A Machine Learning Framework for Stock Trading: Integrating Technical and Economic Indicators

Abstract

The increasing complexity of financial markets has driven the demand for intelligent trading systems capable of making accurate and timely decisions. This study presents a machine learning framework for stock trading that integrates both technical and economic indicators to enhance predictive performance and trading profitability. Utilizing a combination of Random Forest, XGBoost, and Long Short-Term Memory (LSTM) models, the research evaluates the effectiveness of these algorithms in forecasting stock price movements. Among them, the LSTM model demonstrated superior accuracy (71%) and a higher Area Under the Curve (AUC-ROC = 0.77), owing to its ability to capture temporal dependencies in financial time series data. The hybrid feature set included commonly used technical indicators (e.g., RSI, MACD, moving averages) and key macroeconomic variables (e.g., GDP growth, interest rates, CPI), improving the model’s adaptability to both market volatility and economic cycles. Backtesting over a three-year period revealed that the machine learning-based strategy yielded a cumulative return of 32.4% with a Sharpe ratio of 1.45, significantly outperforming traditional approaches like moving average crossover and buy-and-hold. Statistical validation through paired t-tests confirmed the significance of these results (p < 0.001). The findings underscore the potential of integrated machine learning approaches to offer more informed, resilient, and profitable trading strategies. This research contributes to the evolving field of AI in finance by offering a data-driven, interpretable, and scalable solution for stock market prediction and investment decision-making.