Scientists Develop AI-Powered Crypto Portfolio Management Tool Trained on On-Chain Data

A team of scientists from the University of Tsukuba in Japan has developed a groundbreaking artificial intelligence (AI) system for managing cryptocurrency portfolios. The system, known as CryptoRLPM (Cryptocurrency reinforcement learning portfolio manager), leverages on-chain data to train its AI model, making it the first of its kind in the field.

CryptoRLPM utilizes a technique called reinforcement learning (RL), which involves the AI system interacting with its environment—the cryptocurrency portfolio—and updating its training based on reward signals. This feedback-driven approach is integrated throughout the system’s architecture, comprising five primary units: data feed, data refinement, portfolio agent, live trading, and agent updating.

To evaluate CryptoRLPM’s performance, the researchers conducted experiments using three different portfolios over a two-year period, from October 2020 to September 2022. The portfolios consisted of various combinations of Bitcoin (BTC), Storj (STORJ), Bluzelle (BLZ), and Chainlink (LINK).

The scientists compared the system’s performance against a baseline evaluation of standard market performance using three key metrics: accumulated rate of return (AAR), daily rate of return (DRR), and Sortino ratio (SR). AAR and DRR provide insights into an asset’s gains or losses within a specific time frame, while the SR measures the asset’s risk-adjusted return.

According to the research paper, CryptoRLPM demonstrated significant improvements over the baseline performance, including an 83.14% increase in AAR, a 0.5603% improvement in DRR, and a 2.1767 improvement in SR compared to the baseline Bitcoin portfolio.

The successful development and performance of CryptoRLPM showcase the potential of AI-powered tools in optimizing cryptocurrency portfolio management strategies. However, further research and real-world testing are needed to validate and refine the system’s capabilities.

For more news, find me on Twitter or subscribe to my YouTube channel.

What is your opinion on this issue? Leave me your comment below! I’m always interested in your opinion!

Leave a Reply

Your email address will not be published. Required fields are marked *

Recommended for you