Improving the Performance of Stock Price Prediction

Abstract

This research analyzes the importance of accurate stock price forecasting within the global financial markets, specifically emphasizing the growing use of AI methods such as deep and machine learning. The advancement of AI technologies has been accelerated by the limitations imposed by conventional finance models in effectively capturing the complex dynamics of markets, hence aiming to enhance the reliability of forecasts. The current investigation utilized the random forest (RF), XGBoost, and stacked generalization forecasting algorithms to examine a dataset spanning from June 1, 2004 to November 16, 2023 with a particular emphasis on the NASDAQ index. The stacked generalization approach exhibited superior performance compared to both models, with lower error coefficients. This outcome implies an improved predictive capability and reduced bias. Overall, the outcomes of the examination revealed that the stacked generalization model has remarkable efficacy in forecasting stock values, surpassing the RF and XGBoost models in terms of performance.