Optimization of investment strategies through machine learning

Area of Science:

• Quantitative Finance
• Machine Learning Applications
• Investment Strategy Optimization

Background:

• Traditional investment models often lack adaptability to market fluctuations.
• Integrating financial metrics with advanced algorithms can enhance stock selection accuracy.
• Economic Value-Added (EVA) offers a robust framework for fundamental stock appraisal.

Purpose of the Study:

• To develop a sustainable quantitative investing model integrating Machine Learning (ML) and Economic Value-Added (EVA).
• To optimize investment strategies through advanced stock selection and algorithmic trading.
• To evaluate the model's performance in diverse market conditions.

Main Methods:

• Quantitative stock selection using Principal Component Analysis (PCA) and EVA criteria.
• Algorithmic trading employing ML techniques: Moving Average Convergence, Stochastic Indicators, and Long-Short Term Memory (LSTM).
• Model validation on the United States stock market data.

Main Results:

• LSTM networks achieved higher accuracy in forecasting future stock values.
• The proposed investment strategy proved feasible across various market situations.
• The model generated returns significantly exceeding market benchmarks.

Conclusions:

• The combined ML and EVA approach offers a novel and effective method for stock appraisal and selection.
• The developed model provides a realistic and valuable tool for investors seeking substantial returns.
• This strategy promotes rational investing and can contribute to market efficiency.