Evolving deep fuzzy ensemble network for portfolio management
This project aims to design a robust portfolio management system using evolving fuzzy Ensemble Transformer, modified technical indicators and Reinforced Learning. The ensemble of transformer models will predict future stock prices, and explain how predictions are made in the form of if-then logic. N...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/175295 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | This project aims to design a robust portfolio management system using evolving fuzzy Ensemble Transformer, modified technical indicators and Reinforced Learning. The ensemble of transformer models will predict future stock prices, and explain how predictions are made in the form of if-then logic. Next, price forecasts are used to calculate modified technical signals that identifies trend reversal points. Reinforcement Learning optimizes the returns and compensates for delays in trend reversal prediction. The system is applied to a carefully constructed portfolio for both allocation and dynamic rebalancing, and its performance is benchmarked against other popular trading strategies. The system is evaluated on its returns and robustness in changing market conditions. |
|---|