Meta-Learning Reinforcement Learning for Crypto-Return Prediction

Junqiao Wang, Zhaoyang Guan, Guanyu Liu, Tianze Xia, Xianzhi Li, Shuo Yin, Xinyuan Song Chuhan Cheng, Tianyu Shi, Alex Lee · 2025

Read Meta-Learning Reinforcement Learning for Crypto-Return Prediction by Junqiao Wang, Zhaoyang Guan, Guanyu Liu, Tianze Xia, Xianzhi Li, Shuo Yin, Xinyuan Song Chuhan Cheng, Tianyu Shi, Alex Lee (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-09751.txt.

Description

Junqiao Wang and 9 co-authors present Meta-RL-Crypto in a 2025 paper, a system meant to predict cryptocurrency returns and trade on those predictions. Crypto prices shift fast, driven by on-chain activity, news, and social sentiment, while labeled training data are scarce. Their design joins meta-learning with reinforcement learning inside one transformer-based agent that needs no extra human supervision.

The agent rotates through three roles, acting, judging its own actions, then judging the judging, in a closed loop. Market inputs in several forms feed the loop, and the agent keeps refining both its trading policy and its scoring rules. Across varied market conditions the authors report stronger results on real market indicators than other language model based trading baselines.

Catalog record

CREATORJunqiao Wang, Zhaoyang Guan, Guanyu Liu, Tianze Xia, Xianzhi Li, Shuo Yin, Xinyuan Song Chuhan Cheng, Tianyu Shi, Alex Lee
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-09751

Browse the full library · About the Tower of Babel