arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.02300eess.SYcs.SYmath.OCq-fin.CPq-fin.MF
收藏

从半无限约束到结构化鲁棒策略:金融系统的最优增益选择

From Semi-Infinite Constraints to Structured Robust Policies: Optimal Gain Selection for Financial Systems

Chung-Han Hsieh

AI总结:

针对金融交易中不确定性下的增益选择,本文将半无限鲁棒优化转化为平衡与互补两类结构化策略,并提出图解高效求解方法,实证显示其风险调整收益更优、下行风险更低。

摘要:

本文研究金融交易系统的鲁棒最优增益选择问题,该问题在 double linear policy(双线性策略)框架内构建,用于在多头与空头头寸之间配置资本。核心目标是在一系列不确定市场条件下统一保证 robust positive expected(RPE,鲁棒正期望)利润,同时确保风险控制。该问题引出一个带有半无限约束的鲁棒优化表述,其中不确定性由可能收益参数的有界集合建模。我们通过将半无限约束转化为结构化策略来解决该问题,即 balanced policy(平衡策略)和 complementary policy(互补策略),它们能够显式刻画最优解。此外,我们提出一种新颖的图解方法,以高效求解鲁棒增益选择问题,显著降低计算复杂度。基于历史股票价格数据的实证验证表明,与传统策略相比,该方法在风险调整后收益和下行风险方面表现出优越性能。该框架通过纳入鲁棒性考量,推广了经典均值—方差优化,为不确定性下的鲁棒交易提供了系统且高效的解决方案。

英文摘要:

This paper studies the robust optimal gain selection problem for financial trading systems, formulated within a \emph{double linear policy} framework, which allocates capital across long and short positions. The key objective is to guarantee \emph{robust positive expected} (RPE) profits uniformly across a range of uncertain market conditions while ensuring risk control. This problem leads to a robust optimization formulation with \emph{semi-infinite} constraints, where the uncertainty is modeled by a bounded set of possible return parameters. We address this by transforming semi-infinite constraints into structured policies -- the \emph{balanced} policy and the \emph{complementary} policy -- which enable explicit characterization of the optimal solution. Additionally, we propose a novel graphical approach to efficiently solve the robust gain selection problem, drastically reducing computational complexity. Empirical validation on historical stock price data demonstrates superior performance in terms of risk-adjusted returns and downside risk compared to conventional strategies. This framework generalizes classical mean-variance optimization by incorporating robustness considerations, offering a systematic and efficient solution for robust trading under uncertainty.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑