CFA三级知识点必备：Equity Portfolio Management_标准版.docx

寸二6UOHq0CCl一鼠法皿肾一Rgf藤三<ZDCOEOCraO=OJtOd Il m寸TABSUISS>UI IEquityInvestmentStyleClassifications"AThetwomainapproachesinstyleanalysisaretheholdings-basedapproachandthereturns-basedapproach.ATheholdings-basedapproach1undlnveslmenlsiylelooksattheattributesofeachValueBlendGrowthindividualstockinaportfolioandaggregatestheseattributesSize1.argetoconcludetheoverallstyleoftheportfolio.AcommonapplicationofthisideaistheMorningstarStyleBox.MidSmallHoldings-BasedStyleAnalysisAThestyleboxapproachaimstoclassifyapproximatelythesamenumberofstocksineachofthevalue,blend,andgrowthgroups,essentiallydistributingthemarketvalueofeachrowevenlyacrossthegrid.ATheclassificationofstocksintovalue/blend/growthinvolvesassigningastylescoretoeachindividualstock.Forexample,toassignavaluescore,thedividendyieldmaybeused./StockswouldberankedaccordingtotheirdividendyieldandaSCOreallocatedtoaStcICkbasedOntheirPe(CentiIeOfthemarketvalueoftheirparticulargroup./IfthestockisattheIoWer(high)endOfthedividendyieldrange,itWillreceiveaIc)W(high)SCoreCloSet。OQOO).Acomprehensivescoringmodelwouldusemanyindicatorsofvalueandcombinethemtogetherinapre-determinedweighting.Holdings-BasedStyleAnalysisATheclassificationofstocksThedifferencebetweenthestock,sgrowthandvaluescoresiscalledanetstylescore./Ifthenetstylescoreisstronglynegative,thestockisclassifiedasvalue./Ifthenetstylescoreisstronglypositivethenthestockisclassifiedasgrowth./Ifthenetstylescoreisclosetozerothenthestockwillbeclassifiedascore.Onceconstructedforastockuniverse,thegridcanbeusedasavisualaidtohelpcategorizeandtrackmanagedinvestmentportfolios.，Ataglance,aninvestorcanseewhereamanagerispositionedonthegrid,and,ifhistoricaldataexists,howthisstylehaschangedovertime.Returns-BasedStyleAnalysis>Areturns-basedstyleanalysisaimstoidentifythestyleofafundthroughregressionofthefundsreturnsagainstasetofpassivestyleindices.AByimposingaconstraintontheregressionthatthesumoftheslopecoefficientsshouldsumtoavalueof1,theslopecoefficientscanbeinterpretedastheanager,sallocationtothatstyleduringtheperiod.Forexample,areturn-basedstyleanalysismightconductaregressionoffundreturnsversusfourpassiveindicesasfollows:+斗存=tnereturnostylende×smesamepero雪=1 and s > O fors=thefundexposuretostyles(withconstraints.along-onlyportfolio)=aconstantofteninterpretedasthevalueaddedbythefundmanagert=theresidualreturnthatcannotbeexplainedbythestylesusedintheanalysisManagerSelf-IdentificationAThefud,sinvestmentstrategyisusuallyself-describedbythemanager.AComparingthatself-descriptionwithreturns-basedandholdings-basedstyleanalysiswilleitherconfirmaconsistentidentificationorindicateaneedforfurtherinvestigationandanalysistoexplainthediscrepancy.Somestylessuchasequitylong/short,equitymarketneutralandshortbiasdonotfittraditionalstylecategoriesandthemanager'sdescriptionandfundprospectusbecomesthekeysourceofinformationonstyleofsuchfunds.Equitystyleanalysis-Comparisonof2tech.AdvantagesDisadvantagesReturn-basedRequiresminimalinformationCanbeexecutedquicklyCosteffectiveMorewidelyappliedMaybeineffectiveincharacterizingcurrentstyleDifficulttodetectmoreaggressivepositionsHolding-basedMoreaccuratethanreturns-basedFacilitatescomparisonsofindividualpositionsCapturechangesinstylemorequicklyMoredataintensivethanreturns-basedanalysisLesseffectiveforfundswithsubstantialpositionsinderivatives.巨业创新憎值一2.PortfolioConstruction专业创新增值APassively-managedindex-basedequityportfolioscanbeconstructedby:FullReplication:fullreplication(holdallofthesecuritiesintheindex)StratifiedSampling:holdasampleofthesecuritiesbasedonstratifiedsamplingOptimization:usemorecomplexoptimizationtomaximizedesirablecharacteristicswhileminimizingundesirablecharacteristics.BlendedApproach:inpracticeablendoftheseapproachesmaybeused.AFullreplicationcanbecostlywhentherearelargenumbersofstockandliquidityislimited.Theportfoliomustberegularlyreconstitutedandrebalanced.Theadvantageoffullreplicationisthatitcloselymatchestheindexreturn(beforetransactioncosts).AToavoidthehighcostoffullreplication,itoftenmakesmoresenseforthemanagertousestratifiedsampling,inwhichheholdsasubsetoftheconstituentstocks,withthesampleselectedinsuchawayastoreplicatetheindexreturn/riskcharacteristics.Toimplementstratifiedsampling,themanagercreatesstrataacrosstheconstituentstocksthataremutuallyexclusiveandexhaustive.Themanagermustconsidersizeofthesampleused.Asmorestocksareaddedandtheportfolioapproachesfullreplication;lessliquidstocksareadded,increasingtransactioncostandtrackingerror.>Optimizationusesthetoolsofmodernportfoliotheorytoaddresstheproblemofminimizingtrackingerror.Theoptimizerseeksthecombinationofstocksthatwouldhaveminimizedtrackingerrorandpossiblymaximizedreturn.Theadvantagesofoptimizationtechniquesistheytypicallyexhibitlowertrackingerror,andthattheyexplicitlyaccountforthecovarianceamongconstituentstocks.Theobviousdrawbackofoptimizationisthatitisbasedonhistoricalrelationshipsandthosecanchange.Maintaintheoptimizationasthedatachangecanbecostly.Anotherdrawbackisthatitcancreateportfoliosthatarenotmeanvarianceefficientrelativetothebenchmark./Thesolutionistoaddaconstraintthattotalportfoliovarianceisequaltothevolatilityofthebenchmark.ABlendedApproachFullreplicationispreferredforindexeswithsmallnumbersofliquidstocks,whilestratifiedsamplingoroptimizationispreferableforindexeswithlotsofheterogeneous,thinlytradedstocks./ForlargeindexesliketheWilshire5000,theconstituentstocksrunthegamutfromlargeandliquidtosmallandthinlytraded.Inthatcaseacombinationoftwoapproaches,fullreplicationandstratifiedsamplingoroptimization.