本项目应用稀有事件模拟技术把市场风险因子厚尾特征以多项式衰减的金融衍生品组合市场风险度量模型推进到市场风险因子以指数类型衰减情形，把金融衍生品组合市场风险度量模型推进到场外组合度量模型。创新体现：（1）建立Delta非中性下风险因子为非对称Laplace分布的金融衍生品组合市场风险度量模型，并应用两步重要抽样与分层抽样技术相结合进行有效Monte Carlo模拟计算； (2)利用多元厚尾分布来刻画不同类型风险因子相依关系，把信用风险因子嵌入基于多元厚尾分布的金融衍生品组合市场风险度量模型，并运用方差减少技术进行有效数值模拟计算；（3）在标的资产对数价格变化为多元跳跃扩散过程情形下，构建出带跳跃成分和信用风险的场外金融衍生品组合风险度量模型，把傅里叶变换技术和重要抽样技术相耦合进行有效稀有事件模拟；（4）应用本项目成果建立基于稀有事件模拟技术的我国结构性理财产品、存贷款隐含期权风险度量模型。

With application of the rare-event simulation techniques, this project develops efficient models for market risk measurement on derivative portfolio when the risk factors with hevy-tailed behaviors decayed at a rate of exponential type rather than of the traditional polynomial type, and extends market risk measurement model on exchange-traded derivative portfolio into risk measurement model on over-the-counter derivative portfolio being face of market risk and default risk with counterparties. The concrete new ideas are as follows. Firstly, we establish market risk measurement model on derivative portfolio with the risk factors having multivariate Laplace distributions under Non-Delta-Hedging case, and combine the two-step importance sampling with stratified sampling technique for effective Monte Carlo simulation. Secondly, we describe the dependences among different types of risk factors using multivariate heavy-tailed distributions, incorporate credit risk factors into the market risk measurement model on derivative portfolio with multivariate heavy-tailed distributions, and apply the variance reduction techniques for effective numerical simulation calculation.Thirdly, under the log price changes in underlying assets having multi-factor jump-diffusion process, we construct risk measurement model on OTC derivative portfolio containing jump component and credit risk, and coupled Fourier transforms technique with importance sampling technique for effective rare-event simulation. Finally, applying the research results of this project, we establish risk measurement model for structured financial products, deposits and loans with implied option based on rare-event simulation techniques.