ACCURACY DRIVEN ARTIFICIAL NEURAL NETWORKS IN STOCK MARKET PREDICTION

International Journal on Soft Computing (IJSC) Vol.3, No.2, May 2012
DOI: 10.5121/ijsc.2012.3203 35
ACCURACY DRIVEN ARTIFICIAL NEURAL
NETWORKS IN STOCK MARKET PREDICTION
Selvan Simon1
and Arun Raoot2
1
Asst. Professor, P. G. Dept. of Computer Sci., SNDT Women’s University, Mumbai
selvansimon@gmail.com
2
Professor, National Institute of Industrial Engineering, Vihar Lake, Mumbai
adraoot@gmail.com
ABSTRACT
Globalization has made the stock market prediction (SMP) accuracy more challenging and rewarding for
the researchers and other participants in the stock market. Local and global economic situations along
with the company’s financial strength and prospects have to be taken into account to improve the
prediction accuracy. Artificial Neural Networks (ANN) has been identified to be one of the dominant data
mining techniques in stock market prediction area. In this paper, we survey different ANN models that have
been experimented in SMP with the special enhancement techniques used with them to improve the
accuracy. Also, we explore the possible research strategies in this accuracy driven ANN models.
KEYWORDS
Artificial Neural Networks, Multilayer Perceptron, Back Propagation, Stock market prediction &
Prediction accuracy.
1. INTRODUCTION
Investors in stock market want to maximize their returns by buying or selling their investments at
an appropriate time. Since stock market data are highly time-variant and are normally in a
nonlinear pattern, predicting the future price of a stock is highly challenging. With the increase of
economic globalization and evaluation of information technology, analyzing stock market data for
predicting the future of the stock has become increasingly challenging, important and rewarding.
With the development of ANN, researchers are hoping to demystify the stock market because of
its great capability in pattern recognition and machine learning problems such as classification
and prediction.
This paper surveys the ANN models used in SMP. It is organized as follows: section 2 introduces
the ANN. Section 3 introduces the stock market fundamentals and describes common models
used in the stock market prediction. Section 4 explores ANN models, their variations, and special
techniques applied in SMP. Finally, in section 5, we discuss a number of strategies for improving
the prediction accuracy.

36
2. ARTIFICIAL NEURAL NETWORKS
Nowadays, ANN is considered as a common data mining method in different fields like finance,
economy, medical, business, industry, science, and so on. ANNs are computer models built to
emulate the human pattern recognition function.
2.1. Structure of an ANN
It consists of fundamental processing elements called neurons. These neurons are distributed in
few hierarchical layers. Most of the neural networks are three layered: input, middle or hidden,
and output. Generally, there occurs no data processing at the input layer. The input layer takes the
inputs and passes to the middle layer. There can be more than one middle layers. These hidden
layers are where all the complexity resides and the computations are done. The data or
information is distributed through the network and stored in the form of interconnections. These
interconnections between artificial neurons are called weights. Fig.1 shows a representation of an
ANN and a simplified neuron. A neuron, like other linear or polynomial approximation, relates a
set of input variables {Xi}, i=1,……,n to set of one or more output variables {Yj}, j=1,…,m. But,
in case of ANN, the input variables are mapped to the output set by squashing or transforming by
a special function f, known as activation function. Each neuron also has a bias assigned to it. Each
neuron receives an input signal, which transmits through a connection that multiplies its strength
by the scalar weight w. A bias is added to the weighted input and is then passed through the
activation function to get the desired output [1].
2.2. Characteristics of ANN
The weight w and the bias b are adjustable parameters of the neuron. The weights between two
neurons in two adjacent layers are adjusted to reduce errors through an iterative training process
while training samples are presented to the network. This training process lasts till the error
between actual output and expected value meets the requirements, so that the satisfactory weights
and threshold can be achieved. Thus, they learn the relationship inherent in variables from a set of
training sample. Back propagation (BP) algorithm is the most widely used training algorithm for
ANN. They store the captured knowledge and make it available for future use. ANN has the
capability to adapt the network parameters to the changes in the studied system. Moreover, when
the system under study is non-stationary and dynamic in nature, the ANN changes its network
parameters in real time [1].
Fig. 1 Representation of an ANN and a simplified neuron

37
3. APPLICATIONS OF ANN IN STOCK MARKET
In this section, we briefly introduce the stock market and the evolution of the stock market
prediction models.
3.1. Stock market basics
A stock market is a public market for the trading of company stock and derivatives at an agreed
price. Stock market gets investors together to buy and sell shares in companies. Share market sets
prices according to supply and demand. Hence, a stock that is highly in demand will increase in
price, whereas a stock that is being heavily sold will decrease in price. Primary market deals with
the new issues of securities directly from the company. An official prospectus is published under
the corporation law that contains all the information about the company that is reasonably
required by the investors to make an informed investment decisions. The existing securities are
bought and sold in the secondary market among traders. A share is a document issued by a
company, which entitles its holder to be one of the owners of the company. By owning a share
one can earn a portion of the company's profit called dividend. Also, by selling the shares one
gets the capital gain. However, there is a risk of making a capital loss, if selling price of the share
is lower than the buying price. Stock is a collection or a group of shares. Stock Exchanges act as
the clearing house for each transaction, that is, they collect and deliver the shares, and guarantee
payment to the seller of a security. The smooth functioning of all these activities facilitates
business expansion, economic growth, employments and promotes production of goods and
services. To be able to trade a security on a certain stock exchange, it has to be listed there.
Listing requirements are the set of conditions imposed by a given stock exchange on companies
that want to be listed. Stock brokers are licensed agents to trade shares. They have direct access to
the share market to do share transactions. They charge a fee for this service. Traders buy and sell
financial instruments such as stocks, bonds and derivatives. Traders are either professionals
working in a financial institution or a corporation, or individual investors [2].
3.2. SMP using traditional models
The characteristic that all stock markets instruments have in common is the uncertainty related to
their future states. This feature is undesirable for the investors, but unavoidable when stock
market is chosen for investment. While numerous attempts have been made, no method has been
discovered so far to accurately predict the stock price. However, stock analysts have been using
various approaches for predicting stock market. In this section, we briefly explain the important
approaches used in the stock market prediction.
3.2.1. Technical Analysis
Technical analysis is based on mining rules and patterns from the past prices of stocks. It analyses
the financial time series data to forecast stock market using indicators of technical analysis.
Technical analysts utilize charts and modelling techniques to identify trends in price and volume.
But charts or numeric data contain only the events and not the cause why it happened. It is
believed that market timing is critical and opportunity can be found through the careful averaging
of historical price and volume movements and comparing them against current price [3].
3.2.2. Fundamental Analysis
Fundamental analysis investigates the factors that affect supply and demand. The goal is to gather
and interpret this information and act before the information is incorporated in stock price.

38
Fundamental analysis is based on economic data of companies that they have to publish regularly,
for example, annual and quarterly reports, auditor’s reports, balance sheet, income statements, etc.
This method is concerned more with the company rather than the actual stock. The analysts make
their decisions based on the past performance of the company, its earnings forecast, the particular
industry sector and the overall economy [3].
3.3. SMP using ANN
ANN is nonlinear in nature and since the stock market returns change in a nonlinear pattern, the
ANN is more appropriate to model these changes [4]. Indeed, ANN is significantly more accurate
than its traditional competitive model, multiple linear regression analysis for SMP [5]. Moreover,
predicting stock index with traditional time series analysis has proven to be difficult. Because of
its easy adaptation to noisy data and ability to extract useful information from large data set and
to solve complex non-linear problems, ANN suits in predicting stock returns [6].
4. COMPETITIVE ANN MODELS IN SMP
While attempting to improve the SMP accuracy, researchers have experimented with many
variations of ANN models. Here, we discuss these ANN models and techniques:
4.1. Improving ANN accuracy by relevant ANN model
By comparing various ANN models and one can choose the model and its learning algorithm
most suitable for the given application, prediction target and problem situation to get the best
result:
Saad et al. [7] compared three networks for stock trend predictions. Authors compared the time
delay neural networks (TDNN), recurrent neural networks (RNN), and probabilistic neural
networks (PNN) neural networks, utilizing conjugate gradient and extended Kalman filter training
for TDNN and RNN. The history of the daily closing price was analysed. The three networks
showed comparable results. TDNN was moderate with respect to implementation complexity and
memory requirement. PNN was more suitable for stocks which do not need training on long
history, like Apple stock. RNN had the capability to dynamically incorporate past experience.
Quah [8] presented methodologies to select equities based on soft-computing models which focus
on applying fundamental analysis for equities screening. The performance of three models was
compared: multilayer perceptron (MLP), adaptive neuro-fuzzy inference systems (ANFIS) and
general growing and pruning radial basis function (GGAP-RBF). The author applied several
benchmark matrices to compare performance. The author also suggested how equities can be
picked systematically by using relative operating characteristics (ROC) curve. ROC curve has
two variables from confusion matrix, which are True Positive (TP) rate and True Negative (TN)
rate. One can maximize TP and minimize TN for optimal performance. The average appreciation
of equities was almost doubled for all three soft-computing models than the average market
appreciation.
Charkha [9] used the feed forward back propagation neural networks (BPNN) with early stopping
and radial basis networks (RBN) to predict the trend of stock price (i.e. classification) and to
predict the stock price (i.e. value prediction). Fundamental or technical indicators were not used
in this research as basic objective of this research was to determine the usability of ANN in
predicting the future prices based on past prices alone. It was observed that BPNN was better for
trend prediction and RBN was better for the value prediction.

39
Naeini et al. [10] used two kinds of neural networks, a feed forward MLP and an Elman recurrent
network to predict a company’s stock value based on its stock share value history. The
experimental results showed that the application of MLP neural networks was more promising in
predicting stock value rather than Elman recurrent networks and linear regression method.
However, based on the standard measures it was found that the Elman recurrent network and
linear regression can predict the direction of the changes of the stock value better than the MLP.
4.2. Improving ANN accuracy by statistical techniques
By integrating various standard statistical multivariate analytical techniques with ANN models,
one can improve the prediction accuracy:
Wang and Leu [11] used recurrent neural networks (RNN) with BP for forecasting mid-term price
trend of Taiwan stock market. The network was trained using features extracted from Auto
Regressive Integrated Moving Average (ARIMA) analyses. An ARIMA model is a linear non-
stationary model that uses difference operator to convert non-stationary series to stationary. It
does not work well in modelling non-linear series by itself. The experimental results showed that
the ARIMA-based recurrent neural network was capable of predicting the market trend with
acceptable accuracy.
Thenmozhi [12] applied neural networks to predict the daily returns of the Bombay stock
exchange (BSE) Sensex. An MLP using back propagation network was used. It was found that
the predictive power of the network model was influenced by the previous day. The study showed
that satisfactory results could be achieved by applying MLP to predict the BSE Sensex. A
sensitivity analysis was done in order to determine the relative importance of each input on the
output, once the network was fully trained. Sensitivities were determined by cycling each input
for all training patterns and computing the effect on the MLP output response. It indicated that the
immediate previous day return contributed significantly in predicting the returns compared to the
first three-day returns used for prediction.
Al-Luhaib et al. [13] examined Saudi stock market (SSM) to predict the direction of daily price
changes. BPNN was applied to predict the direction of price changes for the stocks listed in SSM.
The target had a representation of three classes 1, -1 and 0 that respectively represent the increase,
decrease or insignificant change in the stock prices. The use of dynamic target was a novel
enhancement to the traditional objective function. The weights updates through different training
epochs were slowly changed to smooth the convergence and achieve the generalization by
reaching the ideal target. The classification result was improved by using dynamic target.
Zhao et al. [14] predicted the stock price using BPNN by considering a single closing price as the
time series vector. The authors made a two steps forecast approach. First, use Gray correlation
analysis to choose the set of variable which can describe the characteristics of the state of the
stock market from a number of technical indicators. Then classify the state of stock market by the
self-organizing feature map (SOFM) network. And based on this classification, BPNN was used
for prediction. The results showed that the predictive accuracy of SOFM-BP model was better
than that of the traditional BPNN model.
ANN models were widely applied for stock market prediction. However, the parameter settings of
the networks are normally determined through a trial-and-error methodology. Hsieh et al. [15]
integrated the design of experiment (DOE), Taguchi method, and BPNN to construct a robust
engine to improve the prediction accuracy. Adopting data from Taiwan Stock Exchange (TWSE),
the technical analytical indexes of the listed stocks of TWSE were computed. The research results

40
indicated that the DOE-based predictor could effectively improve the forecasting rate of stock
price variations.
4.3. Improving ANN accuracy by special algorithms
It is also possible to combine special algorithms with ANN for denoising, selecting and
optimizing parameters to improve accuracy of ANN models in SMP:
Lee and Lim [16] presented a methodology for forecasting the daily Korea composites to price
index (KOSPI) using neural networks with self-adaptive weighted fuzzy membership functions to
improve forecasting accuracy. The degree of classification intensity was obtained by bounded
sum of weighted fuzzy membership functions extracted by this network. The Haar wavelet
function was used as a mother wavelet. A set of five extracted coefficient features of the Haar
wavelet transform were presented to forecast KOSPI. This model demonstrated an excellent
capability in SMP.
Li and Liu [17] analysed the complexity of the stock price system and used BPNN for stock
market prediction. They presented a method for determining the number of hidden layers,
selection and pre-treatment of sample data and determination of preliminary parameters. To avoid
local extreme and promote convergence speed, the network trained by the Levenberg-Marquardt
BP algorithm, based on numerical optimization. Simulation experiments on Shanghai stock
exchange index indicated that this algorithm could make efficient short-term prediction.
Olatunji et al. [18] used ANN model for the predicting stock market. The model was based on
SSM historical data covering a large span of time. They used only the closing price of the stock as
input to the system. The number of windows gap to determine the number of previous data to be
used was based on an algorithm for searching optimal parameter. The optimized parameter was
then used to build the final ANN model. The experimental results indicated that the proposed
ANN model predicted the next day closing price stock market value adequately accurate.
It is evident that the dominant data mining technique used in SMP are ANN models [19], [20].
Also, MLP with BP training found to be the most widely used ANN model in SMP. Researchers
have combined various statistical, mathematical and special algorithms with ANN models to
improve prediction accuracy. We have summarized these enhancement techniques in Table 1.

41
Table 1. Enhancement techniques for improving accuracy in SMP
Sr. Researchers Year Enhancement Techniques used with ANN models
1 Wang and Leu 1996 ARIMA to convert non-stationary series to stationary
2 Saad et al. 1998 Choosing problem specific ANN model
3 Thenmozhi 2006 Sensitivity analysis to find importance of each input
4 Al-Luhaib et al. 2007 Dynamic target to smooth the convergence
5 Lee and Lim 2007 Self-adaptive weighted fuzzy membership functions
6 Charkha 2008 Choosing target specific ANN model
7 Quah 2008 ROC curve to select equities
8 Li and Liu 2009 Levenberg-Marquardt algorithm to avoid local extreme
9 Zhao et al. 2009 Gray correlation analysis to choose the set of variable
10 Naeini et al. 2010 Choosing target specific ANN model
11 Hsieh et al. 2011 Taguchi method for parameter optimization
12 Olatunji et al. 2011 Heuristic algorithm for parameter optimization
5. STRATEGIES FOR IMPROVING ACCURACY
Despite the extensive research on applying ANN to SMP, this field is still evolving to improve
the accuracy. We have isolated some of the challenges and research strategies in this area:
5.1. Hybrid analysis
When applying machine learning and data mining to stock market data, researchers have mainly
focused in technical analysis to see if our algorithm can accurately learn the underlying patterns
in the stock time series. ANN can also play a major role in evaluating and predicting the
performance of the company and other similar parameters helpful in fundamental analysis.
Neither all possible technical information nor all available fundamental information or
combinations of both have been tested. Thus, both ideas have to be regarded [21]. The most
successful ANN stock prediction models may use some sort of a hybrid analysis model involving
both fundamental and technical indicators [3]. Building an ANN model that combines the process
of the technical and fundamental analyses will improve the prediction results.
5.2. Choice of inputs
The future returns of a stock can be based on a number of factors such as earnings per share,
capital investment, daily transaction volume, market share, stock prices, leading indicators,
macroeconomic data, interest rates, inflation rates, political issues, and many others. [11], [22–
25]. Different stock market parameter are also used: movement of index, major stock market
indices, domestic minimum loan rate, gold price [26], results of fundamental analysis [8], moving
average crossover inputs based on technical analysis rules [27] and so on. Further research is
anticipated to incorporate additional inputs that influence stock returns with this neural approach
[28]. Identifying, categorizing, and sorting the parameters in the order of their influence on stock
price using multivariate techniques will provide the most useful input collection.

42
5.3. Training ANN with stock market data
The usage and training of an ANN is an art [29]. Pre and post data processing issues such as
selecting, sampling, cleaning, filtering, denoising, normalizing, deseasonalising, validating,
pruning, segmenting, organizing, identifying, optimizing data for training ANN, for stock market
data are important. While applying an ANN model in SMP, care should be taken at every single
process. Developing methodologies for processing stock market data will improve the accuracy.
5.4. Lagged data input
In multi lag prediction, some predicted values are also used to predict futures values. The extent
of this lagged data to be incorporated and weights allotted to them are design issues. To predict
the five-day future index value for the market, Walczak [30] used the set of one-day, two-days,
and five-days lags of the closing value, along with the corresponding one-day, two-days, and five-
days normalized average trading volumes for the respective index markets as the input values for
ANN. An alternative input is to use lagged differenced time-series data [11]. One can consider the
lagged data up to a number of weeks plus the current week and use numerical methods like
forward and backward difference, interpolation to prepare input using the past, current and
recently predicted data along with their errors with appropriate weights.
5.5. ANN components optimization
MLP with BP training has been the most commonly used ANN model in SMP. The selection of
appropriate number of hidden layers, number of neurons in each layer, size of the training set,
initial values for weights, inputs to be included, activation function are the key design issues of
this model. Tweaking these components iteratively and applying tools like DOE, one can arrive at
the combination of the components that yields the best prediction.
5.6. Targets in stock market
ANN models should be considered along with the target application. For example, stock market
trend prediction and stock market prediction may require different ANN models and data. Targets
in stock market may be predicting market indices, market trend, market volatility, buy-hold-sell
alarm, high-low risk-return classification, triaging best to worst stock for trading strategy and so
on. The forecasting may be for long term, short term, given period, or instant. It may be for a
stock or a sector. Identifying the most suitable ANN model and data for a specific prediction
target will enhance the accuracy.
5.7. Hybrid models
One can divide the complex stock market prediction tasks into simpler subtasks, perform the task
and integrate the results to get better performance. Every data mining models has its own
strengths and weaknesses. By applying more than one data mining techniques, say genetic
algorithm and neural networks on two different subtasks, we can take the advantages of their
strengths to subsume their weaknesses [4]. Hybridization using two-layer and three-layer
approach have attracted many data mining researchers.

43
6. CONCLUSION AND FUTURE WORKS
ANN plays an important role in SMP. ANN models have outperformed other traditional models.
Also, MLP with BP training found to be the most widely used ANN models in SMP. There are
variations in ANN model. It is important to choose the appropriate ANN model considering the
problem target. Various statistical techniques may be used to pre-process the data for improving
performance. Also, special algorithms may be combined with ANN to improve accuracy. There
are several challenges and research scopes for improving the prediction accuracy. We will be
working on the strategies in section V, in future.
REFERENCES
[1] M. Majumder and A. Hussian, MD, “Forcasting of Indian Stock Market Index Using Artificial Neural
Network,” nseindia.com, pp. 1-21, 2010.
[2] D. V. Setty, T. M. Rangaswamy, and K. N. Subramanya, “A review on Data Mining Applications to
the Performance of Stock Marketing,” International Journal of Computer Applications, vol. 1, no. 3,
pp. 33-43, Feb. 2010.
[3] P. Falinouss, “Stock trend prediction using news articles,” Master’s thesis, Lulea University of
Technology, pp. 1653–0187, 2007.
[4] D. Zhang and L. Zhou, “Discovering golden nuggets: data mining in financial application,” Systems,
Man, and Cybernetics, Part C: Applications and Reviews, IEEE Transactions on, vol. 34, no. 4, pp.
513–522, 2004.
[5] J. Kutsurelis, “Forecasting financial markets using neural networks: An analysis of methods and
accuracy,” 1998.
[6] R. Dase and D. Pawar, “Application of Artificial Neural Network for stock market predictions: A
review of literature,” International Journal of Machine Intelligence, ISSN, vol. 2, no. 2, pp. 0975–
2927, 2010.
[7] E. W. Saad, D. V. Prokhorov, and D. C. Wunsch, “Comparative study of stock trend prediction using
time delay, recurrent and probabilistic neural networks,” Neural Networks, IEEE Transactions on,
vol. 9, no. 6, pp. 1456–1470, 1998.
[8] T. Quah, “DJIA stock selection assisted by neural network,” Expert Systems with Applications, vol.
35, no. 1-2, pp. 50-58, Jul. 2008.
[9] P. R. Charkha, “Stock Price Prediction and Trend Prediction Using Neural Networks,” 2008 First
International Conference on Emerging Trends in Engineering and Technology, pp. 592-594, 2008.
[10] M. Naeini, H. Taremian, and H. Hashemi, “Stock market value prediction using neural networks,”
2010 International Conference on Computer Information Systems and Industrial Management
Applications (CISIM), pp. 132-136, Oct. 2010.
[11] J. H. Wang and J. Y. Leu, “Stock market trend prediction using ARIMA-based neural networks,” in
Neural Networks, 1996., IEEE International Conference on, 1996, vol. 4, pp. 2160–2165.
[12] M. Thenmozhi, “Forcasting Stock Index Returns Using Neural Networks,” dbr.shtr.org, vol. 7, no. 2,
pp. 59-69, 2006.
[13] A. Al-Luhaib, K. Al-Ghoneim, and Y. Al-Ohali, “Dynamic Targets for Stock Market Prediction,” in
Signal Processing and Communications, 2007. ICSPC 2007. IEEE International Conference on,
2007, no. November, pp. 1019–1022.
[14] Q. Zhao, X. Zhao, and F. Duan, “Prediction Model of Stock Prices Based on Correlative Analysis and
Neural Networks,” 2009 Second International Conference on Information and Computing Science,
pp. 189-192, 2009.
[15] L. Hsieh, S. Hsieh, and P. Tai, “Enhanced stock price variation prediction via DOE and BPNN-based
optimization,” Expert Systems with Applications, vol. 38, no. 11, pp. 14178-14184, May 2011.
[16] S. H. Lee and J. S. Lim, “Forecasting KOSPI by weighted average defuzzification based on
NEWFM,” and Advanced Systems, 2007. ICIAS 2007., pp. 66–70, 2007.
[17] F. Li and C. Liu, “Application Study of BP Neural Network on Stock Market Prediction,” 2009 Ninth
International Conference on Hybrid Intelligent Systems, pp. 174-178, 2009.

44
[18] S. O. Olatunji, M. S. Al-Ahmadi, M. Elshafee, and Y. A. Fallatah, “Saudi Arabia stock prices
forecasting using artificial neural networks,” Applications of Digital, pp. 81–86, 2011.
[19] Y. Li and W. Ma, “Applications of Artificial Neural Networks in Financial Economics: A Survey,”
2010 International Symposium on Computational Intelligence and Design, pp. 211-214, Oct. 2010.
[20] A. M. Safer, “Predicting abnormal stock returns with a nonparametric nonlinear method,” in Neural
Networks, 2001. Proceedings. IJCNN’01. International Joint Conference on, 2001, vol. 3, no. 1993,
pp. 1833–1837.
[21] M. Schumann and T. Lohrbach, “Comparing artificial neural networks with statistical methods within
the field of stock market prediction,” in System Sciences, 1993, Proceeding of the Twenty-Sixth
Hawaii International Conference on, 1993, vol. 4, pp. 597–606.
[22] J. Boston, “A measure of uncertainty for stock performance,” in Computational Intelligence for
Financial Engineering (CIFEr), 1998. Proceedings of the IEEE/IAFE/INFORMS 1998 Conference
on, 1998, pp. 161–164.
[23] H. White, “Economic prediction using neural networks: The case of IBM daily stock returns,” in
Neural Networks, 1988., IEEE International Conference on, 1988, pp. 451–458.
[24] C. Ornes, “A neural network that explains as well as predicts financial market behavior,”
Computational Intelligence for Financial, pp. 43–49, 1997.
[25] A. Refenes, A. Zapranis, and Y. Bentz, “Modelling stock returns with neural networks,” in Neural
Network Applications and Tools. Workshop on (1993), 1993, pp. 39–50.
[26] P. Sutheebanjard and W. Premchaiswadi, “Stock Exchange of Thailand Index Prediction Using Back
Propagation Neural Networks,” 2010 Second International Conference on Computer and Network
Technology, pp. 377-380, 2010.
[27] M. Mehrara, A. Moeini, M. Ahrari, and A. Ghafari, “Using Technical Analysis with Neural Network
for Forecasting Stock Price Index in Tehran Stock Exchange,” Middle Eastern Finance and
Economics, no. 6, 2010.
[28] J. Roman and A. Jameel, “Backpropagation and recurrent neural networks in financial analysis of
multiple stock market returns,” in System Sciences, 1996., Proceedings of the Twenty-Ninth Hawaii
International Conference on,, 1996, vol. 2, pp. 454–460.
[29] Y. JingTao and L. T. Chew, “Guidelines for financial forecasting with neural networks,” . Neural
Information Processing, Shanghai,, 2001.
[30] S. Walczak, “Gaining competitive advantage for trading in emerging capital markets with neural
networks,” Journal of Management Information Systems, vol. 16, no. 2, pp. 177–192, 1999.
Authors
Selvan Simon is an Assistant Professor in the Post Graduate Department of Computer
Science, S N D T Women’s University, Juhu, Mumbai. His area of interest is Stock
Market Intelligent Systems, Decision Sciences, Web Technologies, E-Trading and M-
Trading . Email: selvansimon@gmail.com
Arun Raoot is a Professor in the Department of Industrial Engineering and IT
Applications, National Institute of Industrial Engineering, Vihar Lake, Mumbai. His
research interest includes Industrial Engineering, Facilities Planning, and IT Applications.
Email: adraoot@gmail.com

ACCURACY DRIVEN ARTIFICIAL NEURAL NETWORKS IN STOCK MARKET PREDICTION

More Related Content

What's hot (9)

Similar to ACCURACY DRIVEN ARTIFICIAL NEURAL NETWORKS IN STOCK MARKET PREDICTION (20)

Recently uploaded (20)

ACCURACY DRIVEN ARTIFICIAL NEURAL NETWORKS IN STOCK MARKET PREDICTION