报告题目：Monte Carlo Calibration to Implied Volatility Surface
报告人：韩传祥博士，台湾"清华大学"计量财务金融学系副教授

报告时间：2014年10月16日(周四)上午9:00-10:00
报告地点：数学楼二楼学术报告厅
 
 
摘要：
The calibration problem of implied volatility surface under complex financial models can be formulated as a nonlinear high-dimensional optimization problem.  To resolve this problem, we develop a sequential methodology termed two-stage Monte Carlo method. It consists of the first stage-dimension separation for splitting parametric space into two subspaces, and the second stage-standard error reduction for fast and accurate evaluation of option prices. Dimension separation reduces dimensionality of the optimization problem by utilizing the martingale transformation between the historical probability measure of the spot price and an equivalent probability measure of its option prices. Volatility model parameters of the spot price can be estimated a priori by a Fourier transform method and a maximum likelihood method or Markov chain Monte Carlo (McMC) estimation. The second stage standard error reduction aims simultaneously to reduce variance of the option payoff by the algorithm martingale control variate scheme, and to increase the total number of Monte Carlo simulations by the hardware graphics processing unit (GPU) for parallel computing. This two-stage Monte Carlo calibration procedure is capable of solving a variety of complex volatility models, including hybrid models and multifactor stochastic volatility models. Moreover, we extend this approach to joint calibration to market risk, interest rate risk and credit risk.
 
 
报告人简介：韩传祥博士是台湾"清华大学"计量财务金融学系副教授，台湾大学数学系兼任副教授，并主持清华－辉达（Nvidia-NTHU）计算金融联合实验室。他的研究领域为金融工程以及应用概率，特别是蒙地卡罗方法在金融工程之应用以及波动率的研究。