
lo
(MCM
C)
is
car
ried
ou
t.
S
ince
MC
MC
i
s
c
om
pu
te
r
-
base
d
c
al
culat
ion,
the
refo
re
the
a
naly
sis
is
done
by
MC
MC
us
in
g
t
he
al
gorithm
,
discu
ss
ed
in
sub
sect
i
on
2.
3.
As
sta
te
d
be
fore,
the
S
V
m
od
el
in
vo
l
ved
pa
ram
et
er
=
(
,
,
)
.
Sin
ce
the
val
ue
f
or
eac
h
par
am
et
er
is
unknow
n
a
nd
m
us
t
be
s
pecif
ie
d
first
be
for
e
r
unni
n
g
t
he
si
m
ulati
on
,
t
hi
s
stu
dy
re
ferre
d
t
o
a
past
stu
dy
in
identify
in
g
the
par
am
et
er
values.
T
his
stu
dy
fo
ll
ow
e
d
[12
]
wh
o
is
the
a
uthor
of
t
he
stoch
vol
pac
kage.
Th
e
auth
or
us
e
d
=
−
9
,
=
0
.
99
, an
d
=
0
.
1
to ru
n
the
si
m
ulati
on
. Ho
wev
e
r,
with les
s infor
m
at
io
n
about t
he
pr
i
or
pa
ram
et
e
rs,
it
will
ca
us
e
the
poste
ri
or
pa
ram
et
er
bein
g
fa
r
from
the
tr
ue
value
s.
T
he
le
ng
th
of
sim
ulate
d
tim
e
series
is
1000.
A
fter
run
ni
ng
t
he
f
unct
io
n
with
the
pri
or
param
et
er,
resu
lt
will
draw
t
he
init
ia
l
log
vo
l
at
il
ity,
ℎ
0
from
the
s
ta
tio
na
ry
distri
bu
t
ion
of
t
he
AR
(1)
pr
ocess.
Th
en,
us
in
g
ℎ
0
valu
e,
the
value
for
ℎ
1
,
ℎ
2
,
…
,
ℎ
is
gen
e
rated
it
era
ti
vely
.
MC
MC
cal
culat
e
the
new
pr
opos
al
accor
ding
t
o
th
e
m
os
t
recent
sam
ple.
Last
ly,
us
i
ng
norm
al
distrib
ution
with
m
ean
0
an
d
sta
ndar
d
de
viati
on,
exp
(
ℎ
2
)
,
the
lo
g
-
retu
rn
s
are
sim
ulate
d.
From
the
si
m
ulate
d co
nd
it
ion
al
vo
la
ti
li
t
y,
the
in
it
ia
l
va
lue
will
be
generate
d.
F
ro
m
th
e
res
ult
of
sim
ulati
on
,
t
he
val
ue
f
or
init
ia
l
log
-
vo
la
ti
lity
,
ℎ
0
=
0
.
0
0
9
1
0
7
516
.
Using
t
his
init
ia
l
valu
e,
MC
MC
will
it
erate
the
next
value
unti
l
the
1000
th
val
ue.
T
he a
lg
ori
thm
us
ed i
n M
CM
C
is
cal
le
d M
et
ropo
li
s
al
gorithm
. T
he a
lg
or
it
hm
sta
rts
with
possibl
e
init
ia
l
value,
i
n
this
case
ℎ
0
=
0
.
0
0
9
107
5
16
.
Ne
xt,
it
ge
ner
at
e
d
ne
w
pr
opos
al
by
us
i
ng
the
la
st
sam
ple
and
add
i
ng
with s
om
e n
oise. T
he result al
so
giv
e
s the d
esc
riptiv
e stat
ist
ic
s o
f
si
m
ulate
d
condi
ti
on
al
volat
il
it
y.
Th
e
m
ini
m
u
m
valu
e
f
or
sim
ulate
d
co
ndit
ion
al
vo
la
ti
lity
is
0.083
57
,
first
qua
dr
a
nt
e
qu
al
to
0.4
1850
,
m
edian
value
is
0.5
6650
,
0.6
4760
f
or
the
m
ean
an
d
0.8
3020
a
nd
2.1
4000
for
t
hir
d
quad
r
ant
a
nd
m
axi
m
um
value.
Re
s
ults
f
or
si
m
ulate
d
co
ndit
ion
al
volat
il
ity and sim
ulated
data are
v
is
ua
li
zed in
Fi
gur
e
4
.
Figure
4
.
(
a
)
Si
m
ula
te
d
Data
(
log
-
retu
r
ns
)
(
in
%),
(
b)
Sim
ula
te
d
Co
ndit
ion
a
l Vo
la
ti
li
ti
es (
in %
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
108
7
–
109
4
1092
Fr
om
Figure
4
,
ther
e e
xists cl
us
te
rin
g patt
er
n,
where
per
i
ods
of
high
vola
ti
lity
tend
to be
f
ollo
we
d by
per
i
od
s
of
e
ve
n
hi
gher
vola
ti
li
ty
.
Si
m
il
arly
,
per
i
od
s
with
l
ow
vola
ti
li
ty
te
nd
s
to
be
fo
ll
owed
by
low
vola
ti
lity.
The
patte
r
ns
of
Fig
ure
4
(a)
s
hows
t
hat
it
te
nds
to
fluctuate
at
the
m
ean
cl
os
e
to
0.
Fig
ure
4
(
b)
s
hows
that
pe
rio
ds
of
high
vola
ti
li
ty
cause
la
rg
e
changes
in
t
he
returns
,
w
her
e
as
per
i
od
s
of
low
vola
ti
li
t
y
c
ause
sm
al
l
cha
ng
e
s
in
the
r
et
urns
.
T
he
sim
ulate
d
dis
tribu
ti
on
ex
hibi
ts
nu
m
ber
s
of
ou
tl
ie
rs
w
hich
ref
l
ect
fr
e
qu
e
nt
siz
eable
c
ha
nges
i
n
vo
la
ti
li
ty
.
Nex
t
ste
p
is
t
o
si
m
ulate
from
t
he
joint
poste
ri
or
distrib
ution
of
the
S
V
pa
ra
m
et
er
(
,
,
)
us
in
g
th
e
functi
on
“s
vs
a
m
ple”.
Value
of
la
te
nt
l
og
vola
ti
li
ti
es
ℎ
0
,
ℎ
1
,
…
ℎ
a
nd
ret
urns
draw
in
MC
MC
is
us
e
d
i
n
t
hi
s
si
m
ulati
on
. T
he
r
es
ult i
s su
m
m
arized in Ta
bl
e 1
.
Table
1.
Po
ste
r
ior Param
et
er
Me
an
Stand
a
r
d
E
rror
5%
50%
95%
ESS
Mu
-
8.9
44
0.147
24
-
9.1
84
-
8.9
45
-
8.7
03
4295
Ph
i
0.902
0.037
43
0.834
0.907
0.953
117
Sigm
a
0.389
0.080
65
0.273
0.381
0.537
111
exp(m
u/2
)
0.011
0.000
85
0.010
0.011
0.013
4295
sigm
a^2
0.158
0.067
00
0.075
0.145
0.288
111
Table
1
s
how
s
the
value
of
po
ste
rior
par
am
et
ers
gi
ven
by
M
CM
C
sa
m
pler.
The
value
of
m
u
is
equ
al
to
-
8.9
44
wh
il
e
phi
is
e
qual
to
0.902.
Si
nce
va
lue
of
phi
is
hi
gh,
t
hu
s
,
it
s
ho
ws
high
pe
rsiste
nce
i
n
t
he
vo
l
at
il
ity.
Ther
e
f
or
e,
the
r
e
exists
t
he
vo
l
at
il
ity
cl
u
ste
rin
g.
The
sigm
a
or
vo
la
ti
li
ty
of
t
he l
og
-
vola
ti
li
t
y
value
is
0.389.
The
values
f
or
e
xp
(m
u/2
)
a
nd
sig
m
a^2
are
0.0
11
an
d
0.1
58
res
pe
ct
ively
.
The
pe
rcen
ta
ges
give
n
a
re,
5%
,
50
%,
a
nd
95%.
It
re
pr
e
se
nts
t
he
belie
f
of
t
he
possi
ble
va
lues
will
fall
within
it
.
The
va
lue
for
the
m
e
an,
m
u
is
guara
nteed
that
95%
of
it
s
possible
value
s
will
fall
at
-
8.7
03.
ES
S
in
t
he
la
st
colum
n
is
the
ef
fecti
ve
sam
ple
siz
e.
It
is
the
nu
m
ber
of ef
fe
ct
ively
ind
e
penden
t
draws
from
the posteri
or d
ist
rib
utio
n
t
ha
t t
he
MC
MC
is eq
uiv
al
e
n
t t
o.
Trace
plo
ts
in
Figure
5
in
dicat
e
wh
et
he
r
c
onverge
nce
can
be
safely
diag
no
sed
a
nd
how
w
el
l
the
ou
t
put
sam
ples
per
f
orm
.
Trace
plo
t
s
hows
t
he
value
s
that
par
am
et
e
r
to
ok
du
rin
g
MC
MC
ru
ntim
e.
All
th
ree
tra
ce
plot
s
sh
ow
that
the
MC
MC
al
go
rithm
has
converg
ed
.
From
the
t
race
pl
ot,
it
sh
ows
that
the
para
m
et
ers
do
no
t
exh
i
bi
t
sign
ific
a
nt
bias.
Figure
5
.
Tr
ace
Plot
for
MC
M
C Est
i
m
at
e o
f
θ
of the
O
btaine
d
P
os
te
ri
or Sa
m
ple
Stocks
that
m
ai
ntain
a
relat
iv
el
y
sta
ble
pri
ce
is
sai
d
t
o
ha
ve
low
vola
ti
li
ty
. Ho
we
ver,
f
r
om
Fig
ur
e
6
,
it
sh
ows
that
the
vola
ti
li
ty
is
not
relat
ively
sta
ble.
The
refo
r
e,
th
e
KLCI
is
sai
d
to
ha
ve
high
vola
ti
li
ty
,
he
nc
e
ind
ic
at
ing
high
r
isk
in
t
he
Ma
l
ay
sia
n
S
toc
k M
ark
et
for
the
forecast
in
g per
iod
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Mo
delli
ng
v
olat
il
i
ty
o
f ku
ala l
ump
ur
c
omp
osi
te
ind
ex (
KLC
I)
…
(
Siti
Meri
am Z
ahar
i
)
1093
Figure
6
.
Esti
m
at
ed
Vo
la
ti
li
ti
es in P
e
rcen
t
(%)
3.2
Co
m
pari
so
n Be
tween G
ARCH
a
nd S
V
m
od
el
s
The
val
ue
of
R
MSE
a
nd
M
A
E
f
or
both
m
od
el
s
is
c
om
par
ed
t
o
fin
d
the
be
st
m
od
el
.
T
he
com
par
is
on
betwee
n
the
se
two
m
od
el
s i
s
sh
ow
n
in
Ta
ble
2
.
Table
2.
C
om
par
iso
n betwee
n G
ARCH a
nd
SV
m
od
el
GA
RC
H
SV
RM
SE
0.014
218
0.014
213
MAE
0.009
8740
0.009
8702
T
he
val
ue
for
RM
SE
an
d
M
AE
f
or
S
V
a
re
sli
gh
tl
y
lo
wer
com
par
ed
to
G
ARCH
m
od
el
.
This
i
nd
ic
at
es
that
the
S
V
m
od
el
perf
or
m
ed
sli
gh
tl
y
bette
r
than
GA
RC
H
m
od
el
.
The
re
f
or
e
,
SV
m
od
el
is
the
best
m
od
el
to
be
e
m
plo
ye
d for
f
or
ecast
in
g
t
he KLCI
r
et
ur
ns
.
4.
CON
CLUSION
The
pap
e
r
pr
es
ents
the
m
od
el
li
ng
vola
ti
li
t
y
of
the
K
uala
Lu
m
pu
r
Com
po
sit
e
Index
(
KLC
I)
us
in
g
S
V
and
GA
RC
H
m
od
el
s. The
find
i
ngs
s
how
th
at
both
m
od
el
s
are
s
uperi
or
to
each
oth
e
r,
with
sli
gh
t
dif
fer
e
nces
i
n
te
rm
s o
f
RM
S
E and
MAE
va
lues.
For
for
ec
ast
ing
pur
pos
e
s,
it
is
sug
geste
d
t
o
em
plo
y S
V
m
od
el
for
m
anag
i
ng
the
ris
ks
since
it
prov
i
des
m
or
e
in
f
or
m
at
ion
about
t
he
data.
It
al
s
o
f
ocu
s
e
s
on
the
m
ark
et
ne
ws
a
nd
al
s
o
t
he
i
m
pact
of
the
vo
la
ti
li
ty
towa
rd
s
the
ret
urns
.
F
or
fu
t
ur
e
st
ud
ie
s
,
it
is
rec
omm
end
ed
to
us
e
dif
f
ere
nt
pr
i
or
par
am
et
er
val
ue
s
since
the
c
hoic
e
f
or
pa
ram
eter
m
ay
produce
di
ff
e
ren
t
res
ults,
es
pecial
ly
in
te
rm
s
of
persi
ste
nce
of volat
il
it
y.
ACKN
OWLE
DGE
MENT
The
aut
hors
w
ould
li
ke
to
tha
nk
t
he
Mi
nistry
of
Higher
E
du
c
at
ion
Ma
la
ysi
a
(MO
HE)
a
nd
t
he
Re
sea
rc
h
Ma
nag
em
ent
I
ns
ti
tute
(RMI
)
of
U
niv
e
rsiti
T
eknolo
gi
M
AR
A,
Ma
la
ysi
a
f
or
s
upportin
g
t
hi
s
pro
j
ect
unde
r
the
Fund
am
ental
Resea
rch G
ra
nt
Schem
e G
ra
nt
No
:
600
-
RM
I/
FRGS 5/3
(0
086/2016
))
.
REFERE
NCE
S
[1]
Hull
J
.,
W
hit
e
A.
,
"
The
Pric
ing
of
Options
on
As
sets
with
Stoch
astic
Vola
ti
l
it
i
es
,"
The
Journal
of
Finance
,
vol
.
42(2)
,
pp.
281
-
300
,
19
87
.
[2]
Mate
i
M.
,
"
As
sess
ing
Volat
i
li
t
y
Forec
asti
ng
Mo
del
s:
W
h
y
GA
R
CH
Models
T
ak
e
th
e
Lead.
"
Ro
manian
Journal
of
Ec
onomic
Fore
c
asting
.
vo
l.
4(4)
,
42
-
65
,
2009
.
[3]
Franses
P.H.,
v
an
d
er
L
ei
j
,
M.J.,
Paap
R
.
,
"
A
Sim
ple
T
est
for
GA
RCH
ag
ai
n
st
a
Sto
cha
sti
c
Volat
ilit
y
Mode
l
,"
Ec
onometri
c
Ins
ti
tute
Re
search
Pape
rs
EI
2
005
-
41
.
2005
.
[4]
Plat
ani
o
ti
E. J.
,
"
A Re
vie
w of
Sto
cha
sti
c
Vola
ti
l
ity
:
Univar
ia
t
e an
d
Multi
va
riate Models
,"
Imperi
al
Coll
ege London,
London.
Technic
al
Pap
er
,
2005
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
13
, N
o.
3
,
Ma
rc
h 201
9
:
108
7
–
109
4
1094
[5]
Grunbichl
er
,
A,
Longsta
ff
,
F.A
.
"
Valu
ing
Futu
res
and
Options
on
Vola
ti
l
ity
,"
Journal
of
Ban
king
&
Finan
c
e
,
vol.
20(6)
,
pp
.
9
85
-
1001
,
1996
.
[6]
Meli
no
A.
T
.
,
"
Prici
ng
fore
ign
cur
ren
c
y
op
ti
on
s
with
stoch
astic
vo
la
t
il
i
t
y
,"
Jo
urnal
of
E
cono
metric
s
,
vol
.
45
,
pp.
239
–
265
,
19
90
[7]
Anderson
T.
G.
,
Lund
J.
,
"
Esti
m
at
ing
cont
inuous
-
ti
m
e
stoch
asti
c
vola
tilit
y
m
ode
l
s
of
the
short
-
term
int
ere
st
rate
,"
Journal
of
e
cono
metric
s
,
vo
l.
77(
2)
,
pp
.
343
-
377
,
1997
.
[8]
Kim
S.,
Sheph
ar
d
N.,
Chib
S.
,
"
Stocha
stic
vo
la
t
i
li
t
y
:
L
ikelihood
Infe
ren
c
e
and
C
om
par
ison
with
ARCH
Models.
"
Re
v
ie
w
o
f Ec
ono
mic
Stud
ie
s
,
vol.
65
,
pp.
361
–
393
,
1998
.
[9]
Naka
ji
m
a
J
.,
"
Ba
y
esia
n
ana
l
y
sis
of
GA
RCH
and
stocha
sti
c
volatilit
y
:
m
odel
ing
ev
era
ge
,
jumps
an
d
hea
v
y
-
t
ai
ls
for
fina
nc
ia
l
ti
m
e
se
rie
s
,"
Discuss
ion
Pape
r No
.
2008
-
E
-
23
,
2009
.
[10]
Jawed
S.
M
.
,
"
Applicati
on
of
Sto
cha
sti
c
Vol
at
i
li
t
y
Model
to
KS
E
-
1
00
,"
Pa
ki
stan
Jo
urnal
of
Engi
n
eer
ing,
Techno
log
y
&
S
ci
enc
e
,
vol
.
4(1)
,
pp
.
28
-
40
,
2014
[11]
Yu
J.
,
"
Forec
asti
ng
Vol
atilit
y
i
n
th
e
New
Z
eala
nd
Stock
Mar
ket
.
"
App
li
ed
F
inanc
ial
Ec
ono
mics
,
vo
l.
12(3)
.
Pp.
193
-
202
,
20
12
.
[12]
Kastne
r
G.
,
"
De
a
li
ng
wi
th
Stoch
a
stic
Vol
at
i
li
t
y
in
Ti
m
e
Seri
es
Us
i
ng
the
R
Pa
cka
g
e
Stochvo
l.
"
Jou
rnal
of
S
tat
ist
ic
a
l
Soft
ware
,
vol
.
69
,
pp
.
1
-
30
,
201
6
.
[13]
Aı
Y.,
&
Kim
m
el
R.
,
"
Maxi
m
um
li
kel
ihood
esti
m
a
ti
on
of
stocha
sti
c
vo
la
t
i
li
t
y
m
ode
ls
,
"
Jo
urnal
of
fi
nan
cial
ec
onomics
,
vol
.
83
(2)
,
pp
.
413
-
4
52
,
2007
.
[14]
Chib,
S.,
N
ard
ari,
F.,
&
Shephar
d
,
N.
"
Markov
chain
Monte
Car
lo
m
et
hods
for
stocha
stic
vo
la
t
il
i
t
y
m
odel
s
,
"
Journal
of
E
conome
tric
s
,
vol.
108
(2),
pp.
281
-
316
,
2002
.
[15]
Goh
Soo
Khoon,
M.
L.
-
H.
,
"
Th
e
I
m
pac
t
of
the
Glo
bal
Finan
ci
a
l
Cri
sis:
The
Case
of
Malay
s
ia
,
"
TWN
Global
Ec
onomy
Serie
s
,
pp
.
1
-
38
,
2010.
Evaluation Warning : The document was created with Spire.PDF for Python.