Article Preview
Top1. Introduction
Stock price prediction is one of the interesting fields of research in financial world. Since, the stock data are nonlinear, noisy, complicated, dynamic and non-parametric in nature (YakupKara et al., 2011). The ultimate goal of stock market price prediction is to develop a powerful model or system for predicting the stock value to gain high profits. Over the past decades, lots of research works have been devoted to the development and enhancement of stock market price prediction models. One of the most popular and widely used models is the ANN model. ANN is a powerful tool for stock market prediction due to its capability of mapping non-linear input-output data (Awad et al., 2009; Handa et al., 2015; Sharma et al., 2015).
Although soft computing models (ANN) can be a promising statistical tool for stock market prediction, several research studies showed that neural networks had some drawbacks in training the input data due to non-stationary characteristics of stock data (Hiransha et al., 2018; Khare et al., 2017; Mohapatra et al., 2012). To address this issue, many researchers and analysists have attempted to develop more and more efficient models by combining ANN with bioinspired computing algorithms. Some studies proved that the integrated model, ANN with bio inspired algorithms outperform over traditional models such as RBF-GA (Sheta & DeJong, 2001), ANN-FSA (Shen et al., 2011), SVM-PSO (Mohammed et al., 2018; AlaaF et al., 2015), ANN-BFO and ANN-ABFO (Essam, 2016), ANN-PSOCoM (Majhi et al., 2009) and MLP-BBO (Hesam & Mahsa, 2018).
The core objective of this study is to predict the stock price employing ANN and bio inspired computing algorithms and to evaluate them with some statistical metrics. Instead of finding the optimal weights of FFNN by trial and error method, PSO and BBO algorithm is proposed for the FFNN. Ten technical indicators obtained from the historical data are utilized as the inputs for the proposed prediction model. The main focus of this study is also to show and prove the predictability of stock price using soft computing and bio inspired computing algorithms and to compare the efficiency of these models. In addition to this, proposed prediction model is compared with some other existing models to demonstrate its efficiency.
The remainder of the paper is structured as follows: Section 2 provides a brief review of related work in stock market predictions. Section 3 explains the functioning of FFNN followed by the details of PSO and BBO algorithms. Section 4 describes the detailed steps of proposed prediction model. Section 5 reports the empirical findings obtained from the comparative analysis. Finally, Section 6 concludes the contribution of this study followed by relevant references.