Investment Portfolio Optimization: Integrating Portfolio Allocation Methods with RNN LSTM

The investment industry is experiencing significant growth, driven by an increased interest from new investors. However, achieving success in investing heavily relies on effective decision-making skills, which can be particularly challenging for newcomers. To tackle this issue, the research proposes an investment decision support system designed to aid investors in optimizing their investment portfolios. The proposed approach integrates traditional mathematical methods with advanced Long-Short Term Memory (LSTM) machine learning techniques. The study demonstrates that the most effective combination involves employing LSTM alongside Post Modern Portfolio Theory (PMPT), resulting in the highest Mean Return of 0.002854, the highest Sortino Ratio of 0.192646, and the highest Ending Equity of 499.973204. Backtesting also reveals that the LSTM prediction, in combination with MPT and PMPT, has the potential to significantly increase portfolio values in just 1 year and 7 months, reaching 500% of the initial investment. Furthermore, incorporating LSTM predictions for future investment instrument prices enhances the performance of existing mathematical methods.