Papers

Pricing Vulnerable Spread Options under an Intensity-based Model

This paper proposes an intensity-based model to price spread options with default risk. Default risk is captured by a Cox process, whose intensity is correlated with the volatility. We also propose a more general correlation structure than previous works.
Download Paper

RGBoost: A revised gradient direction boosting machine

This paper presents a novel variant of the gradient boosting algorithm, termed RGBoost, that enhances performance by modifying the negative gradient in every iteration. Specifically, RGBoost substitutes the conventional negative error vector with a direction aligned to the true gradient of the loss functional within the context of a Reproducing Kernel Hilbert Space (RKHS). Empirical evaluations demonstrate that RGBoost achieves a notably faster convergence rate compared to traditional methods. This enhancement potentially offers substantial improvements in predictive accuracy and efficiency in various machine learning applications.
Download Paper