A review and analysis of deep learning methods for stock market prediction with variety of indicators

Abstract

The stock market is an open forum in which stocks may be exchanged, bought, and sold on a stock exchange or over-the-counter to gain profits.  However, the stock market may be impacted by a variety of variables like business cycle, economic conditions, changing political and government policies, market volatility and so on, resulting in lesser forecast possibilities. Statistical models have difficulty predicting the stock market because of their alternating nature and relying only on previous data does not give optimal future results.  In recent years, Deep Learning (DL) models have gained attention in stock market forecasting.  DL models accurately estimate both past and current data to improve prediction outcomes. Various DL models are explored in this study and their application to stock market prediction, specifically how past patterns can be used to anticipate future trends. DL models are able to reveal previously unseen patterns and connections by using massive datasets, including market prices, transaction amounts, news mood, and macroeconomic indicators. Furthermore, the article compares and contrasts a number of DL algorithms that are part of stock market prediction systems, outlining their advantages and disadvantages. At last, possible improvisations are made to improve the effectiveness and precision of stock market forecasting.