On the Pricing of American Options under Stochastic Local Volatility and Stochastic Correlation via the RBSDE Framework

Battle Count: 50

Highly relevant for desks dealing with exotic derivatives, particularly American options on assets with stochastic volatility and correlation risk. The ability to price and hedge these instruments accurately is crucial for risk management and pricing in volatile markets.

Implementing RBSDEs with reflection requires careful handling of the obstacle condition. Integrating XGBoost for conditional expectation estimation adds complexity regarding data management and model training at each time step. However, libraries for XGBoost and SDE simulation are widely available.

The paper provides detailed parameter settings for XGBoost (tree depth, learning rate, number of trees), time discretization schemes, and specific model parameters for SABR and Heston extensions. It compares against established benchmarks (CTMC, DPDB). Code availability is not explicitly linked in the text, but the methodology is standard enough for reproduction given the details.