A tier one investment bank sought to compute MVA using ISDA’s SIMM methodology. The existing code-base was inefficient, with the sensitivities for SIMM calculated using finite differences (bumping), and relying on basic software optimisations. This created a major performance bottleneck and couldn’t scale with the increasing number of trades requiring SIMM.

To boost the computational performance – using algorithmic differentiation (AD), and advanced software optimisations – the client turned to Xcelerit to provide a full solution from training, advisory, and software tools.

top