arXiv：cs.AI· Gennadiy Chistiakov, Yury Yanovich·· 6 小时前
模块化金融强化学习中的信用分配：等量信息评估框架
Equal-information evaluation of credit assignment in modular financial reinforcement learning
AI 导读
研究者提出一套等量信息评估框架，结合等量信息重构对照、固定效用比较与预设下游护栏，用于评估模块化金融强化学习中的信用分配，并在归因引导的模块化交易架构上开展三项对照研究。结果显示，基于残差的分配在披露接口下与普通损失预算控制完全一致；直接缺口监督提升了组件预测却未能保持解码结果；组件级 critic 监督在固定效用下相较标量监督恶化了风险条件确定性等价与回撤。
正文
Abstract:Credit assignment in modular financial reinforcement learning is difficult to evaluate because an attributed signal may improve a component loss without adding decision-relevant information or improving economic outcomes. We propose an evaluation framework combining equal-information reconstruction controls, fixed-utility comparisons and prespecified downstream guardrails. The framework is instantiated in an attribution-guided modular trading architecture and tested in three controlled studies covering update allocation, supervision-target representation and portfolio risk control. We find that residual-based allocation is exactly reproduced by an ordinary loss-budget control under the disclosed interface; direct shortfall supervision improves component prediction but fails decoded-outcome preservation; and component-wise critic supervision worsens risk-condition certainty equivalent and drawdown relative to scalar supervision under fixed utility. These results identify three distinct evaluation failures: redundant attribution signals, representation-induced prediction trade-offs, and misalignment between component supervision and downstream risk outcomes. A preserved historical engineering case further shows why module-level learning claims require run-level provenance. The contribution is a reusable methodology for evaluating credit assignment in expert systems operating under partial observability, delayed feedback and economic risk constraints.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.14750 [cs.AI] |
| (or arXiv:2610.14750v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.14750 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yury Yanovich [view email]
[v1]
Sat, 10 Oct 2026 09:14:23 UTC (2,193 KB)
来源：arXiv：cs.AI · arxiv.org