基于多分形的资产组合风险度量建模与实证研究

考虑资产收益率的多分形特征及资产组合收益率间的复杂相依结构,运用Markov switching multifractal(MSM)模型对资产收益建模并结合Copula函数刻画相依结构,构建了资产组合市场风险度量的Copula-MSM模型.以风险价值(VaR)和期望损失(ES)作为市场风险度量工具,选取上证指数和恒生指数构成的资产组合进行实证分析,并比较Copula-MSM, Copula-GARCH和Copula-FIGARCH模型对VaR和ES风险测度的估计精度差异.实证结果表明,与Copula-GARCH和Copula-FIGARCH模型相比Copula-MSM能更准确的估计VaR和ES值,提高风险度量精度.     

混合分布的VaR非参数估计:对期货市场的实证分析

基于价格走势的不同将金融市场分为平稳行情,趋势与震荡行情,用混合分布刻画资产收益的母体分布特征,提出一种新的VaR非参数估计方法.利用高频数据对市场状态分类,基于Bootstrap的核密度方法估计子分布,最后计算总体分布的VaR.选取股指期货,铜期货,橡胶期货,豆粕期货等四个期货品种进行实证研究,估计了不同置信水平下的多头和空头VaR.实证结果表明:新方法相较传统的GARCH族模型,多种基于犀尾分布的GARCH模型,混合正态GARCH模型以及方差一协方差法能够更准确地估计空头VaR与尾部风险.     

基于偏t分布realized GARCH模型的尾部风险估计

借助偏t分布realized GARCH模型,提出同时考虑高频和低频信息的尾部风险估计方法,分别纳入RV、RRV和BPV构成尾部风险估计的对比模型,并利用上证综指高频数据进行实证分析.实证结果表明:相比EGARCH模型,realized GARCH模型能够提供更准确的VaR和ES估计;纳入对微观结构噪声和跳跃稳健的已实现测度有助于提高VaR和ES估计的准确性;realized GARCH模型在尾部风险估计中的表现对次贷危机前和次贷危机后两个不同的样本期间稳健.     

证券投资基金的风险资本确定研究

研究基金管理人如何配置风险资本以抵御来自资产组合收益风险的问题。建立基金管理人确定风险资本的模型，给出关于投资组合收益率分布函数的最小风险资本比率。用跳跃一扩散过程来描述市场收益率，在此基础上建立基金的投资组合的收益率模型，并给出模型参数的估计方法。利用随机模拟方法获得投资组合收益率的模拟样本．可以获得投资组合收益率的经验分布，进而计算最小风险资本率。用模拟方法获得的投资组合收益率的经验分布可以克服用常方差正态分布假设的误差。本文的结果对基金管理人的风险管理具有指导意义。     

跳跃-扩散模型框架下基于鞍点近似的信用组合一致性风险度量

在债务人资产价值的跳跃-扩散模型框架下,应用鞍点近似方法计算了信用组合损失的概率分布,并以算例具体说明了计算过程,得出了信用组合风险的VaR量度和一致性量度ES.然后,将鞍点近似方法得出的损失分布与蒙特卡罗模拟得出的结果相比较,得出了对于要求精度较高、资产数目较多的信用组合,基于鞍点近似方法计算一致性风险量度是信用风险管理的合适方法的结论.     

基于GJR模型的EVT动态风险测度研究

通过运用ARMA-GJR模型捕获上证综指的损失序列的自相关、波动集聚性和杠杆效应特征,用极大似然估计(MLE)估计模型参数以求出条件均值和条件方差以及标准残差序列;然后假设沪市指数损失标准残差序列近似满足EVT条件,分别取175、105和35个极值数据并运用MLE来估计广义帕累托分布(generalized Pareto distribution,GPD)的参数,进而估计出q分位数对应的动态风险值VaR(value at risk)和ES(expected shortfall);最后对风险测度方法的估计效果进行分析.实证结果表明:标准残差序列的极值尾部近似服从GPD,ES是相对于VaR更保守的风险测度方法.     

高频数据下基于PGARCH模型的VaR估计方法及应用

高频数据在风险价值VaR度量和预测方面的价值日益凸显,文中基于高频数据为嵌入日内收益过程的PGARCH模型提出一类稳健M估计,同时给出相应的VaR估计方法,并基于沪深300指数和恒生指数的5分钟高频数据对时间内和时间外的VaR进行估计预测.实证结果表明,高频数据下PGARCH模型的M估计所提供的VaR估计方法可更加准确的预测VaR,预测结果均优于日间低频数据的估计结果和基于高频数据的QMLE估计结果,该方法可以很好地应用于风险管理中.     

基于EGARCH和C-F扩展的VaR模型及应用

分别运用EGARCH和Cornish—Fisher扩展的VaR模型对中国深、沪股市的期望收益率与风险进行了实证比较研究，发现结合异方差的Cornish—Fisher扩展的VaR模型与仅用波动率描述的VaR计量方法比较，具有较好修正作用；同时对中国股票市场风险成因作了初步探讨。     

市场交易机制导致的流动性风险——考虑涨跌停制度产生的流动性风险

价格稳定机制作为市场特征的一部分,对于流动性有着重要的影响.通过对收益进行调整,将涨跌幅导致的流动性风险整合在风险度量VaR框架中.对比调整前后的离位点数目、VaR值等评估涨跌停限制对流动性风险的影响,并对考虑涨跌幅限制下的VaR计算方法选择、BIS资本乘数设置等进行讨论.结果表明,涨跌幅限制确实产生了流动性风险,支持VaR的估计过程中需要考虑涨跌停限制的影响.     

国际有色金属期货市场VaR和ES风险度量功效的比较

采用历史模拟法、蒙特卡罗模拟法、指数加权法、等权重法、GARCH法和极值理论法六种参数方法和非参数方法对国际有色金属期货市场风险值VaR建模, 并引入基于极值理论的预期不足ES对风险度量进行补充. 在利用返回检验对实证结果的功效进行比较研究的基础之上, 得出结论: 从对ES和VaR的检验结果比较而言, 当有色金属期货市场价格出现大幅跌涨时, ES比VaR更能准确地度量市场风险; 从对各种方法VaR的检验结果比较而言, 在95\%的置信度下, GARCH法和指数加权法能够更加有效地度量市场风险, 在99\%的置信度下, 指数加权法、GARCH法和历史模拟法能够更好地估计市场风险.     

Fast DOA estimation algorithm for MIMO sonar based on ant colony optimization

The maximum likelihood (ML) estimator demonstrates remarkable performance in direction of arrival (DOA) estimation for the multiple input multiple output (MIMO) sonar.However,this advantage comes with prohibitive computational complexity.In order to solve this problem,an ant colony optimization (ACO) is incorporated into the MIMO ML DOA estimator.Based on the ACO,a novel MIMO ML DOA estimator named the MIMO ACO ML (ML DOA estimator based on ACO for MIMO sonar) with even lower computational complexity is proposed.By extending the pheromone remaining process to the pheromone Gaussian kernel probability distribution function in the continuous space,the proposed algorithm achieves the global optimum value of the MIMO ML DOA estimator.Simulations and experimental results show that the computational cost of MIMO ACO ML is only 1/6 of the MIMO ML algorithm,while maintaining similar performance with the MIMO ML method.     

Feature Screening for Nonparametric and Semiparametric Models with Ultrahigh-Dimensional Covariates

This paper considers the feature screening and variable selection for ultrahigh dimensional covariates. The new feature screening procedure base on conditional expectation which is used to differentiate whether an explanatory variable contributes to a response variable or not, without requiring a specific parametric form of the underlying data model. The authors estimate the marginal conditional expectation by kernel regression estimator. The proposed method is showed to have sure screen property. The authors propose an iterative kernel estimator algorithm to reduce the ultrahigh dimensionality to an appropriate scale. Simulation results and real data analysis demonstrate the proposed method works well and performs better than competing methods.     

Smoothed estimator of quantile residual lifetime for right censored data

It is of great interest to estimate quantile residual lifetime in medical science and many other ?elds. In survival analysis, Kaplan-Meier(K-M) estimator has been widely used to estimate the survival distribution. However, it is well-known that the K-M estimator is not continuous, thus it can not always be used to calculate quantile residual lifetime. In this paper, the authors propose a kernel smoothing method to give an estimator of quantile residual lifetime. By using modern empirical process techniques, the consistency and the asymptotic normality of the proposed estimator are provided neatly.The authors also present the empirical small sample performances of the estimator. De?ciency is introduced to compare the performance of the proposed estimator with the naive unsmoothed estimator of the quantile residaul lifetime. Further simulation studies indicate that the proposed estimator performs very well.     

基于LMKL和OC-ELM的航空电子部件故障检测方法

针对航空电子部件故障样本获取困难以及检测准确率不高的问题,提出基于局部多核学习(localized multiple kernel learning, LMKL)和一类超限学习机(one-class extreme learning machine, OC-ELM)的故障检测方法。仅运用正常状态的小样本数据,给出了LMK-OC-ELM的数学表达形式,并在不同的门模型下推导了LMK-OC-ELM中局部核权重的优化方法;在获取局部核权重的基础上,定义了离线故障检测所需的统计检验量与阈值,以便工程实现。将所提方法应用于某型接收机,结果表明,在训练时间可控的前提下,与4种常见的一类分类(one-class classification, OCC)算法相比,所提方法可均衡地提高召回率、查准率和特异度,以LMK-OC-ELM-sig为代表,其在F1、曲线下方面积(area under curve, AUC)、G-mean和准确率4个指标上,比最近提出的局部多核异常检测(localized multiple kernel anomaly detection, LMKAD)方法分别提高了1.60%、1.57%、1.53%和2.23%。     

Improved H-infinity channel estimator based on EM for MIMO-OFDM systems

H-infinity estimator is generally implemented in timevariant state-space models,but it leads to high complexity when the model is used for multiple input multiple output with orthogonal frequency division multiplexing (MIMO-OFDM) systems.Thus,an H-infinity estimator over time-invariant system models is proposed,which modifies the Krein space accordingly.In order to avoid the large matrix inversion and multiplication required in each OFDM symbol from different transmit antennas,expectation maximization (EM) is developed to reduce the high computational load.Joint estimation over multiple OFDM symbols is used to resist the high pilot overhead generated by the increasing number of transmit antennas.Finally,the performance of the proposed estimator is enhanced via an angle-domain process.Through performance analysis and simulation experiments,it is indicated that the proposed algorithm has a better mean square error (MSE) and bit error rate (BER) performance than the optimal least square (LS) estimator.Joint estimation over multiple OFDM symbols can not only reduce the pilot overhead but also promote the channel performance.What is more,an obvious improvement can be obtained by using the angle-domain filter.     

自回归模型的加权复合Expectile回归估计及其应用

本文基于充分利用多个Expectile信息能提高参数估计效率的假设,提出了AR模型的加权复合Expectile回归(WCER)估计,探讨了该估计的最优权重,建立了其大样本性质,发现根据由数据驱动的最优权重所获得的WCER估计与最优权重已知时所获得的WCER估计具有相同的渐近有效性.数值模拟表明,当误差为厚尾或非对称分布,所提出的WCER估计大大优于传统最小二乘估计.即使误差分布未知,依然可以得到像极大似然估计一样具有优良统计性质的WCER估计.应用所提出的方法分析恒生指数和标准普尔500指数,实证分析表明:所提出的WCER估计在有效性意义下非常具有竞争力.     

Smooth estimation of ROC curve in the presence of auxiliary information

Receiver operating characteristic (ROC) curve is often used to study and compare two-sample problems in medicine. When more information may be available on one treatment than the other, one can improve estimator of ROC curve if the auxiliary population information is taken into account. The authors show that the empirical likelihood method can be naturally adapted to make efficient use of the auxiliary information to such problems. The authors propose a smoothed empirical likelihood estimator for ROC curve with some auxiliary information in medical studies. The proposed estimates are more efficient than those ROC estimators without any auxiliary information, in the sense of comparing asymptotic variances and mean squared error (MSE). Some asymptotic properties for the empirical likelihood estimation of ROC curve are established. A simulation study is presented to demonstrate the performance of the proposed estimators.     

一个智能集成的故障诊断系统

针对故障智能诊断中单一专家系统方法和单一神经网络方法中存在的不足 ,本文开发了一个集专家系统、神经网络以及人的“心智”为一体的较为通用的故障诊断系统 ,并将其成功地应用于一个“智能集成的可控硅整流稳压电源故障诊断系突”中 .     

基于蚁群算法的最大似然方位估计快速算法

针对最大似然(maximum likelihood, ML)方位估计方法多维非线性搜索计算量大的问题,将连续空间蚁群算法与最大似然算法相结合,提出基于蚁群算法的最大似然 (ant colony optimization based maximum likelihood, ACOML) 估计新方法。该方法将传统蚁群算法中的信息量留存过程拓展为连续空间的信息量高斯核概率密度函数,得到最大似然方位估计的非线性全局最优解。仿真结果表明,ACOML方法保持了原最大似然方位估计方法算法的优良估计性能,而计算量只是最大似然方法的1/15。     

一种噪声背景下的雷达目标识别方法

雷达高分辨距离像(high resolution range profile, HRRP)包含了丰富的目标结构信息,在雷达目标识别领域有良好的应用前景。针对传统的HRRP识别方法对噪声环境适应性差的问题,选用具有时移不变性的紧支撑小波自相关作为支持向量机(support vector machine, SVM)分类器的核函数,研究了幂次变换(power transform, PT)参数的选取对识别效果的影响,给出了参数选取经验公式,结合信噪比实时估算自适应地进行数据预处理以增强算法的抗噪性能。仿真表明,所提出的方法与传统的高斯径向基核SVM相比,提高了目标识别率,并且具有较好的噪声稳健性。