Stock Price Prediction using Artificial Neural Network

Stock Price Prediction using Artificial Neural Network

Chirag Modi

U.G. Student, Dept of Electronics &

Communication Engineering, BTL Institute of Technology, Bangalore.

Shah Khalander Pasha

U.G. Student, Dept of Electronics &

Communication Engineering, BTL Institute of Technology,Bangalore.

Dr. Manju Devi

HOD,

Dept of Electronics & Communication Engineering, BTL Institute of Technology,Bangalore.

Abstract: A stock market is a public market for the trading of company stock. It is an organized set-up with a regulatory body and the members who trade in shares are registered with the stock market and regulatory body SEBI. 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. Prediction provides knowledgeable information regarding the current status of the stock price movement. Thus this can be utilized in decision making for customers in finalizing whether to buy or sell the particular shares of a given stock. Many researches have been carried out for predicting stock market price using various data mining techniques. This work aims at using of Artificial Neural Network techniques to predict the stock price of companies listed under National Stock Exchange (NSE). The past data of the selected stock will be used for building and training the models. The results from the model will be used for comparison with the real data to ascertain the accuracy of the model.

Keywords: Stock prediction, Neural Network, NSE, Multi Layer Perceptron (MLP), MATLAB

INTRODUCTION:

From the beginning of time it has been mans common goal to make his life easier. The prevailing notion in society is that wealth brings comfort and luxury, so it is not surprising that there has been so much work done on ways to predict the markets. Various technical, fundamental, and statistical indicators have been proposed and used with varying results. However, not one technique or combination of techniques has been successful enough. With the development of neural networks, researchers and investors are hoping that the market mysteries can be unraveled. A stock market is a public market for the trading of company stock and derivatives at an agreed price; these are securities listed on a stock exchange as well as those only traded privately. It is an organized set-up with a regulatory body and the members who trade in shares are registered with the stock market and regulatory body SEBI. The stock market is also called the secondary market as it involves trading between two investors. Stock market gets investors together to buy and sell their shares. Share market sets prices according to supply and demand. Stocks that are in demand will increase their price, whereas as stocks that are being heavily sold will decrease their price. Companies

that are permitted to be traded in this market place are called listed companies. 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 evolution of information technology, analyzing stock market data for predicting the future of the stock has become increasingly challenging, important and rewarding. Prediction provides knowledgeable information regarding the current status of the stock price movement. Thus this can be utilized in decision making for customers in finalizing whether to buy or sell the particular shares of a given stock.

Artificial Neural Network (ANN): It is often just called a neural network, is a set of interconnected links that have weights associated with them. The concept of ANN was derived from biological neural networks. Neural networks open up a new foray into the field of making efficient and usable predictions in order to optimize profits. Artificial Neural Networks are being used in numerous areas, as it is an irrefutably effective tool that aids the scientific community in forecasting about probable outcomes. Any ANN can be thought of as a set of interconnected units broadly categorized into three layers. These three layers are the input layer, the hidden layer and the output layer. Inputs are fed into the input layer, and its weighted outputs are passed onto the hidden layer. The neurons in the hidden layer (hidden neurons) are essentially concealed from view. Using additional levels of hidden neurons provides increased flexibility and more accurate processing. However, the flexibility comes at the cost of extra complexity in the training algorithm. Having more hidden neurons than necessary is wasteful, as a less number of neurons would serve our purpose just fine. On the other hand, having less hidden neurons than required would cause reduced robustness of the system, and defeat its very purpose. An illustration of a neural network is shown in Figure 1.

MLP Neural Network: Multi Layer Perceptron (MLP) is a feedforward neural network with one or more layers between input and output layer. MLP maps sets of input

data onto a set of appropriate outputs. Feedforward means that data flows in one direction from input to output layer (forward). An MLP consists of multiple layers of nodes in a directed graph, with each layer fully connected to the next one. Except for the input nodes, each node is a neuron (or processing element) with a nonlinear activation function. This type of network is trained with the backpropagation learning algorithm. MLPs are widely used for pattern classification, recognition, prediction and approximation. Multi Layer Perceptron can solve problems which are not linearly separable. MLPs separate classes via Hyperplanes. MLPs use distributed learning. MLPs have one or more hidden layers.

Fig. 1 MLP Neural Network

PROPOSED METHODOLOGY:

• To study the current stock market trend and collect trend data.

• To build prediction model for the companies listed under NSE using multilayer perceptron (MLP) Neural Network technique.

• To compare the model with real data for its accuracy.

• The past three years data of the companies listed under NSE are used for building, training and testing the prediction model for MLP algorithms.

  Step 1: Data collection – Past data of stock market are collected from various authorized sources. The data of companies listed under NSE, for the duration of 36 months (1-1-2011 to 1-1-2014) have been collected.

  Step 2: Developing MLP neural network algorithm to predict future stock price. With Neural Network Toolbox MATLAB, MLP neural network is built and trained for different combinations of data and parameters as shown in below snapshots.

  Fig. 2 Front end GUI

  • Once the program is run, the GUI appears as shown below.

  • Now, clicking on Create Network in MLP gives the following.

Fig. 3 Create MLP

Fig. 4 MLP Network

Fig. 4 shows MLP Neural Network Model that is built. For different combinations of input values, the model varies. Now, it gives performance curve, error curve and output graph. Here, as shown in above snapshot, Name of the Network can be any name. The other six parameters namely, No. of input Neurons, No. of output Neurons, No. of Hidden Layers, No. of Neurons in each, Learning Rate and Epochs are used for different combinations of values and the result of all these combinations are taken into the consideratio to check which combination gives the more accurate result. After entering all the values, click on create network. Now, the network is created.

Once the network is created, the next step is to train the network.

Clicking on Train Network gives the following.

Fig. 5 Performance Curve MLP

Fig. 6 Error Curve MLP

Fig. 7 Output Curve MLP

For different combinations of data and parameters, this performance curve varies. Training of the model stops either it reaches to mentioned number of epochs (shown in Fig. 3) or when Mean Squared Error (MSE) is almost never improving after certain epochs. The circle in the performance curve shows the best validation performance. Error curve shows how much accurate the network is. For the target input index, error curve shows how much the predicted value varies from the actual value. More the grouping of curve towards 0 (zero), better the network is. Sudden up/down movement in the curve means, predicted value varies significantly from the actual value for that particular input index

Here, actual value is compared with the redacted value. In the output curve as shown in Fig. 7, blue line shows actual value for that particular input index and red line shows predicted value.

RESULTS:

For different combinations of data and parameters, the different results and outputs are taken for all the companies that are listed under NSE.

Performance analysis is the most important part. The numbers here are very deceptive and careful analysis of the same is important.

MSE: Mean Square Error is obtained by squaring the difference of obtained output and the actual output. The main drawback of MSE is that the value of MSE increases with the increase in the stock price. Eg: Assume that a neural network is able to predict the price with 90% accuracy. For stock like Idea cellular whose price is in the range of 150-200 the maximum possible MSE is 400. If the same network is used for Infosys whose price is in the range of 2000 – 3000 the maximum MSE is around 90,000. Hence MSE is a very deceptive number. It is used by the system for training the weights.

MSE various with stock price hence cannot be used for comparison. We need a number which can be compared, for this Normalized MSE is the solution. Normalized MSE is obtained by dividing MSE with the stock price. Correct Direction %: Let us assume that the actual closing price of the stock is 100 Rs even if the predicted price is 99.9999 Rs it is not accounted in the above percentage. Hence correct direction % should alone not be used for the analysis. It is used along with the normalized MSE for comparing neural networks. Standard Deviation is used for identifying the range for the accuracy

Table 1: Result Analysis

of the network. It is used along with Correct Direction %. Thus, Normalized MSE along with Correct Direction % and SD are used for comparing performance of the system.

CONCLUSIONS

Stock market data are highly time-variant and are normally in a nonlinear pattern, predicting the future price of a stock is highly challenging. Prediction provides knowledgeable information regarding the current status of the stock price movement. In the literature review, different data mining techniques for stock market prediction are reviewed. It is noticed that Artificial Neural Network technique is very useful in predicting stock indices as well as stock price of particular company. Many different algorithms have been used with neural network.

Feedforward MLP neural network technique is considered to predict the stock price of companies listed under LIX15 index of NSE. From the result table is can be concluded that MLP neural network technique gives the satisfactory output with

Median Normalized Error 0.05995 Median Correct Direction % 51.06

Median Standard Deviation 6.39825

REFERENCES

1. Darmadi Komo, Chein-I Chang, Hanseok KO, Neural Network Technology for Stock Market Index Prediction, International Symposium on Speech, Image Processing and Neural Networks, 13-16 April 1994

2. D. Venuopal Setty, T.M.Rangaswamy and K.N.Subramanya, A Review on Data Mining Applications to the Performance of Stock Market, International Journal of Computer Applications, (0975

   8887) Volume 1 No. 3, 2010

3. Dase R.K. and Pawar D.D., Application of Artificial Neural Network for stock market predictions: A review of literature, International Journal of Machine Intelligence, ISSN: 09752927,

   Volume 2, Issue 2, pp-14-17, 2010

4. Akhter Mohiuddin Rather, A prediction based approach for stock returns using autoregressive neural networks, IEEE, 978-1- 4673-0126-8, 2011

5. D. Ashok Kumar and S. Murugan, Performance Analysis of Indian Stock Market Index using Neural Network Time Series Model, International Conference on Pattern Recognition, Informatics and Mobile Engineering (PRIME), IEEE, 978-1-

   4673-5845-3, 2013

6. Aditya Nawani, Himanshu Gupta, Narina Thakur, Prediction of Market Capital for Trading Firms through Data Mining Techniques, International Journal of Computer Applications (0975 8887) Volume 70 No.18, May 2013

7. Chi Kin Chow, Tong Lee, Construction of multi-layer feed forward binary neural network by a genetic algorithm Neural Networks, IJCNN Proceedings of the 2002 International Joint Conference, Honolulu, HI, USA, 2002

8. Cao Q, Leggio KB, Schniederjans Mj., A comparison between Fama Frenchs model and artificial neural network s in predicting the Chinese stock market, Computers & Operations Research,

   Vol. 32, pp. 2499-2512, 2005

9. [9] Satyajit Dhar, Tuhin Mukherjee, Arnab Kumar Ghoshal, Performance Evaluation of Neural Network Approach in Financial Prediction: Evidence from Indian Market, Proceedings of the International Conference on Communication and Computational Intelligence, pp.597-602, 2010

10. Gitansh Khirbat, Rahul Gupta, Sanjay Singh, Optimal Neural Network Architecture for Stock Market Forecasting, International Conference on Communication Systems and Network Technologies, IEEE, 978-0-7695-4958-3, 2013

11. Binoy B. Nair, M. Patturajan, V.P Mohandas, Sreenivasan R.R, Predicting the BSE Sensex: Performance Comparison of Adaptive Linear Element, Feed forward and Time Delay Neural Networks, IEEE, 978-1-4673-0449-8, 2012

12. Kumar Abhishek, Anshul Khairwa, Tej Pratap, Surya Prakash, A Stock Market Prediction Model using Artificial Neural Network,

13. J. G. Agrawal, Dr. V. S. Chourasia, Dr. A. K. Mittra, State-of- the-Art in Stock Prediction Techniques, International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering, Vol. 2, Issue 4, April 2013

14. Pratap Kishore Padhiary and Ambika Prasad Mishra, Development of Improved Artificial Neural Network Model for Stock Market Prediction, International Journal of Engineering Science and Technology (IJEST), Vol. 3 No. 2, Feb 2011

    ABOUT THE AUTHOR:

    1. Chirag Modi:

       Born on 23rd Sep 1994 in Rajasthan, currently pursuing Bachelors degree in Electronics and Communication (2016) from Visveshwaraya Technological University, Karnataka.

    2. Shah Khalander Pasha:

       Born on 8th Sep 1994 in Karnataka, currently pursuing Bachelors degree in Electronics and Communication (2016) from Visveshwaraya Technological Uniiversity, Karnataka.

    3. Dr.Manju Devi:

Born on 10th Dec 1974 in Uttar Pradesh, obtained her B.E degree(1996) in Electronics and Communication Engineering from Anna University, Chennai and M.Tech degree(2004) in Applied Electronics from Visvesvaraya Technological University (VTU), Karnataka. She has completed her Ph.D degree from VTU in the field of analog and mixed mode VLSI. At present working as Professor & head in the department of ECE, Vice Principal at BTLIT, Blore. Almost eighteen years of teaching experience in engineering colleges. Her areas of interest are VLSI design, analog and mixed mode VLSI design and digital electronics.