A Deep Reinforcement Learning Approach to Automated Stock Trading, using xLSTM Networks

题目： 使用 xLSTM 网络的自动化股票交易深度强化学习方法。
作者： Faezeh Sarlakifar 等（2025）。
摘要要点： 论文提出一个端到端的深度强化学习交易框架，采用扩展的 LSTM（xLSTM）以增强时间序列特征提取。作者在多个股票集合上进行了评测，与传统 RL 基线和监督学习模型比较，在考虑交易成本的仿真下显示出更高的夏普比率和风险调整后收益。
A Deep Reinforcement Learning Approach to Automated Stock Trading, using xLSTM Networks
Authors: Faezeh Sarlakifar et al. — 2025.
Source / arXiv: arXiv:2503.09655.
English summary: This paper proposes an end-to-end deep reinforcement learning framework for automated stock trading that integrates an extended LSTM variant (xLSTM) for richer temporal feature extraction. The authors evaluate the agent on multiple equity universes, compare against classic RL baselines and supervised forecasting models, and report improved Sharpe ratio and risk-adjusted returns under realistic transaction-cost simulations.