 Open Access
 Total Downloads : 237
 Authors : C. Ugwu, Onwuachu Uzochukwu C
 Paper ID : IJERTV3IS041849
 Volume & Issue : Volume 03, Issue 04 (April 2014)
 Published (First Online): 16052014
 ISSN (Online) : 22780181
 Publisher Name : IJERT
 License: This work is licensed under a Creative Commons Attribution 4.0 International License
A Hybrid Factor Based Neural Network Model Application for Stock Price Prediction
C. Ugwu and Onwuachu Uzochukwu C
Department of Computer Sciences, University of Port Harcourt, Choba, Nigeria.
Abstract – This paper demonstrates the use of neural network error back propagation algorithm in analyzing and predicting of Nigeria Stock Market Prices. Nigeria stock market prices were collected for the period of one thousand, two hundred and three days and subjected into training, validation and testing. A zero mean unit variance transformation was used to normalize the input variables in order to allow the same range which makes them to differ by order of magnitude. A 14j1 network topology was adopted because of fourteen input variables in which variable j was determined by the number of hidden neurons during network selection. The technical and fundamental data served as input into the error back propagation algorithm which was simulated with MATLAB and implemented with Java programming language. From the results obtained, Nestle Nigerian Plc. recorded a mean squared error (MSE) and regression (R) values of 466186e6 and 0.999923 respectively, Guinness Nigerian Plc. recorded a mean squared error (MSE) and regression

values of 8.29839e7 and 0.999873 respectively and Total Nigerian Plc. recorded a mean squared error (MSE) and regression (R) values of 3.07993e6 and 0.999193 respectively. The result shows that using hybridize factors input variables gives a better result in stock market prediction with minimum amount of error.
Keywords: Neural network, regression value, mean square, technical analysis and fundamental analysis.

INTRODUCTION
Prediction in stock market has been a hot research area for many years [6]. If any system which can consistently predict the trends of the dynamic stock market be developed, it would make the owner of the system wealthy. Many investments professionals and market participants have met the efficient market hypothesis with skepticism and regard it purely as conservative academic opinion. They believed that mechanism can be devised to predict market prices. The characteristic that all stock markets have in
common is the uncertainty, imprecision which is related with their short and longterm future state. The purpose of prediction is to reduce uncertainty associated to investment decision making. [11]. This feature is undesirable for the investor but it is also unavoidable whenever the stock market is selected as the investment tool. Stock market follows a random walk, which implies that the best prediction you can have about tomorrows value is todays value. Indisputably, forecasting stock indices is very difficult because of the market volatility that needs accurate forecast model. The stock market indices are highly fluctuating and it affects the investors belief. Determining more effective ways of stock market index prediction is important for stock market investors in order to make more informed and accurate investment decisions. Many studies on stock market prediction using Artificial Neural Networks or statistical methods were performed on technical raw data. These data might have been affected by inflation or fluctuation of exchange rates especially in developed countries such as Nigeria [2].
Back propagation neural network is commonly used for price prediction [14]. The objective of this paper is to demonstrate the use of neural network back propagation learning algorithm in training and simulating stock market data using hybrid factor variables as inputs to generate the optimal prediction.

LITERATURE REVIEWS Predicting the stock market is very difficult since it depends on several known and unknown factors. So many methods like technical analysis, fundamental analysis, time series analysis and statistical analysis have been used in attempting to predict the price in the stock market, but none of these methods are proved as a consistently acceptable prediction tool. This paper will not be complete without mentioning researches done by scholar in this area; Kyoungjae proposes genetic algorithms (GAs) approach to feature discrimination and the determination of
connection weights for artificial neural networks (ANNs) to predict the stock price index. In their study, GA is employed not only to improve the learning algorithm, but also to reduce the complexity in feature space. Experiment results show that GA approach to the feature discrimination model outperform the other two conventional models. [4]
Qing et al. in their study used artificial neural networks to predict stock price movement (i.e., price returns) for firms traded on the Shanghai stock exchange. We compare the predictive power using linear models from financial forecasting literature to the predictive power of the univariate and multivariate neural network models. Our results show that neural networks outperform the linear models compared. These results are statistically significant across our sample firms, and indicate neural networks are a useful tool for stock price prediction in emerging markets, like China. [8]
YiHsien, in his study integrated new hybrid asymmetric volatility approach into artificial neural networks optionpricing model to improve forecasting ability of derivative securities price. Owing to combines the new hybrid asymmetric volatility method can be reduced the stochastic and nonlinearity of the error term sequence and captured the asymmetric volatility simultaneously. Hence, in the ANNS optionpricing model, the results demonstrate that GreyGJR GARCH volatility provides higher predictability than other volatility approaches. [12]
PeiChan et al. in their study, an integrated system, CBDWNN by combining dynamic time windows, case based reasoning (CBR), and neural network for stock trading prediction is developed and it includes three different stages, beginning with screening out potential stocks and the important influential factors; and using back propagation network (BPN) to predict the buy/sell points (wave peak and wave trough) of stock price and adopting case based dynamic window (CBDW) to further improve the forecasting results from BPN. The empirical results show that the CBDW can assist the BPN to reduce the false alarm of buying or selling decisions.[7]
ShengHsun et al., in their study employs twostage architecture for better stock price prediction.
Specifically, the selforganizing map (SOM) was first used to decompose the whole input space into regions where data points with similar statistical distributions are grouped together, so as to contain and capture the nonstationary property of financial series. After decomposing heterogeneous data points into several homogenous regions, support vector regression (SVR) is applied to forecast financial indices. The proposed technique was empirically tested using stock price series from seven major financial markets. The results show that the performance of stock price prediction can be significantly enhanced by using the twostage architecture in comparison with a single SVR model. [11]
Zhang in their paper proposed an improved bacterial chemo taxis optimization (IBCO), which is then integrated into the back propagation (BP) artificial neural network to develop an efficient forecasting model for prediction of various stock indices. Experiments show its better performance than other methods in learning ability and generalization. [13]
Akinwale examined the use of error back propagation and regression analysis to predict the untranslated and translated Nigeria Stock Market Price (NSMP). The autor was used 5 j 1 network topology to adopt the five input variables. The number of hidden neurons determined the variables during the network selection. Both the untranslated and translated statements were analyzed and compared. The Performance of translated NSMP using regression analysis or error propagation was more superior to untranslated NSMP. The result was showed on untranslated NSMP ranged for 11.3% while 2.7% for NSMP. [2]
Haven examined other scholars work; this research looks at a hybrid factor based Neural Network model application for Stock Price Prediction. In this research important fundamental and technical factors that affect stock market will form the input for the Neural Network. This fundamental factor includes the gross domestic product and inflation rate and then the technical factor includes the close price, opening price, lowest price, the highest price and the volume.

MATERIALS AND METHODS
For the purpose of training our data, the back propagation algorithm was used, we obtained data from the daily index values of the Nigerian stock exchange (NSE) for three selected companies. The selected companies include Total Nigeria Plc., Nestle Nigeria Plc. and Guinness Nigeria Plc. The data were collected from January 2008 to march 2013. The source of data used is www.cashcraft.com which provides daily stock market data reports of these companies. The system requirement includes the system input variables which comprises of both the technical and the fundamental input variables. [1] The technical and the fundamental input variables are listed as
Oi1 the opening price of day i1 Oi2 the opening price of day i2 Hi1 the daily high price of day i1 Hi2 the daily high price of day i2 Li1 the daily low price of day i1 Li2 the daily low price of day i2 Ci1 the closing price of day i1
Ci1 the closing price of day i2 Vi1 the trading volume of day i1 Vi2 the trading volume of day i2
Gi1 thegross domestic product of year i1 Gi2 thegross domestic product of year i2 Ii1 the inflation rate of year i1
Ii2 the inflation rate of year i2
We have a total of 14 input variables, these inputs were normalized which is an appropriate stage in training the data obtained using neural networks applications that was developed. The input data is normalized into the range of [0, 1] or [1, 1] according to the activation function of the neurons. [5]. In this paper the value of the stock market is normalized into the range of [0, 1] using a sigmoid function and the neural networks are trained and tested using the back propagation algorithm.
A. The Architectural Model of the Proposed System
The architecture of this model consists of a 14j 1network topology because of fourteen input variables in which variable j was determined by the
number of hidden neurons during network selection.
X13,…X114. This is the combination of both technical and fundamental variables. The hidden layer of X21, X22, and X23 are intermediate variables which interact by means of weight matrices with adjustable weights to produce the output.
Information Flow
Fig: 3.1: Neural Network Model Architecture of the System
Forward propagation is a supervised learning algorithm and describes the "flow of information" through a neural net from its input layer to its output layer.[3] The feed forward algorithm was used to calculate the optimal weights of the stock prediction. The mathematical models for the feed forward algorithm are as follows:
Figure: 3.1 depicts a schematic diagram of a 1431
Inputj x j yi wij
3.1
topology and fourteen input variableare: X11, X12,
yi is the generated output and wi j represents weights
f (x)
1
1 e x j
3.2
f (x) is a sigmoid that is used as the activation function
Error Tk Ok
3.3
Tk is the observed (True) output while Ok is the calculated (actual) output
The error in the output layer is calculated by using the formula in equation 3.4
k ok (1 – ok )(Tk ok )
3.4
Fig 4.1 Neural Network Fitting Tool for Data Selection
Where Ok is the calculated (actual) output expressed
in equation 3.5
Figure 4.2 shows the Neural Network fitting tool for selection of network size. This interface gives the
Ok
1
1 e xk
3.5
user the opportunity to select the number of neuron in the networks hidden layer. The user can return to this panel and change the number of the neuron if the
Tk is the observed (True) output
The back propagation error in the hidden layer is calculated by using the formula in equation 3.6
j o j (1 – o j ) k * w jk
network does not perform well after training.
k 3.6
Where wjk is the weight of the connection from unit j to unit k in the next layer and k is the error of unit k.
The weight adjustment formula in equation (3.7) is used to adjust the weights to produce new weights which are fed back into the input layer.
Wnew Wold * * input
3.7
Where is a constant called the learning rate. The learning rate takes value between 0 and 1.

EXPERIMENTS AND RESULTS
The simulation was done using Matlab. Figure 4.1 shows the Neural Network fitting tool for data selection. This Interface helps in collection of input data and the target data from the work space.
Fig 4.2 Neural Network Fitting Tool for Network Size Selection
Figure 4.3 shows the Neural Network training. The Neural Network model was trained using Levenberg Marquardt back propagation. The network is trained to fit the inputs and the target. This means that neural network map between a data of numeric inputs and a set of numeric targets. Training automatically stops
when generalization stops improving as indicated by the increase in the mean square error of the validation samples.
Fig 4.3 Neural Network Training Tool
The neural network fitting tool will help in training network and evaluation its performance using mean square error and regression analysis. Training multiple times will generate different results due to different initial condition and sampling. Figure 4.4 shows the result of the trained Network.
Fig 4.4 Neural Network Fitting Tool for Displaying the Result of the Trained Network.
Table: 4.1. The Performance Analysis of the Nestle Nigerian Plc. Prediction
No of
Samples
Mean Squared Error (MSE)
Regression
(R) Values
Training
842
4.66186e6
0.999923
Validation
180
3.25520e5
0.999466
Testing
180
4.25777e5
0.99244
Table: 4.2 the Performance Analysis of the Total Nigerian Plc. Prediction
No of Samples
Mean Squared Error (MSE)
Regression (R) Values
Training
842
3.07993e6
0.999193
Validation
180
4.91378e6
0.998718
Testing
180
8.15871e6
0.998029
Table: 4.3 The Performance Analysis of the Guinness Nigerian Plc. Prediction
No of Samples
Mean Squared Error (MSE)
Regression
(R) Values
Training
841
8.29839e7
0.999873
Validation
180
1.07882e6
0.999852
Testing
180
2.44880e6
0.999690
Tale: 4.4 Stock Predictions for Nestle Nigerian Plc.
Sample of Daily Stock Price Prediction for Nestle Nigerian Plc.
Sample Date
Actual value
Predicted Values with Different Neural Network Predictive Models
14211
14201
14191
14181
14171
14161
14151
14141
20/3/2013
834.00
835.01
834.50
834.48
843.66
835.28
834.76
840.82
830.89
21/3/2013
870.00
864.64
861.95
861.99
861.87
863.36
866.46
860.40
864.96
22/3/2013
860.00
863.87
861.01
861.65
861.57
863.11
862.69
861.55
864.90
25/3/2013
860.00
861.84
861.98
863.91
866.07
859.55
865.94
864.45
862.21
26/3/2013
916.00
912.00
911.46
914.42
916.25
913.44
915.46
910.74
911.56
27/3/2013
940.00
946.30
941.92
941.39
940.76
945.55
952.76
944.13
945.24
28/3/2013
950.00
945.23
946.31
949.01
949.11
949.46
947.80
950.35
951.12
2/4/2013
960.02
957.74
956.99
957.39
957.82
957.16
958.59
957.61
958.12
3/4/2013
960.11
960.05
958.62
958.56
956.00
958.42
959.09
960.04
959.91
4/4/2013
960.15
959.93
958.40
959.13
959.54
956.73
959.06
957.90
959.15
5/4/2013
960.50
961.13
960.75
961.33
961.89
959.15
961.81
960.65
961.17
8/4/2013
970.00
969.12
968.77
969.51
970.38
967.88
969.38
968.21
969.39
9/4/2013
973.00
971.28
970.76
970.48
971.03
970.02
970.93
970.68
971.90
10/4/2013
972.00
971.70
970.67
970.61
971.02
969.25
971.01
970.38
971.65
1
0.95
0.9
0.85
14211 Predicted
14211 Actual
1
0.95
0.9
0.85
14201 Predicted
14201 Actual
0.8
0.75
0.8
0.75
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
Fi g 4.5 Graph of 14 21 1 Pr edictive mod el.
1
0.95
0.9
0.85
14191 Predicted
14191 Actual
0.8
0.75
20/3/ 2013
21/3/ 2013
22/3/ 2013
25/3/ 2013
26/3/ 2013
27/3/ 2013
28/3/ 2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/ 2013
Fi g 4.7 Graph of 14 19 1 Pr edictive mod el.
Fi g 4.6 Graph of 14 20 1 Pr edictive mod el.
1
0.95
0.9
0.85
14181 Predicted
14181 Actual
0.8
0.75
20/3/ 2013
21/3/ 2013
22/3/ 2013
25/3/ 2013
26/3/ 2013
27/3/ 2013
28/3/ 2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/ 2013
Fi g 4.8 Grap h of 14 18 1 Pr edictive mod el.
Table: 4.5 Stock Predictions for Total Nigerian Plc.
Sample of Daily Stock Price prediction for Total Nigerian Plc.
Sample Date
Actual value
Predicted Values with Different Neural Network Predictive Models
14211
14201
14191
14181
14171
14161
14151
14141
20/3/2013
161.00
160.70
161.03
161.5
161.18
161.15
161.26
161.48
161.30
21/3/2013
161.00
160.65
161.05
161.41
161.13
161.18
161.28
161.49
161.30
22/3/2013
161.00
160.66
161.05
161.36
161.11
161.17
161.29
161.48
161.25
25/3/2013
161.00
160.66
161.05
161.36
161.11
161.17
161.29
161.48
161.25
26/3/2013
161.00
160.00
160.77
161.57
161.69
161.12
160.26
161.19
160.86
27/3/2013
161.00
161.42
160.58
161.37
161.88
161.15
161.24
162.46
161.90
28/3/2013
169.05
169.59
169.49
167.82
169.57
169.38
170.11
170.37
170.60
2/4/2013
180.00
178.16
180.19
180.14
180.88
179.80
179.69
178.69
180.39
3/4/2013
180.00
180.48
180.33
180.41
179.73
180.16
180.30
179.26
180.46
4/4/2013
180.00
180.19
180.45
180.09
179.72
179.86
179.59
180.16
179.96
5/4/2013
180.00
180.14
180.46
179.94
179.67
179.84
179.65
180.13
179.95
8/4/2013
180.00
180.30
180.41
180.05
179.78
179.88
179.62
180.21
179.97
9/4/2013
180.00
180.32
180.41
180.26
179.84
179.84
179.54
180.23
179.98
10/4/2013
180.00
179.55
180.67
179.47
179.18
179.99
179.82
179.82
179.88
0.185
0.18
0.175
0.17
0.165
0.16
0.155
0.15
0.145
14211 Predicted
14211 Actual
0.185
0.18
0.175
0.17
0.165
0.16
14181 Predicted
14181 Actual
0.155
0.15
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
Fi g 4. 9 Graph of 14 21 1 Pr edictive mod el.
0.185
0.18
0.175
0.17
0.165
0.16
14191 Predicted
14191 Actual
0.155
0.15
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
Fig 4.11 Graph o f 14191Predictive model.
Fi g 4.10 Graph of 14 18 1 Pred ictive mod el.
0.185
0.18
0.175
0.17
0.165
0.16
14201 Predicted
14201 Actual
0.155
0.15
20/3/ 2013
21/3/ 2013
22/3/ 2013
25/3/ 2013
26/3/ 2013
27/3/ 2013
28/3/ 2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/ 2013
Fi g 4.12 Graph of 14 20 1 Pred ictive mod el.
Table:4.6 Stock Predictions for Guinness Nigeria Plc.
Sample of Daily Stock Price prediction for Guinness Nigeria Plc.
Sample Date
Actual value
Predicted Values with Different Neural Network Predictive Models
14211
14201
14191
14181
14171
14161
14151
14141
20/3/2013
265.00
265.25
264.92
265.43
265.07
265.07
265.35
265.52
265.24
21/3/2013
267.00
267.13
266.86
267.45
266.94
266.96
267.23
267.22
267.29
22/3/2013
265.00
264.88
264.79
264.99
264.65
264.95
265.03
265.28
264.62
25/3/2013
265.00
265.25
265.03
265.64
265.04
264.99
265.01
265.21
265.23
26/3/2013
263.01
262.64
262.68
263.07
262.64
262.79
262.90
263.33
262.50
27/3/2013
263.80
263.84
263.84
264.27
263.78
263.74
263.75
264.09
263.90
28/3/2013
265.00
265.09
264.79
265.38
264.93
265.00
265.23
265.27
265.14
2/4/2013
265.02
265.08
264.78
265.24
264.88
265.03
265.28
265.38
265.03
3/4/2013
266.00
266.05
265.82
266.34
265.86
265.94
266.15
266.24
266.12
4/4/2013
265.00
265.58
265.10
266.03
265.53
265.16
265.67
265.71
265.53
5/4/2013
265.00
265.02
264.87
265.34
264.97
264.95
265.05
265.30
264.95
8/4/2013
265.00
264.96
264.85
265.26
264.81
264.92
265.08
265.30
264.96
9/4/2013
265.00
265.03
264.96
265.28
264.82
264.94
265.02
265.23
264.99
10/4/2013
264.00
263.85
263.85
264.13
263.65
263.87
263.84
264.12
263.74
14141 Actual
14141 Predicted
0.268
0.267
0.266
0.265
0.264
0.263
0.262
0.261
0.26
0.261
0.26
14171 Predicted
14171 Actual
0.268
0.267
0.266
0.265
0.264
0.263
0.262
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
20/3/ 2013
21/3/ 2013
22/3/ 2013
25/3/ 2013
26/3/ 2013
27/3/ 2013
28/3/ 2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/ 2013
Fi g 4.13 Graph of 14 16 1Predictive model.
Fi g 4.15 Graph of 14 15 1 Predictive model.
Fi g 4.14 Graph of 14 17 1 Predictive model.
0.268
0.267
0.266
0.265
0.264
0.263
0.262
0.261
0.26
14161 Predicted
14161 Actual
20/3/2013
21/3/2013
22/3/2013
25/3/2013
26/3/2013
27/3/2013
28/3/2013
2/4/2013
3/4/2013
4/4/2013
5/4/2013
8/4/2013
9/4/2013
10/4/2013
Fi g 4.16 Graph of 14 16 1 Predictive model

DISCUSSION OF RESULTS
The mean squared error was used to describe the performance of the neural network prediction. Table
4.1 shows the performance of Nestle Nigerian Plc. Prediction. The mean square error of the training, validation and testing are 4.6618e6, 3.25520e5 and 4.25777e5 respectively. The values from the predictive model show a minimum amount of error. Table 4.2 shows the predicted values with different neural network predictive model. The table contains the sample data, the actual value, and values of different predictive models. The results from the different predictive models where compared with the actual value and it was observed that 14191 predictive model gave the best prediction for Nestle Nigeria Plc.The result of 14191 predictive model and the actual value are in bold for easy comparison and identification. Figure 4.5, 4.6, 4.7 and 4.8 shows the graphs of 14211, 14201, 14191 and 14181 predictive models respectively. The graphs demonstrate the closeness of the predicted value against the actual value, and it were seen from the graph that the prediction was done with minimum amount of error.
Table 4.4 shows the performance of Total Nigerian Plc. Prediction, the mean square error of the training; validation and testing are 3.07993e6, 4.91378e6 and 8.15871e6 respectively. The values from the predictive model show a minimum amount of error. Table 4.5 shows the predicted values with different neural network predictive model. The table contains the sample data, the actual value, and values of different predictive model. The results from the different predictive models where compared with the actual value and it was observed that 14201 predictive model gave the best prediction for Total Nigeria Plc.The result of 14201 predictive model and the actual value are in bold for easy comparison and identification. Figure 4.9, 4.10, 4.11 and 4.12
shows the graphs of 14211, 14181, 14191 and 14201 predictive models respectively. The graphs demonstrate the closeness of the predicted value against the actual value, and it were seen from the graph that the prediction was done with minimum amount of error.
Table 4.3 shows the performance of Guinness Nigerian Plc. Prediction, the mean square error of the training; validation and testing are 8.29839e7, 1.07882e6 and 2.4488e6 respectively. The values from the predictive model show a minimum amount of error. Table 4.2 shows the predicted values with different neural network predictive model. The table contains the sample data, the actual value, and values of different predictive model. The results from the different predictive models where compared with the actual value and it was observed that 14171 predictive model gave the best prediction for Guinness Nigeria Plc. The result of 14171 predictive model and the actual value are in bold for easy comparison and identification. Figure 4.13, 4.14,
4.15 and 4.16 shows the graphs of 14161, 14171, 14151 and 14161 predictive models respectively. The graphs demonstrate the closeness of the predicted value against the actual value, and it were seen from the graph that the prediction was done with minimum amount of error.

CONCLUSION

The result shows that using hybridize factors input variables gives a better result in stock market prediction with minimum amount of error. It was also observed from the experimental results that optimal prediction can be achieved by varying the number of the hidden neuron. The initialization scheme may be improved by estimating weights between input nodes and hidden nodes, instead of random initialization. Enrichment of more relevant inputs such as fundamental data and technical data from derivative markets may improve the predictability of the network. Applying Neural Network back propagation learning algorithm in training data for stock prediction has been shown in this paper to be an efficient tool for stock prediction.
REFERENCES

A. Adebiyi Ayodele, K. Ayo Charles., O. Adebiyi Marion., and O. Otokiti Sunday, Stock Price Prediction using Neural Network with Hybridized Market Indicators,
Journal of Emerging Trends in Computing and Information Sciences, Vol3 No 1, Pp 19, 2012.

T. Akinwale Adio, O.T. Arogundade and F Adekoya Adebayo. Translated Nigeria Stock Market Price Using Artificial Neural Network for Effective Prediction, Journal of Theoretical and Applied Information TechnologyVol1 No 1. Pp 3643. 2009

N. C. Ashioba, E.O. Nwachukwu, and O. Owolobi. Finding the Optimal Solution for a Transportation Problem Using Neural Network. Microwave, International Journal of Science and Technology, Vol. 3 No 1, Pp 3640, 2012.

KyoungJae Kimand and Ingoo Han, Genetic Algorithms Approach to Feature Discretization in Artificial Neural Networks for the Prediction of Stock Price Index, Institute of Science and Technology, Vol19, .Pp 125132, 2000

Mahdi Pakdaman Naeini, Homa Baradaran Hashemi and Hamidreza Taremian, .Stock Market Value Prediction Using Neural Networks, International Conference on Computer Information Systems and Industrial Management Applications (CISIM), Pp132 136 2010.

Neelimab Budhani, C. K. Jha, and K Sandeep. Budhani,. Applicationof Neural Network In Analysis of Stock Market Prediction, International Journal of Computer Science And Engineering Technology (IJCSET), Vol 3No 4,, Pp6168. Vol. 3 No. 4 , April 2012.

PeiChann Chang and ChenHao Liu,. A Neural Network With A Case Based Dynamic Window for Stock Trading Prediction; Expert Systems With Applications Vol 36,
No3,Pp 68896898, 2008

Qing Cao, B. Karyl. Leggio, and J Marc. Schniederjans, A Comparison Between Fama and French's Model and Artificial Neural Networks In Predicting The Chinese Stock Market, Computers &Operations Research , Vol32. Pp24992512, 2005.

Reza Gharroie Ahangar, Mahmood Yahya Zadehfar and Hassan Pournaghshband. The Comparison of Methods Artificial Neural Network With Linear Regression Using Specific Variance for Predicting Stock Price In Tehran Stock Exchange, International Journal of Science and Information Security, Vol 7 No 2. Pp 3846. .2010.

Salim Lahmiri , Neural Networks and Investor Sentiment Measures for Stock Market Trend Prediction, Journal of Theoretical and Applied Information Technology. ISSN 19928645, Vol 27 No 1, 2011.

ShengHsun Hsu and JJ PoAn Hsieh, A TwoStage Architecture For Stock Price For Recasting By Integrating SelfOrganizing Map and Support Vector Regression, Expert Systems with Applications, Vol36, No 4, Pp 79477951. 2008

YiHsien Wang, Nonlinear Neural Network Forecasting Model for Stock Index X Option Price: Hybrid GJRGARCH Approach, Expert Systems with ApplicationsVol36.,No1, Pp564570. 2007.

Zhang Yudong and Wu Lenan,. Stock Market Prediction of S&P 500 Via Combination of Improved BCO Approach and BP Neural Network, Expert Systems With Applications, Vol 36, No5, Pp88498854. 2008

B. Manjul, S.S.V.N. Sarma, R. Lakshman Naik and
G. Shruthi, Stock Prediction using Neural Network, International Journal Of Advanced Engineering Sciences and Technologies (IJAEST) Vol No. 10, No. 1, Pp 013 018, 2011