Stochastic Grid Bundling Method for Backward Stochastic Differential Equations

We apply stochastic grid bundling method (SGBM) to numerically solve backward stochastic differential equations.SGBM takes advantage of both regress-later and the adaptive local basis to improve on traditional Monte-Carlo... [ view full abstract ]

We apply stochastic grid bundling method (SGBM) to numerically solve backward stochastic differential equations.

SGBM takes advantage of both regress-later and the adaptive local basis to improve on traditional Monte-Carlo regression.

The dependent variable is regressed in SGBM on basis functions at the end of the interval instead of the beginning as in traditional method, and the conditional expectation is computed exactly. This results in better accuracy.

With the adaptive local basis approach, the regression basis is defined on a partition of the domain, and the partition depends on the simulated examples. This results in better result and scalability.