Deep learning-based predictive models for forex market trends: Practical implementation and performance evaluation.

Area of Science:

• * Computational Finance
• * Machine Learning Applications
• * Financial Market Analysis

Background:

• * Growing interest in financial market trend prediction for real-world applications.
• * Foreign Exchange (Forex) market complexity simplified to binary classification.
• * Need for practical deep learning models in Forex trading.

Purpose of the Study:

• * Propose practical deep learning-based predictive models for Forex trading.
• * Enhance trader decision-making by minimizing losses and anticipating risks.
• * Emphasize profit return as a key evaluation metric over accuracy.

Main Methods:

• * Implementation of deep learning models for Forex market prediction.
• * Utilization of realistic Yahoo Finance datasets for experimental validation.
• * Comparison of traditional two-value labels with proposed three-value labels.

Main Results:

• * Demonstrated effectiveness of implemented deep learning prediction mechanisms.
• * Validation through extensive experimental studies on realistic financial data.
• * Superior performance observed with three-value labels compared to two-value labels.

Conclusions:

• * Deep learning models offer practical solutions for Forex market prediction.
• * Prioritizing profit return is crucial for evaluating trading models.
• * Three-value labels enhance prediction accuracy and optimize trading strategies.