AD for MVA
Margin Valuation Adjustment (MVA) is a pricing adjustment for derivatives, associated with the funding costs of posting initial margin. The computation of MVA in a SIMM (Standard Initial Margin Model) framework calls for the simulation of a trade’s sensitivities with respect to different risk factors. This imposes substantial computational costs when traditional finite difference estimators are employed. In this session, we will dive into how algorithmic differentiation (AD) can be used to quickly compute the necessary sensitivities and speed up the calculation of MVA. Furthermore, we will explore how machine learning techniques can be utilised to speed up the computation even further.
Sebastian Schneider is a quant researcher at Robeco, where he focuses on equity selection research. Before that, he completed his thesis on algorithmic differentiation for MVA at ING. He holds a Master’s degree in Finance from the Vrije Universiteit Amsterdam, where he completed the Duisenberg Honours Programme in Quantitative Risk Management.