Machine learning with kernels for portfolio valuation and risk management (original) (raw)
Abstract:We introduce a simulation method for dynamic portfolio valuation and risk management building on machine learning with kernels. We learn the dynamic value process of a portfolio from a finite sample of its cumulative cash flow. The learned value process is given in closed form thanks to a suitable choice of the kernel. We show asymptotic consistency and derive finite sample error bounds under conditions that are suitable for finance applications. Numerical experiments show good results in large dimensions for a moderate training sample size.
Submission history
From: Lotfi Boudabsa [view email]
[v1] Sun, 9 Jun 2019 22:17:46 UTC (1,701 KB)
[v2] Wed, 19 Jun 2019 05:16:38 UTC (1,702 KB)
[v3] Tue, 25 Jun 2019 17:32:41 UTC (1,702 KB)
[v4] Fri, 19 Jul 2019 10:13:21 UTC (1,702 KB)
[v5] Thu, 15 Aug 2019 19:31:12 UTC (1,702 KB)
[v6] Wed, 15 Jan 2020 17:14:32 UTC (1,703 KB)
[v7] Sun, 30 Aug 2020 20:00:13 UTC (289 KB)
[v8] Wed, 26 May 2021 18:45:45 UTC (556 KB)