基于BP神经网络和RBF神经网络的期权定价

Option Pricing with BP Neural Network and RBF Neural Network

摘要: 建立BP神经网络模型和两种径向基函数(RBF)神经网络模型，对以花旗集团为标的股票的看涨和看跌期权进行了模拟定价，并利用四项误差指标比较三种模型和B-S公式的定价预测能力。结果表明，在四种模型中，神经网络模型对期权的定价结果优于B-S模型；而在三种神经网络模型中，基于K-均值聚类算法和伪逆法的径向基函数神经网络对看涨期权的预测精度最高。

Abstract: A BP neural network model and two radial basis function (RBF) neural network models are established to price the call and put options whose underlying stock is the Citigroup (Citigroup Inc.), and the predictive abilities of the three models with those of the B-S formula by four error evaluation criteria are compared. Our results demonstrate that the neural network models are better than the B-S option pricing model, while the predictive accuracy of the RBF network based on the K-means clustering algorithm and the pseudo- inverse technique is the highest in pricing the call options.