Efficient Simulation and Calibration of the Rough Bergomi Model via Wasserstein Distance

Battle Count: 89

The paper is highly relevant to quantitative finance and derivatives pricing. The rBergomi model is widely used for volatility modeling in trading desks. The efficient Monte Carlo pricing engine and robust calibration methodology directly impact option pricing accuracy, risk management, and trading strategy development. The Wasserstein-based calibration improves out-of-sample performance and tail pricing, which is critical for barrier options and exotic derivatives. However, the paper focuses on computational methodology rather than direct trading strategy development.

The mSOE scheme requires careful implementation of the SOE approximation (nodes and weights from Jiang et al.), Cholesky decomposition of the covariance matrix, recursive updates of historical components, and exact Gaussian simulation. The Wasserstein calibration requires automatic differentiation (TensorFlow), L-BFGS-B optimization with box constraints, and SVI fitting for distribution recovery. The GitHub repository provides example code, but the overall pipeline involves multiple interconnected components requiring numerical expertise.

The paper provides a GitHub repository with example code for the mSOE scheme and calibration procedure. Detailed algorithm descriptions (Algorithm 1), parameter settings, node/weight tables, and numerical experiments are provided. However, some implementation details like the exact SOE node/weight construction from Jiang et al. [16] require external reference. The use of TensorFlow for automatic differentiation and L-BFGS-B for optimization is specified.