The U.S. Stock Market's Short-Term Trends Prediction Using Historic Data

Short-term stock price forecasting is a field in high demand for investors and academics due to increased volatility in the worlds financial markets. It is a tool that could significantly increase investors confidence and help them reduce insecurity when doing financial contraction. JPMorgan Chase stock (JPM) is the subject of this study. This paper uses the stocks data to predict prices in the models, including XGBoost and LightGBM, and by aggregating them with weighted fusion, which may enhance prediction accuracy. Therefore, this study is directed at model training, feature selection, as well as fusion methodology design and improvement. In the end, the outcome of the modelling of the JPM stock price shows that the fusion model can indeed enhance the accuracy of the prediction. The fusion model could aim for prediction via decreasing risk, since backtesting for prediction-based trading strategies is expected to gauge performance in terms of risk and return direction. Empirical research is at the cornerstone of investment decisions within the United States market; for this reason, this research holds promise for real-world applicability.