Key points are not available for this paper at this time.
The challenging yet traditional field of stock market forecasting has garnered significant interest from data scientists and economists. Given the high risks associated with trading, where investors can lose part or all of their initial capital, there is a growing need for advanced methods to support investment decisions. By leveraging the capability to predict future stock performance patterns by machine learning or AI, the return on investment for short-term trading can potentially be increased. This research reviews the structures of various AI-based machine learning techniques with a case study of Apple stock data spanning from 2010 to 2021 and the effectiveness of the algorithms in terms of various real-time factors. Among several machine learning algorithms, the SVM algorithm is shown to be the most accurate, with an accuracy rate of 76% for the case study. However, the effectiveness of machine learning systems in predicting stock prices is influenced by several factors. The inherent volatility and unpredictability of financial markets make it challenging for models to anticipate sudden, unforeseen events. Overfitting is a common issue where models that perform well on historical data fail to generalize to new, unseen data because they capture noise instead of genuine market trends. Data quality issues, such as noisy or incomplete data, further complicate predictions. Additionally, the complexity of machine learning models can hinder interpretability, making it easier for analysts to trust and utilize their outcomes. This research underscores the necessity for careful and ongoing refinement of ML techniques in stock price forecasting.
Islam et al. (Mon,) studied this question.