arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

金融AI的应急暴露路由:中断风险与不可分割决策的成本

Contingent Exposure Routing for Financial AI: Outage Risk and the Cost of Indivisible Decisions

Shivam Gupta

arXiv 2610.00239首次发表:更新:

AI 中文总结

针对金融AI故障切换中的决策错误分担问题,提出基于市场影响矩阵的应急路由方法,推导不可分割决策的风险界限,实验显示风险降低达10.53%,提供可审计的压力测试框架。

AI 中文摘要

模型故障切换恢复了可用性,但改变了哪些金融机构分担决策错误。我们通过局部市场影响响应矩阵构建了中断应急路由,并研究了期望平方价格位移。一个对称构造表明,随着主端点数目的增长,共享备份可能留下一个一阶集中下限,而平衡的故障转移风险则随幸存端点数量的增加而反向减少。对于不可分割的决策,我们推导了独立随机路由的精确二阶矩,以及决定其与分数分配差距的有效暴露粒度。条件期望舍入给出了一个无需耦合配额的有限智能体界限;一个独立的交换程序保留了端点数量,并针对双重下界进行了评估。在60个合成投资组合网络和11,340个场景评估中,后者在中心反馈设置且独立误差下,对单端点和双端点移除分别降低了6.57%和10.53%的风险。对1,024条记录的API响应在构造的再平衡任务上进行回放,给出了较小的留出集降低3.30%(配对自助区间2.07--4.57%)。强对齐误差、劣质端点和不可分割性限制了多样化。贡献是一个可审计的路由压力测试和实现分析,而非对真实市场崩溃概率的估计。

英文摘要

Model failover restores availability, but changes which financial institutions share decision errors. We formulate outage-contingent routing through a local market-impact response matrix and study expected squared price displacement. A symmetric construction shows that a shared backup can leave an order-one concentration floor as the number of primary endpoints grows, while balanced fallback risk decreases inversely with the surviving endpoint count. For indivisible decisions, we derive the exact second moment of independent randomized routing and an effective-exposure granularity that determines its gap from fractional allocation. Conditional-expectation rounding gives a finite-agent bound without coupled quotas; a separate swap procedure preserves endpoint counts and is assessed against dual lower bounds. Across 60 synthetic portfolio networks and 11,340 scenario evaluations, the latter reduces risk by 6.57% and 10.53% for single and double endpoint removals at the central feedback setting with independent errors. A replay of 1,024 recorded API responses on constructed rebalancing tasks gives a smaller held-out reduction of 3.30% (paired bootstrap interval 2.07--4.57%). Strongly aligned errors, inferior endpoints, and indivisibility limit diversification. The contribution is an auditable routing stress test and implementation analysis, not an estimate of real-market crash probabilities.

Comments15 pages, 6 figures, 3 tables. Code and data: https://github.com/shi1720/contingent-exposure-routing

论文原文

arXiv 摘要页 · PDF 原文 · HTML 原文

↑