Convergence of reward functionals in a reselling model for a European option (Q2890728)

In the context of the classical Black-Scholes model, a Cox-Ingersoll-Ross process is used to describe the implied volatility \(\sigma(t)\), i.e., \(\sigma(t)=\sqrt{\widetilde\sigma^2(t)+\delta_0^2}\), where \(\widetilde \sigma^2(t)\) is the CIR-process correlated with the Black-Scholes asset price \(S(t)\). If the Black-Scholes call-price is given by \(C(t, S(t), \sigma)\), then the optimal reselling problem is defined like the optimal exercise problem for the American option: NEWLINE\[NEWLINE \Phi(\mathcal{M}_T)=\sup_{\tau\in\mathcal{M}_T} \operatorname{E}\mathrm{e}^{-r\tau}C(\tau, S(\tau), \sigma(\tau)), NEWLINE\]NEWLINE where \(\mathcal{M}_T\) is the class of Markov stopping times. A two-dimensional approximation, binomial for the log-price and trinomial for the implied volatility, is constructed. The weak convergence in the Skorokhod space of the reward functional is proved. The proposed discrete model is arbitrage-free.