您好,欢迎访问三七文档
51FRM100%PASS51FRM过关百分百2009学员专用资料第0页第1页PartIIQuantitativeAnalysisChapter2:FundamentalsofProbability本章英文参考读物:FRMHandBookDamodarGujarati,EssentialsofEconometrics,3rdEdition(NewYork:McGrawHill,‐2006).Chapter1–TheNatureandScopeofEconometricsChapter2–ReviewofStatistics:ProbabilityandProbabilityDistributionsChapter3–CharacteristicsofProbabilityDistributionsChapter4–SomeImportantProbabilityDistributions本章中文参考读物:《概率论与数理统计》盛骤等编(第一章至第五章),高等教育出版社本章掌握目标1.计量经济学引论(仅在Gujarati书中有此内容)描绘计量经济学的方法论区分用于经验分析的不同类型数据描绘计量模型中识别、解释和有效性检验的过程2.计量经济学基本概念了解随机变量的定义,区分连续变量和离散变量。了解一个事件概率的定义描绘相对频率(relativefrequency)或概率的经验解释。了解贝叶斯定理的定义并应用贝叶斯公式以确定一个事件的概率(仅在Gujarati书中有此内容)描绘和解释随机变量的概率函数(pdf),概率密度函数和累积密度函数区分单变量和多变量概率密度函数描绘边缘概率函数和条件概率函数解释统计独立性(statisticalindependence)与统计相关性(statisticaldependence)的区别3.概率分布了解随机变量期望值的定义并进行计算和解释。了解随机变量方差的定义并进行计算和解释。了解Chebyshev不等式的定义,并用它确定随机变量处于某个特定区间的概率(仅在Gujarati书中有此内容)了解两个随机变量协方差和相关性的定义,并能计算和解释了解随机变量均值和方差的定义,并能计算和解释描绘条件期望和非条件期望的区别了解偏度(skewness)和峰度(kurtosis)的定义,并能计算和解释描绘和识别platykurtic分布和leptokurtic分布(仅在Gujarati书中有此内容)了解正态分布随机变量的偏度(skewness)和峰度(kurtosis)的定义区分总体和样本,并计算样本均值、方差、协方差、相关性、偏度和峰度会计算VAR值略微了解Copula理论4.一些重要的概率分布描述正态分布和标准正态分布的关键性质了解随机抽样的概念和一个估计值的抽样分布建立频率分布并从一个频率分布计算相对频率了解样本均值的标准误的定义并能计算了解对数正态、t分布、chi‐square(卡方)分布和F分布的关键性质并能识别每种分布的通常在哪些情形出现。另外,把它们与正态分布联系和区别了解Binomial分布,Possion分布并能与正态分布联系和区别5.其它了解中心极限定理了解ExtremeValueTheory本章重点与难点本章重点内容:概率定义和贝叶斯法则,数学期望、方差、偏度、峰度和VAR值的计算,协方差和相关系数的计算,随机变量的线性转换。正态分布和其它各种分布期望和方差的简单计算,各种分布的特征比较。本章难点:VAR的理解,Copula理论,对数正态分布,t分布,chi方分布与正态分布的关系,中心极限定理、极值理论。2.1CHARACTERIZINGRANDOMVARIABLES2.1.1UnivariateDistributionFunctions(本小节简单运用微积分知识)ArandomvariableXischaracterizedbyadistributionfunction,(2.1)whichistheprobabilitythattherealizationoftherandomvariableXendsuplessthanorequaltothegivennumberx.Thisisalsocalledacumulativedistributionfunction.WhenthevariableXtakesdiscretevalues,thisdistributionisobtainedbysummingthestepvalueslessthanorequaltox.Thatis,(2.2)第2页wherethefunctionf(x)iscalledthefrequencyfunctionortheprobabilitydensityfunction(p.d.f.).Whenthevariableiscontinuous,thedistributionisgivenby2.3Thedensitycanbeobtainedfromthedistributionusing2.4Thesefunctionshavenotableproperties.Thedensityf(u)mustbepositiveforallu.Asxtendstoinfinity,thedistributiontendstounityasitrepresentsthetotalprobabilityofanydrawforx:2.52.1.2Moments(矩)提示:矩(Moments)是高等数学的概念。矩最常见的是一阶矩、二阶矩、以后统称高阶矩。一阶矩又叫静矩,是对函数与自变量的积xf(x)的积分(连续函数)或求和(离散函数)。统计学中一阶矩叫做数学期望(均值),二阶中心矩叫做方差。theexpectedvalueforx,ormean,isgivenbytheintegral2.6Thedistributioncanalsobedescribedbyitsquantile,whichisthecutoffpointxwithanassociatedprobabilityc:2.7So,thereisaprobabilityofcthattherandomvariablewillfallbelowx.DefinethisquantileasQ(X,c).The50%quantileisknownasthemedian.valueatrisk(VAR)canbeinterpretedasthecutoffpointsuchthatalosswillnothappenwithprobabilitygreaterthanp=95%,提示:Var从统计的意义上讲,本身是个数字,是指面临“正常”的市场波动时“处于风险状态的价值”。即在给定的置信水平和一定的持有期限内,预期的最大损失量。例如,某一投资公司持有的证券组合在未来24小时内,置信度为95%,在证券市场正常波动的情况下,VaR值为100万元。其含义是指,该公司的证券组合在一天内(24小时),由于市场价格变化而带来的最大损失超过100万元的概率为5%。第3页Iff(u)isthedistributionofprofitandlossesontheportfolio,VARisdefinedfrom2.8wherepistheright-tailprobability,andctheusualleft-tailprobability.VARcanbedefinedasminusthequantileitself,oralternatively,thedeviationbetweentheexpectedvalueandthequantile,2.9NotethatVARistypicallyreportedasaloss,i.e.apositivenumber,whichexplainsthenegativesign.variance2.10standarddeviation2.11skewness,whichdescribesdeparturesfromsymmetry.2.12提示:skewness,偏度,表征概率分布密度曲线相对于平均值不对称的程度。Negativeskewnessindicatesthatthedistributionhasalonglefttail,whichindicatesahighprobabilityofobservinglargenegativevalues.kurtosis,whichdescribesthedegreeof“flatness”ofadistribution,orwidthofitstails.2.13提示:kurtosis,峰度,是用来反映分布密度曲线顶端尖峭或扁平程度的指标,峰度值越小,顶端越尖峭,两侧为thintail(瘦尾),反之峰度值越大,顶端越扁平,两侧为fattail(肥尾)。Becauseofthefourthpower,largeobservationsinthetailwillhavealargeweightandhencecreatelargekurtosis.Suchadistributioniscalledleptokurtic,orfat-tailed.Highkurtosisindicatesahigherprobabilityofextrememovements.Akurtosisof3isconsideredaverage.2.2.1JointDistributionsjointbivariatedistributionfunction第4页2.14Inthecontinuouscase,thisisalso2.15wheref(u1,u2)isnowthejointdensity.Theanalysissimplifiesconsiderablyifthevariablesareindependent.2.16andtheintegralreducesto2.17Itisalsousefultocharacterizethedistributionofx1abstractingfromx2.Byintegratingoverallvaluesofx2,weobtainthemarginaldensity:2.18andsimilarlyforx2.Wecanthendefinetheconditionaldensityas2.19Here,wekeepx2fixedanddividethejointdensitybythemarginalprobabilityofx2.Thisnormalizationisnecessarytoensurethattheconditionaldensityisaproperdensityfunctionthatintegratestoone.ThisrelationshipisalsoknownasBayes’rule(贝叶斯法则).2.2.2CopulasItisrarelythecasethatfinancialvariablesareindependent.Dependenciescanbemodeledbyafunctioncalledthecopula,whichlinks,orattaches,marginaldistributionsintoajointdistribution.Formally,thecopulaisafunctionofthemarginaldistributionsF(x),plussomeparameters,θ,thatarespecifictothisfunction(andnottothemarginals).Inthebivariatecase,copulahastwoarguments:2.20Thelinkbetweenthejointandmarginaldistribution
本文标题:FRM过关百分百
链接地址:https://www.777doc.com/doc-5401271 .html