- Convolutional Recurrent Neural Networks (CRNN) outperform other models in maximizing cumulative returns, while Transformers and Self-Attention models excel in risk-adjusted returns and drawdown minimization.
- Energy and Health sectors show strong upward momentum with favorable risk-adjusted returns, making them ideal for long-term Buy & Hold strategies. Financial, Consumer Non-Cyclical, and Industrial sectors exhibit mixed performance with periods of recovery and volatility, while Basic Materials, Consumer Cyclical, Property, and Technology sectors experience higher drawdowns and market fluctuations, requiring active monitoring and dynamic strategies.
- Machine Learning (ML)-based technical analysis reacts more quickly to short-term price movements, generating frequent buy and sell signals. Market-driven strategies rely on fundamental analysis and macroeconomic trends, offering a more stable long-term approach to portfolio management.
- To maximize investment performance, future research should focus on: 1. Real-time model updates – ensuring adaptability in volatile markets. 2. Sentiment Analysis Integration – leveraging social media and financial news. 3. Hybrid AI models – combining reinforcement learning with ANN for dynamic decision-making. Integrating sentiment analysis from news and social media can enhance predictive accuracy by capturing broader market sentiment.
- Combining reinforcement learning with ANNs could optimize trading strategies by adapting to market trends dynamically. Regular model recalibration with updated financial data is essential to ensure consistent investment performance and risk management.
Toward Stronger Financial Industry in Indonesia