What question did this study set out to answer?

To develop an intelligent portfolio management system leveraging multimodal machine learning and reinforcement learning techniques.

April 10, 2026Open Access

Portfolio Management Application based on Ranking and Reinforcement Learning

View Full Paper

Ask AI

Bookmark

Authors

Thesis demonstrates an algorithmic approach that optimizes portfolio management using ranking and reinforcement learning methods.

To develop an intelligent portfolio management system leveraging multimodal machine learning and reinforcement learning techniques.
Utilized a ranking algorithm with technical and fundamental indicators to select stocks monthly.
Applied the XGBoost ranking algorithm to categorize S&P 500 stocks into ten quantiles based on predicted returns.
Constructed a filtered portfolio ensuring diversification with sector constraints.
Employed reinforcement learning algorithms (A2C, DDPG, PPO, TD3, SAC) for dynamic capital allocation.
Compared the performance of different reinforcement learning algorithms through backtesting.
The proposed framework outperforms both the S&P 500 index and traditional ranking models.
Demonstrated robust and adaptive policies by the agent in capital allocation.
Achieved effective investment strategies that respond to market trends and news.

Φερδινάντος Φ. Κόττας (2025) studied this question.

View Full Paper

Ask AI

Bookmark

April 10, 2026Open Access

View Full Paper

Ask AI

Bookmark

Authors

Thesis demonstrates an algorithmic approach that optimizes portfolio management using ranking and reinforcement learning methods.

To develop an intelligent portfolio management system leveraging multimodal machine learning and reinforcement learning techniques.
Utilized a ranking algorithm with technical and fundamental indicators to select stocks monthly.
Applied the XGBoost ranking algorithm to categorize S&P 500 stocks into ten quantiles based on predicted returns.
Constructed a filtered portfolio ensuring diversification with sector constraints.
Employed reinforcement learning algorithms (A2C, DDPG, PPO, TD3, SAC) for dynamic capital allocation.
Compared the performance of different reinforcement learning algorithms through backtesting.
The proposed framework outperforms both the S&P 500 index and traditional ranking models.
Demonstrated robust and adaptive policies by the agent in capital allocation.
Achieved effective investment strategies that respond to market trends and news.

Φερδινάντος Φ. Κόττας (2025) studied this question.

View Full Paper

Ask AI

Bookmark