Skip to content
DraconiticResearch
Search Sources API
203,203 studies31 sourcesUpdated 51 min ago
PreprintCC BY 4.0arXiv

Neural Network Methods in the 2BSDE Formalism for Solving the High-Dimensional Black--Scholes Equation with Transaction Costs

  • Guhesh Kumaran
Open at sourcePDFDOICite

Abstract

A 2BSDE system is written down for a (D+1)-dimensional Black--Scholes partial differential equation (PDE) with the Leland transaction cost model which incorporates the sign of the curvature. The 2BSDE system takes input from the ``speed'' (derivative of the curvature). This system is solved using a single ResNet neural network which learns both the solution to the PDE as well as the degrees of freedom in the curvature matrix. The network is used to learn the price of derivatives such as call options and butterfly spreads on geometric baskets for D=10,100 achieving a time-averaged RMSE of less than $0.1$ in each case.

Related

  1. A potential exploring of a multi-dimensional coupled Burgers–Fisher–Huxley equation using finite volume and element methods (FVM/FEM)
    Tarik Chakkour
    Results in Applied MathematicsPeer-reviewedCC BY-NC-ND 4.0Full textAnalysis & PDEsMathematical physicsHAL
  2. Boundary-preserving randomized neural network methods for Poisson equations on general two- and three-dimensional domains
    Huifang Zhou
    arXivPreprintCC BY 4.0Full textNumerical analysis & scientific computingarXiv
  3. Statistical comparison of reconstruction methods for the inverse boundary problem of the one-dimensional wave equation
    Samuel Agenorwoth, Emilia Blåsten
    arXivPreprintarXiv licenceFull textAnalysis & PDEsNumerical analysis & scientific computingarXiv
  4. Subdivision-based collocation methods for time fractional convection diffusion reaction equation with environmental and biomedical applications
    Saima Bibi, Syeda Tehmina Ejaz
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textGeneral mathematicsCrossref
  5. Some Methods to Solve the Diophantine Equation ay^2=b(x^k-c^k)
    Xiaodong Zhuang, Nikos E. Mastorakis
    arXivPreprintCC BY 4.0Full textAlgebra & number theoryarXiv

Cite

BibTeXRIS
BibTeX
@misc{kumaran2026neural,
  title = {{Neural Network Methods in the 2BSDE Formalism for Solving the High-Dimensional Black--Scholes Equation with Transaction Costs}},
  author = {Guhesh Kumaran},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.41019},
  eprint = {2610.41019},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.41019},
  note = {CC BY 4.0},
}
RIS
TY  - UNPB
TI  - Neural Network Methods in the 2BSDE Formalism for Solving the High-Dimensional Black--Scholes Equation with Transaction Costs
AU  - Guhesh Kumaran
PY  - 2026
DA  - 2026/10/29
PB  - arXiv
DO  - 10.48550/arxiv.2610.41019
UR  - https://arxiv.org/abs/2610.41019
AB  - A 2BSDE system is written down for a (D+1)-dimensional Black--Scholes partial differential equation (PDE) with the Leland transaction cost model which incorporates the sign of the curvature. The 2BSDE system takes input from the ``speed'' (derivative of the curvature). This system is solved using a single ResNet neural network which learns both the solution to the PDE as well as the degrees of freedom in the curvature matrix. The network is used to learn the price of derivatives such as call options and butterfly spreads on geometric baskets for D=10,100 achieving a time-averaged RMSE of less than $0.1$ in each case.
ER  -