A Nonparametric Approach to Pricing Convertible Bond via Neural Network

The paper proposes a nonparametric method for estimating the price of convertible bonds using artificial neural networks (ANNs). Market convertible bonds prices quoted on the Shanghai stock exchange are used for performance comparison between the parametric Black-Scholes (BS), binary tree model and the proposed ANN model. The input variables of model are investigated and the results are compared. The results show that the performances of the proposed model produce often better convertible bonds price than other parametric models. The model simulation results slightly lower than actual market prices generally, which are significant and differ from previous literatures.