This thesis deals with the numerical methods for backward stochastic differential equations (BSDEs) and their applications to the problem of hedging in finance. We analyze the Theta-scheme proposed by Zhao et al. [36, 37], the Euler scheme with Monte carlo simulation proposed by Bounchard and Touzi [6], and the Euler scheme with scaled random walk analyzed by Briand et al. [8]. The aim of this thesis is to provide a guideline along basic properties of BSDEs, a detailed error estimation of the numerical methods and their implementation and usage. We implement the three numerical methods for BSDEs in pure mathematical examples and for BSDEs which are related to the problem of hedging in the Black-Scholes model and the problem of hedging with higher interest rate for borrowing. The scientific contribution of this master's thesis is threefold. First, we implement the Euler scheme for BSDEs, which combines non-parametric regression estimation with the Monte Carlo simulation. Secondly, we compare the accuracy and efficiency of all three numerical methods presented in the master's thesis. Thirdly, we compare the performance of the Theta-scheme combined with Gauss-Hermite quadrature and linear interpolation with different choices of theta as well.
«
This thesis deals with the numerical methods for backward stochastic differential equations (BSDEs) and their applications to the problem of hedging in finance. We analyze the Theta-scheme proposed by Zhao et al. [36, 37], the Euler scheme with Monte carlo simulation proposed by Bounchard and Touzi [6], and the Euler scheme with scaled random walk analyzed by Briand et al. [8]. The aim of this thesis is to provide a guideline along basic properties of BSDEs, a detailed error estimation of the nu...
»