Scholarlink

<p>We propose a fast Monte Carlo simulation (FMCS) for the pricing of four underlying asset equity-linked securities (ELS) on the Android application environment. Structured notes serve as yield-enhancement and risk-transfer instruments across interest rate regimes. Their valuation and sensitivity analysis require repeated Monte Carlo revaluation under multiple market scenarios, which motivates low-latency execution on commodity devices. The proposed method combines MCS with the Brownian bridge technique. It evaluates paths at observation dates, applies coarse barrier checks, and uses Brownian bridge refinement only when those checkpoints do not determine knock-in status. This design avoids fine-grid barrier checks for most paths and reduces per-path computational cost. Since Monte Carlo paths remain independent, the workload is embarrassingly parallel and suits multi-core CPUs, GPUs, and distributed systems. The method also allows a GPU kernel implementation through straightforward path-level parallelization, and it can serve as an efficient algorithmic layer in GPU-based Monte Carlo pipelines. Numerical experiments on an Android device confirm a substantial runtime reduction relative to a standard Monte Carlo baseline, with speedups on the order of tens of times for large sample sizes.</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

Fast pricing in Android platform calculator for four-asset step-down structure equity-linked securities using Brownian bridge

  • 期刊论文

  • 发表日期：

    2026年2月18日

查看全文
  • Hyundong Kim,
  • Juho Ma,
  • Hyunho Shin,
  • Sanghyun Lee,
  • Junseok Kim

摘要

We propose a fast Monte Carlo simulation (FMCS) for the pricing of four underlying asset equity-linked securities (ELS) on the Android application environment. Structured notes serve as yield-enhancement and risk-transfer instruments across interest rate regimes. Their valuation and sensitivity analysis require repeated Monte Carlo revaluation under multiple market scenarios, which motivates low-latency execution on commodity devices. The proposed method combines MCS with the Brownian bridge technique. It evaluates paths at observation dates, applies coarse barrier checks, and uses Brownian bridge refinement only when those checkpoints do not determine knock-in status. This design avoids fine-grid barrier checks for most paths and reduces per-path computational cost. Since Monte Carlo paths remain independent, the workload is embarrassingly parallel and suits multi-core CPUs, GPUs, and distributed systems. The method also allows a GPU kernel implementation through straightforward path-level parallelization, and it can serve as an efficient algorithmic layer in GPU-based Monte Carlo pipelines. Numerical experiments on an Android device confirm a substantial runtime reduction relative to a standard Monte Carlo baseline, with speedups on the order of tens of times for large sample sizes.