Article Preview
Top2.Literature Survey
Atsalakis et al. (Atsalakis & Valavanis, 2009; Zopounidis et al., 2013), have defined three major components for the accurate prediction of stock prices. These components include: Fundamental analysis, Technical analysis and the choice of appropriate technological methods. Fundamental Analysis includes the study of various parameters indicating the performance and credibility of the company over a period of time. Technical analysis is based on Dow Theory and uses price history for prediction. It is a form of a time series analysis, which determines the future price of a stock on the basis of the past price, exponential moving average (EMA), oscillators, support and resistance levels or momentum and volume indicators (Agrawal et al., 2019; Teixeira & De Oliveira, 2010).
The choice of appropriate technological methods for stock price forecasting has always been a challenging task. The Artificial Neural Networks (ANN), with their ability to learn, have always attracted the data scientists to use them in prediction problems with Stock Market being no exception (Shah, 2019; Wang, 2011). With the time, the machine learning techniques have improved and it has been possible to use multiple hidden layers or to cascade various layers, as in case of Convolutional Neural Network (CNN) making it feasible to extract the factual relationships which are otherwise hidden from the normal human analysis.
Singh et al. (Singh & Srivastava, 2016), have used Deep Neural Network algorithm to gain future price information. The performance was evaluated on Google stock price multimedia data (chart) from NASDAQ (National Association of Securities Dealers Automated Quotations exchange) and demonstrated that deep learning can improve stock market forecasting accuracy.
Guresen et al.(2011), analyzed the performance of the MLP (Multilayer perceptron), DAN2 (Dynamic Architecture for Artificial Neural Network, Hybrid Neural Network models for obtaining accurate prediction results. Studies have been mostly preoccupied with forecasting volatilities.