A Decision-Making Framework for Financial Trading Using Linear Tree-Based Kernel Support Vector Machine Classifier

In today's rapidly evolving financial landscape, numerous investors and hedge funds have gravitated towards stock markets, captivated by their extensive range of instruments and lucrative investment opportunities. Despite their allure, navigating the stock market landscape is intricate. It demands financial traders to remain vigilant, identifying thriving firms and diligently tracking stock price oscillations to devise potent trading strategies. In recent times, cutting-edge technological advancements like artificial intelligence and machine learning have revolutionized stock price predictions. Machine learning (ML) algorithms, especially Support Vector Machines (SVM), have become instrumental in predicting stock movements. However, an often-underacknowledged limitation of SVM is its susceptibility to overfit in the presence of noisy and intricate datasets. Addressing this gap, our research introduces the innovative Linear Tree-based Kernel Support Vector Machine (LT-KSVM) for enhanced stock market price forecasts. We empirically validated our approach by creating a portfolio of stocks from the Saudi Stock Exchange (Tadawul). Simulated trades over a span of 3.7 years were deployed to assess the portfolio's predicted performance. We meticulously executed three preprocessing phases to enhance stock price movement forecasts. Remarkably, our proposed framework demonstrated astounding potential, boasting over 90% returns.