the resulting outputs will be compared with the known outputs. If the result and the unknown output are not equal, changing the weights of the
connections will be continued until the difference is minimized. After acquiring the desired convergence for the networks in the learning process,
the validation dataset is applied to the network for the validating step Shahkarami A. et al. 2014.
Fig. 1 An artificial neural network is an interconnected group of nodes. Source : SPE International, Colorado, USA, 16–18April 2014.
2.2 Review of previous researches
Several economists advocate the application of neural networks to different fields in financial markets and economic growth methods of analysis Kuan,
C.M. and White, H. 1994. We focus the review of prior studies on prediction of financial market. Chen et al 2003 attempted to predict the trend of return
connections will be co o
nt nt
in inued until
l th
th e
e difference is minimized. After
acquiring the de desired convergence for the networ
ks ks
in the learning process, the vali
i da
dation dataset is a
appl p
ie ie
d d
to to
t t
he e
network for t the
he validating step
S S
ha hahkarami A.
e e
t al al
. 2
2014.
Fig. 1
1 A
An ar ar
ti ti
fi fici
ci l
al neura l
l ne
netw twor
ork k is
is an interc rcon
onne ne
ct d
ed group of f
no node
de s.
S So
ur ce
e :
: SP
SPE Internatio o
na n
l, Color r
a ado, USA, 16
6 –
– 18
18 Ap
Ap i ri
l l
20 2014
14.
2.2 Review of previous research h
es e
Several economists advocate the e
a application of neural networks to different
on the Taiwan Stock Exchange index. The probabilistic neural network PNN is used to forecast the trend of index return. Statistical performance of
the PNN forecasts is compared with that of the generalized methods of moments GMM with Kalman filter and random walk. Empirical results
showed that PNN demonstrate a stronger predictive power than the GMM– Kalman filter and the random walk prediction models. Kim 2003 used SVM
to predict the direction of daily stock price change in the Korea composite stock price index KOSPI. This study selected 12 technical indicators to
create the initial attributes. The indicators are stochastic K, stochastic D, Slow D, momentum, ROC, Williams’ R, AD oscillator, disparity 5,
disparity 10, OSCP, CCI and RSI. In addition, this study examined the feasibility of applying SVM in financial prediction by comparing it with
back-propagation neural network BPN and case-based reasoning CBR. Experimental results proved that SVM outperform BPN and CBR and
provide a promising alternative for stock market prediction. Altay Satman 2005 compared the forecasting performance artificial neural network and
linear regression strategies in Istanbul Stock Exchange and got some evidence of statistical and financial outperform of ANN models. Kumar Thenmozhi
2006 investigated the usefulness of ARIMA, ANN, SVM, and random forest regression models in predicting and trading the SP CNX NIFTY
Index return. The performance of the three nonlinear models and the linear model are measured statistically and financially via a trading experiment. The
the PNN forecasts is com ompared with
h t
t ha
ha t of the generalized methods of
moments GMM MM
with Kalman filter and random m walk. Empirical results
showed d
t t
h hat PNN demonstr
tr at
t e
e a
a st
st ro
ro ng
ng er
r p
p redictive powe
er r
than the GMM– Ka
a lm
lman filter an nd
d th th
e r
random walk prediction on m
m od
od e
els. Kim 2003 03
used SVM to pre
e di
dict ct t
t he
h direc
ec ti
ti on
of daily st
oc k price
ch ch
an an
ge in n
th th
e e
Ko K
rea co omp
m osite
st st
oc ock
k p
price e
i in
dex KOSPI. Thi s
study selected 12 te
techni i
ca a
l l in
indi di
cators rs to
create t t
h he
initial attri bu
tes. The ind
icators are st
ochastic K , sto
o ch
ch as
as tic D
, Sl
ow w
D, momentum, ROC, W
il liams’
R, AD oscillato
r r, dis
is pa
pa ri
ri ty
t 5,
di s
spar ity 10,
OS CP
, CCI
an d
RS I.
I n
addition , th
is study e
x xamined th
the e
fe e
as ibility of applyin
g SV
M in finan
ci al
p rediction by compa
r ring it
with h
ba ba
ck c
-propaga ti
on n
n n
eu eu
ra ra
l l
ne n
twork BPN N
an an
d d
ca ca
se se
-b as
ed reaso
ni nin
ng CB B
R R
. Experimental results proved
th that
t S
SVM outperform BPN and CBR R
an an
d d
pr prov
o ide a promising alternative for stock market prediction. Altay
ay S
Sat at
ma m
n 2
00 5
5 c
om pare
re d
d th th
e e foreca
a st
stin in
g g pe
perf rfo
ormance e
ar ar
ti ti
fi fi
i ci
l al
neu ra
l l netw
twor ork
k and li
li ne
ne ar reg
g re
re ss
ss io
ion strategies in n
I I
stanbul S
Stock Exchange e
a a
nd nd
g g
ot som
ome evidence of statistical and financial ou
utperform o of ANN models. Kumar Thenmozhi
2006 investigated the usef efulness o
f f ARIMA, ANN, SVM, and random
forest regression models in pr r
edic c
ti ting and trading the SP CNX NIFTY
Index return The performance of f
the three nonlinear models and the linear
empirical result suggested that the SVM model is able to outperform other models used in their study.
Hyup Roh 2007 introduces hybrid models with neural networks and time series model for forecasting the volatility of stock price index in two vision
points: deviation and direction and the results showed that ANN-time series models can increase the predictive power for the perspective of deviation and
direction accuracy. His research experimental results showed that the proposed hybrid NN-EGARCH model could be improved in forecasting
volatilities of stock price index time series. Adebiyi Ayodele A. et al. 2009 presented a hybridized approach which
combines the use of the variables of technical and fundamental analysis of stock market indicators for daily stock price prediction. The study used three-
layer one hidden layer multilayer perceptron models a feedforward neural network model trained with backpropagation algorithm. The best outputs of
the two approaches hybridized and technical analysis are compared. Empirical results showed that the accuracy level of the hybridized approach is
better than the technical analysis approach. Liao Wang 2010 applied a Stochastic Time Effective Neural Networks in predicting China global index
and their study results showed that the mentioned model outperform the regression model. Kara et al 2011 compared neural networks performance
and SVM in predicting the movement of stock price index in Istanbul Stock Exchange. The input variables in suggested models include technical
indicators such as CCI, MACD, LW R, etc. The results revealed that neural Hyup Roh 2007 in
n tr
troduces hybrid models ls
w w
ith neural networks and time series model
l f
for forecasting g
the volatility of stock p ri
ri ce
c index in two vision
points s
: :
deviation n
an n
d d
direct t
io io
n n an
a d
d th
th e
re re
su sult
lt s
s showed that
AN A
N-time series m
models can an
incre as
ase the predic ic
ti ti
ve ve
p p
ow ow
er for the p
perspec ec
tive of de vi
v ation and
dire e
ct ct
io io
n n
accura ra
cy . His research experimen
tal l
re re
sults s
sh sh
ow ow
ed tha hat the
pr pr
op op
osed h
hyb rid NN-EGARCH
mo del could
be imp
ro ove
v d
in in
f forecasti
ting vo
v lati
i l
li ti
es of stock pr ic
e index ti me
series. Ad
d eb
iyi Ay od
ele A. et al
. 200
9 pre
sent ed a hyb
ri di
ze d ap
pr proach
ch w
w hich
ch co
o mb
ines the use o f the
va ri
ab le
s of
t echn
ic al
and fundamental
analysi i
s s o
of sto
oc k
mark et ind
ic ators for daily stock price pr
ed icti
on. Th
e stud
y y us
us e
ed three ee-
- layer one hidden layer m
ul ul
ti tila
la yer pe
pe rc
rceptron models a feedforward neu eu
ra ra
l network model trained with backpropagation algorithm. The best o
o u
utpu puts
ts of
th th
e e tw
tw o
o ap
ap pr
pr oa
oa ch
ches hy
hybr br
id id
iz iz
ed a
a nd
nd t
tec ec
hn hn
i ic
al al a
a na
na ly
ly si
si s
s a a
re re
c om
om pa
pared. Em
Empi piri
rica cal
l re result
lt s
s showed tha hat the accura
ra cy
level o f
f th
th e
hy hy
br br
id id
iz iz
ed ed
a a
pp pproach is
better than the technical anal alysis appro
roach. Liao Wang 2010 applied a Stochastic Time Effective N
Neural Netw works in predicting China global index
and their study results show ed
e tha
a t
t the mentioned model outperform the regression model. Kara et al 20
1 11 compared neural networks performance
networks work better in prediction than SVM technique. Zhou Wang et al 2011, propose a new model to predict the Shanghai stock price. They used
Wavelet De- noising- based Back propagation WDBP neural network. For demonstrating superiority new model in predicting, the results of it is
compared with Back Propagation neural network and the total results showed that the WDBP model for forecasting index is better than BP model.
Putra and Kosala 2011 try to predict intraday trading Signals at IDX they used technical indicators - the Price Channel Indicator, the Adaptive Moving
Averages, the Relative Strength Index, the Stochastic Oscillator, the Moving Average Convergence-Divergence, the Moving Averages Crossovers and the
Commodity Channel Index. The result of their experiments showed that the model performs better than the naïve strategy. Also Veri and Baba 2013
forecasting the next closing price at IDX, they used opening price, highest price, lowest price, closing price and volume of shares sold as experimental
variables. The result showed that the most appropriate network architecture is 5-2-1 with dividing the data into two parts, with 40 training data with 95
accuracy of data and 20 test data with 85 accuracy of data.
2.3 Learning Paradigms in ANNs
