A survey of recent machine learning techniques for stock prediction methodologies

The prime purpose of the research is to investigate stock price prediction techniques and their shortcomings concerning particular characteristics and performance measures. The research uses performance metric analysis, dataset analysis, and bibliographic analysis to determine the current state of recently published research on financial market prediction. The research examines how well machine learning models predict stock market performance, emphasizing how accuracy, precision, and recall are often used as performance measures. The researchers thoroughly analyzed 24 publications, detailing the data elements that were employed, such as historical datasets and technical indicators, and criticized related studies for frequently omitting the adjusted closing price. The research indicates that since Adj Close captures closing opinions from important market participants, it is essential for precise stock prediction. This research opens the door for further research into feature selection and how it affects prediction accuracy by illuminating how these machine learning models behave when other characteristics are added. Previous research has shown that machine learning methods such as long short-term memory and support vector machines are often used for stock price prediction with some data optimization. The performance metrics that were employed to assess the performance were also examined. The researchers have reported that rather than being regression-based, the most often utilized metrics are classification-based. Performance is also measured via other metrics, such as the Sharpe ratio and accumulated error. The findings will assist financial market researchers in developing creative concepts and selecting the most useful criteria from the data that have been provided.