Article Preview
TopIntroduction
The trading pool involving the buying or selling of shares is referred to as a stock market. Here, the dealers make profits if the trading goes in their favor (Pun & Shahi, 2018). This market is a means for raising capital for the firms involved (Billah, Waheed, & Hanifa, 2017) and impacting the global economy where it is currently revolving. Publicly listed companies and businesses thrive in the financial markets while analyzing the trends either by profits generated or losses incurred in the light of stocks, foreign exchanges, and bonds to market indicators. Share or stakeholders find the financial markets fascinating as returns and risk could be too high. This makes research in this domain necessary to equip the business owners and investors with the needed information (Samarawickrama & Fernando, 2018). Investors crave insights about the stock market to decide whether to buy or sell portions of stocks and look to swell profit on investment capital. However, forecasting stock prices can be problematic since it is highly volatile with factors ranging from the global economy, events, politics to the investor(s) involved (Oncharoen & Vateekul, 2018). According to Hoseinzade & Haratizadeh (2019), financial markets are presumed a vital part of the world's economy. An economy's growth can be triggered by stocks depending on how well it is doing. The insights will, therefore, be helpful to businesses that revolve around the market's performance.
In stock price forecasting, the objective is to foresee an organization's financial stocks' future estimation. A correct prediction of stocks can prompt enormous benefits for the brokers involved. Now and again, estimating draws out the thought that it is noisy instead of arbitrary (Parmar et al., 2018), and predicting future trends can minimize investment risk (Long, Chen, He, Wu, & Ren, 2020). Predictions on stock prices have been the object of studies for many decades. However, due to its chaoticness and dynamism, studies have concluded that forecasting stock price is difficult (Nelson, Pereira, & Oliveira, 2017). Typically, traders employ technical and fundamental analysis to predict stocks (Singh & Srivastava, 2017). However, artificial intelligence (AI) has been a proficient method to incorporate such procedures. Its presentation in the stock forecast zone has captivated numerous kinds of research due to its dynamic and exact estimation showcased (Parmar et al., 2018). Despite that, the crucial piece of machine learning is the dataset utilized. The dataset is expected to be clean since a minute distortion of the data can negatively influence the outcome and render the predictions inaccurate (Parmar et al., 2018). To aid prospective owners of shares in making appropriate decisions, predicting changes in stock prices in the future can be done by studying the patterns of an earlier time (Kumar, Dogra, Utreja, & Yadav, 2018). One of the methods with the potential to resolve this problem is the Neural Network (NN). NN's design mimics how the human mind processes information; it is one of the information systems for solving a problem (Prastyo, Junaedi, & Sulistiyo, 2017). Neural network and machine learning techniques are suitable for the projection of tremendously volatile time series data with strong noise, non-linearity, and temporary correlation (Li, Bu, Li, & Wu, 2020). Conventional methods adopted for predicting stock prices, such as fundamental and technical analysis, are a constraint since they cannot learn the previous data dynamics in detail (Selvin, Vinayakumar, Gopalakrishnan, Menon, & Soman, 2017). Entire life savings could be lost as the stock market's dynamic and chaotic attribute makes the forecasting a gamble (Nayak et al., 2016; Misra & Chaurasia, 2019; Livieris, Pintelas, & Pintelas, 2020). Another technique that will be of interest is Deep learning.