Scholarlink

<p>This paper studies an approximate method for computing optimal hedging strategies in the mean-variance hedging problem with model risk. Matsumoto and Suyama (<i>Applied Mathematical Finance,</i> <i>31</i>(6), 365–384, 2024) used deep learning to compute mean-variance hedging strategies under model risk, but encountered increasing computational complexity as the number of periods increased. To address this issue, we propose a new approximate method within deep learning that reduces the computational complexity. We perform numerical validation to demonstrate the effectiveness of our approximation by analyzing hedging strategies and hedging errors over shorter periods. We also highlight the necessity of performing approximate computations as the number of periods increases. Finally, we examine methods for adjusting delta-hedging strategies in the continuous-time model, confirming the validity of the proposed approximate method.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Approximate Method for Mean-Variance Hedging Strategy with Model Risk

  • 期刊论文

  • 发表日期：

    2025年12月10日

查看全文
  • Koichi Matsumoto,
  • Tatsuhiko Suyama

摘要

This paper studies an approximate method for computing optimal hedging strategies in the mean-variance hedging problem with model risk. Matsumoto and Suyama (Applied Mathematical Finance, 31(6), 365–384, 2024) used deep learning to compute mean-variance hedging strategies under model risk, but encountered increasing computational complexity as the number of periods increased. To address this issue, we propose a new approximate method within deep learning that reduces the computational complexity. We perform numerical validation to demonstrate the effectiveness of our approximation by analyzing hedging strategies and hedging errors over shorter periods. We also highlight the necessity of performing approximate computations as the number of periods increases. Finally, we examine methods for adjusting delta-hedging strategies in the continuous-time model, confirming the validity of the proposed approximate method.