微觀市場(chǎng)結(jié)構(gòu)上課課件class trading costs and order strategies print

Trading TradingUseofpastorderexecutiondatayzethefactorsinfluencingtradingEvaluateexecutionComparethecostsofalternativestrategies&EstimateexpectedtradingDeveloporderexecutionstrategiestominimizetradingcosts/izenetreturnsAlearningtooltobetteruetioftrading executionAbadlyexecutedtradecannegativelyimpacttheperformanceofeventhebestidea.Trading ysisisacrucialpartofinvestment2 一 普遍的績(jī)效不佳可能是因 成本 過成本來區(qū)別使自己做的更好。 問題不同，最優(yōu)能夠降低成本，盡管會(huì)增加短 1.WhatistradingSlideExplicitcostsoftrading:commissions,taxes,ImplicitcostsofTransactioncostsotherthancommissionsandExecutionBid-askMarketOpportunityReducedportfolioreturns(canresultfromalloftheseInflatedshort-periodpricevolatility(duetobid-ask“bounce,”marketimpact,pricediscovery)TradingExplicitExecutionCosts(theimplicitcostsofBid-askMarketDelay/Opportunitythespreadbetweenthedealer’saskpriceandthedealer’sbidpricethepriceimpactthataninvestorcancreatebytradingonanasset,pushingthepriceupwhenbuyingtheassetandpushingitdownwhileselling.theopportunitycostassociatedwithwaitingtoCost TotalCost +Impact(intra-+Delay(inter-approx.157bps(one-314bps(round-Whatisthecumulativeimpactona10%annualreturnover:1,5,and10-yearhorizons?Total

TotalCostofBasisBasisMidBasisBasis0BasisBasis0DelayBasisBasis0

Mid

ExecutionExecutionThefirsttypeofimplicitcostisexecutionMarketimpactcosts–dealer’sbid-askspreadandpriceconcessionduetoriskofinformationasymmetryMarkettimingcosts–pricevolatilityofstockduringtimeofthetransactionduetootheractivityinthestockThesecostsarisefromdemandforimmediacywhichdiffersformarkettradingconditions,investorstyleandinvestors’objectives(-motivatedtrades)OpportunityOpportunityTheotherimplicitcostisofNOTtransactingrepresentsanopportunitycost,reputed“hiddencostoftrading”Naturaltradeoffbetweenexecutioncostsandopportunitycosts,sinceexecutioncostsarepositivelyrelatedtoimmediacy,butopportunitycostsarepositivelyrelatedtothedelayinexecutionInformation-motivatedinvestorsmayweighrelevanceofopportunitycostsmoreheavily;growthorvalueinvestorsmayweighexecutioncostsmoreheavilyMeasuringMeasuringMeasurementofcostsiscriticalforinvestmentmanagersinformulatingstrategies& sevaluatingperformanceMeasuringindirectcomponentshavenouniquesolutionandeachproposalhasmanyproblemsIndirectCosts=Executionprice–FairMarket ogoustotheHeisenbergprincipleinphysics,theprocessofexecutingatrademaskswhatwouldexistwithoutthetradetaking ce.Thedifferencebetweentheexecutionpriceandthatpricewhichwouldhaveprevaliedinitsabsenceisforeverunobservable.” Arnott&Wagner,FAJ,1990Performance VWAPVWAP=VolumeWeightedAverageLHOC=[Low+High+Open+Close]÷ADHOC=[VWAP+LHOC]÷“Fair“FairPrice”Pre-trade ark–beforetransaction,e.g.previousclose,lastAdvantages:capturesprevailingmarketDisadvantages:gaming,ortradingtoshowlowPost-trade ark–aftertransaction,e.g.nexttrade,tradedateAdvantages:gamingDisadvantages:neglectsrelevantpretradeIntraday–averageofdailyhi/lo,trade-weightedaverageAdvantages:rewardsstrategictradingatopen,close,aroundDisadvantages:gaming,byspreadingtradesoutduringOtherfactors–adjustmentsformarketNewNewSeveralrecentstudiesofequitytradingcostsaresurveyedinKeim&Madhavan,1998:Chan&Lakonishok1993/95usedSEICorpdatafor1986-88on115,000packagetradesbyinstitutionstoshowlowmarketimpactcostsof0.34%forbuys,-0.04%forsellsKeim&Madhavan,1996/97usedPlexusGroup’s21institutionalequitytradersandDFAupstairstradestoshowmarketpriceimpactsoverall0.2%to2.87%,smallstocksofupto5%,andvarywithinvestmentstyle,tradesize,marketcapandtraderskillEdwards&Wagner,1993,64,000PlexusGroupordersshowimplicitandopportunitycostsaresignificantandvarywithmarketconditionsDoestradingcostreallyCost

Goal:MinimizeIndirectSteeperslopeforinformation-Motivatedtraders

ExecutionSteeperslopeforgrowth,ElapsedTimesinceInvestmentDealersprovidefourfunctionsforOpportunitiesforinvestorstotrade Maintenanceofshort-runpriceBetterpriceinformationtomarketOrderlyandfairmarkets–exchangeagents,Designedto modatespecialtradingneedsofinstitutions(e.g.blocktrades,programtrades)Measurementandmonitoringoftradingcostsisimportant,butdifficulttaskbecauseimplicitcosts(executioncosts,opportunitycosts)areasimportantasexplicitcosts(commissions,fees)DoImplicitCostsReallyAverageTradingCostsbyTradeSizeQuartilefor26,000EquityBy21Institutions,January1991-March(Source:Keim&Madhavan,FinancialystsJournal,TradeSizeNYSE/AmexNasdaqTotalImplicitExplicit#ofTotal#of1-234-1-234-Implicitcostsequals(Pa/Pd)-1wherePaisaveragepriceofalltradesinanorderandPdispreviousdayExplicitcostsequalcommissionsperQuartilecutoffsdifferentforsellandbuyNoOdd-EighthDealersimplicitlycolludetoproduceunusuallylargespreadsinApple,Lotus,etc.Alternative BillChristie&PaulSchultz“WhydoNASDAQMarketmakersavoidOdd-eighthquotes?(JournalofFinance,1994)Theimpactofthestudyresultedin$1.03bclassactioncivilsuits,SECfinesof$26.3magainst26securitiesfirms,andDeptofJusticecriminalinvestigation;settledinJanuary1999.Schultz&Christie“WhyDoNasdaqMarketMakersAvoidOdd-EighthQuotes?”JournalofFinance,Dec1994,pp.1813-1840.NumberofInsideSpreadsfor100Nasdaqand100NYSEStocksofSimilarSize&Price0

$$per

TimeSeriesofDailyAverageInsideSpreads:AppleComputer InsideSpread

TheHiddenCostsofTradingCostComponentsofTradingBid-AskPriceOpportunityCostofManagementofTradingTheTradingCostThepayofftoactivemoneyAverageReturnonActiveMoneyManagers-Returnon=ReturnfromActiveTrading-TradingMoneymanagerseitherunder-estimatethetradingcosts,over-estimatethereturnstoactivetradingorboth.Tradingcostsareacriticalingredienttoanyinvestmentstrategy,andcanmakethedifferencebetweenasuccessfulportfolioandanunsuccessfulone.Bid-askSpreadinInvestmentBid-askspreadsaffectthereturnsfrominvestmentstrategies,buttheeffectwillvary,dependinguponthestrategy.PriceWhyisthereapricemarketsarenot yinformationalHowlargeisthepriceingstudiesofblocktradesontheexchangefloorseemtosuggestthatmarketsareliquidandthatthepriceimpactissmallandisreversedquicklyothersarguethatthepriceimpactislikelytobelarge,especiallyforsmallerandlessliquidstocks.PriceImpactinInvestmentthevariablesthatdeterminethatpriceimpactoftradingseemtobethesamevariablesthatdrivethebid-askspread.thesizeofthe esmuchmoreastrategyofinvestinginlow-pricedstockswhicharenotfollowedby ystsmayyieldexcessreturns,evenafterthebid-askspreadisconsidered,foraportfolioof$25millionbutceasetobeprofitableifthatsameportfolioes$500million.OpportunityCostoflikelytobegreatestforshortterminvestmentstrategies,baseduponprivateinformationormomentum,inmarketswithactiveinformationItwillbelessofanissueforlongterminvestmentstrategiesbaseduponpublicinformationandforcontrarianstrategies.FactorsthataffectthisIsthevaluationassessmentbaseduponprivateinformationorpublicinformation?HowactiveisthemarketforHowlong-termorshort-termistheIstheinvestmentstrategya"contrarian"or"momentum"4.ManagementofTradingStep1 Developacoherentinvestmentphilosophy aconsistentinvestmentstrategyStep2 EstimatethecostofwaitinggiventheinvestmentstrategyStep3 Lookatthealternativesavailabletominimizetransactionscosts,giventhecostofwaitingStep4 StaywithinaportfoliosizethatisconsistentwiththeinvestmentphilosophyandtradingstrategythathasbeenStep5 Considerwhetheryourinvestmentstrategyisyieldingreturnsthatexceedthecostsanintegralpartofanyinvestmentportfolioandcanmakethedifferencebetweenaportfoliothatbeatsthemarketandonethatdoesnotimposeasignificantdragonportfolioreturns,andmay inwhyactivemoneymanagersunder-performtheincludenotjustbrokeragecosts,butalsothecostsassociatedwiththebid-askspread,thepriceimpactcreatedbytradingandthecostofwaiting.difficulttocontrolbecauseactionstakentoreduceonecomponentofthetradingcosttendtoincreasetheotherdonotimposeauniformburdenonallinvestmentstrategies.Theypunishshort-term,information-basedstrategiesfarmorethantheydolongtermvalue-basedstrategiesaffectstrategiesthatfocusonsmall,less-liquidassetsfarmorethantheydostrategiesthatarebuiltaroundliquidassets.HowdidyougetabetterpricethanITheInvestingGuys "DidyouhearaboutChankoWireless?""Yeah,Igotinat$25.10.""Well,Igotinat"HowdidyougetabetterpricethanIChooseyourTherearemanygoodbrokeragesoutthere,butnotallofthemwillbegearedtothewayyouinvestDetermineyourYourinvestingstyleisoneofthebiggestfactorstoconsiderwhenselectinganonlinebroker.Todetermineyourstyle,youmustdefineyourneeds.Askyourselfquestionssuchasthefollowing:Doyouneed aladvice,orcanyoudoyour withresearchreports?HowlongdoyoutypicallyholdanWhatisthesizeoftradesyoutypicallyHowimportantisittohavedirectIsfastandefficientorderexecutionabsolu Categoriesof IndividualShort-termInvestorsDayTradingCost ysis&OrderExecutionStrategiesApril19,2004Provideinnovativeresearchontradingcosts,orderexecutionstrategiesandmarketstructurePublishreportsontradingcostsandmarketDeveloptradingcostmodelsandotherpre-tradetoolstohelpinternalandexternal strademoreefficientlyDeveloptheright yticalframeworkandstatisticalmethodologyfortradingcostcomparisonsDetailedexecution ysisfor Help schooseamongexecutionIntroductiontotrading ChoiceoforderexecutionTypesof yedSmallandlargeDirectandindirecttradingSmallQuotedEffectiveLargeImplementationExecutionLiquidityDeviationsfromHowfasttoTrading yzethefactorsinfluencingtradingEvaluateexecutionComparethecostsofalternativestrategies&executionvenuesEstimateexpectedtradingDeveloporderexecutionstrategiestominimizetradingcosts/ izenetreturnsAbadlyexecutedtradecannegativelyimpacttheperformanceofeventhebestidea.Trading ysisisacrucialpartofinvestmentChoiceoforderexecutionHowto WheretoFastor OurAgencyorMarketorlimitSingle-stocktradesorprogramtrades?AlgorithmicorWhento

CrossingDirectaccessorthroughaValue-addedofTradingcost informallthesechoicesTypesof NYSEbook,DealerNot- yedatmarketNYSEfloorECNreserveNot- yedatbroker-AgencyordersheldbyNot- yeddealerNot- yedat“Latent潛在的

Mostliquidityisnot

SmallandlargeTradingcostsforsmallEasytoEasytoEasytoMainWhenWhenisanordersmallandwhenlarge?QuotedEffectiveTradingcostsforlargeDifficulttoDifficulttoDifficulttoMainImplementationDeviationsfrom DirectandindirecttradingWhyareWhyareindirectcostshardtoIndirectEasytoHardtoOurIndirectOurSpreads,implementationshortfall,deviationsfromHardtomeasure,forlargeDirectSmall

Large SmallQuoted

YEDSymbolAassetmanagementfirm,wantstobuy1,000sharesofBestofferLowest yedpriceatwhichAAcanInourexample:AAcanbuyXYZatBestbidHighest yedpriceatwhichAAcanInourexample:AAcansellXYZatDefinitionofquotedDifferencebetweenbestbidandInourexample:RoundtripcostwhenAAbuysatofferandsellsatRewardtoliquidityQuotedspreadappliesuptoquotedInourexample:2,000AAwillhavetopaymoreforlarger

Effective

YEDSymbolDefinitionofeffectiveBuyorders:executionpriceminusmidquote,doubledInourexample:RoundtripcostofAAtradeintoaccountpriceBecauseofnot-disyedliquidityinsidethequote,smallordersmayexecuteatpricebetterthanquotePriceAAbought1,000

Sellat

Effectivespreadsatdifferent864

SEC-mandatedpubliclyavailableRule11Ac1-ButonlyforsmallEffectiveEffective

NYSE“overlap”stocksMarketorders500-1999August2002Selectedvenues20

WhyWhyaretheKnight,Madoffeffectivespreadssolow?

broker-

LargeYEDYEDSymbolSellCanstart“walkingup”thebookusingdisyedliquidityBuy2,000sharesatBuy1,500sharesatBuy1,000sharesatBut1,200sharesatBigliquidityAAneedstoaccessnot-disyedliquidityandworkorderovertimeAA“shops”AAfinallysendsordertoanbroker-dealer,XX,toworkoverthe yedSupposeAAwantstobuy600,000sharesofProfileofAAlargebuyorderWhataretheWhatarethepros&consofshop theorder?0:50first

13:10execution 9:45AAtobuy600,000

10:45orderarrivesatGoldman

$25.75averageexecution

GoldmanSachssplitsshareexecutions

Whatarethepros&consofsplittingtheorder?Source:GoldmanSachs,TradingandMarketStructure ImplementationDefinitionofImplementationShortfallBuyorders:averageexecutionpricepluscommissionsminuspriceatinvestment-decisiontime“All-in”tradingcostfrompointofviewofIncludescommissions,costof order,executionshortfallandopportunityofnon-filled

decision

($25.75–35￠+Source:GoldmanSachs,TradingandMarketStructure DefinitionofPre-TradeCostBuyorders:pricewhenfinalexecutionvenuereceivesorderminuspriceatinvestment-decisiontimeCostof theDelayedInformationSachsreceives

Whoisresponsibleforthepre-tradecost?decision

20￠Source:GoldmanSachs,TradingandMarketStructure ExecutionDefinitionofExecutionShortfallBuyorders:averageexecutionpriceminuspricewhenexecutionvenuereceivesorderTradingcostfrompointofviewofexecutingExecutingvenueisnotresponsibleforpre-tradeIncludesopportunitycostofslowPricewhenGoldman Sachsreceivesorder

AverageexecutionGoldmanSachsExecution15￠or60Source:GoldmanSachs,TradingandMarketStructure

Whatarethedifferencesbetweenimplementationandexecutionshortfall?PuttingitallGoldmanSachs9:45AAdecidestobuyIS=Implementationshortfall(includingPC=Pre-tradeIS=PC+ES+Source:GoldmanSachs,TradingandMarketStructure FactorsaffectingimplementationLiquidity Strainonliquiditycaused bythethreeGoldmanSachsInformationAcceleratesadversepricemoveagainstAA

ChangeinpriceofstockoverexecutionhadAAnotdecidedtoCanbefavorableorNotcausedbyAAorGoldmanAAfacesanunfavorablebase-priceBase-pricechangescreateexecutionSource:GoldmanSachs,TradingandMarketStructure Liquidityimpact,ordersizeandstockLiquidityimpactincreaseswithorder Agivenordersizehashigherliquidityimpactinsmall-cap

LiquiditySampleofGoldmanSachsSampleofGoldmanSachsLargecapSmall0< .25- .5-

.75- 1- 2- 3- 5-

10- 15- Ordersizerelativetoaveragedailyvolumelargecap=$10billionormore;smallcap=$1billionor Trade-offbetweenliquidityriskandexecutionSlice-uporderandexecuteovertimetoreduceliquidity ExecutionriskincreasessharplywithtimetoBasisBasis0

Intra-dayexecutionto 10to 30to 1to 5to 15to 30minTimeto

to 2hour

-

1.SampleofGoldmanSachsinstitutionalnot-heldmarketorders,USSharestrading Howfastto

historicaldataEasytoget

HardtoTrader’s“feelforthemarket”extremelyBroker-dealerBroker-dealervalue-Expectedliquidityinbaseprice slicing-uporderandspreadingovertime

Expectedbase-priceAssumingtraderexpectsbasepricetoriseoverthenextfourhours00to 10to 30to 1to 5to 15to 30minTimeto

1to2hour

流動(dòng)性提供者須知道是否這位管理者擁有特別的非公開,但能迅速改變 存貨風(fēng) 反向到了前的每一分鐘風(fēng)險(xiǎn)都在增加。流動(dòng)性提供這種風(fēng)險(xiǎn)的計(jì)算包括許多因素，規(guī)模越大，持有存implementationJackTreynor首先提出，后來AndrePerold修正 優(yōu)化目標(biāo)是 的效用最效用＝短期的a短期風(fēng)險(xiǎn)調(diào)整- 考慮將使的時(shí)序接近基準(zhǔn)組合，也即，要求 無線分拆,提前或者市價(jià)或限 我們應(yīng)該在什么時(shí) 如 但是成本并不為0，考慮到收益、風(fēng)險(xiǎn)和成本都隨著時(shí)間而動(dòng)態(tài)變化，我們應(yīng)該選擇什么時(shí)間來呢？實(shí)際成本模型應(yīng)該包含：廣度、深度和彈性。廣度用來量度的價(jià)差，深度用來量度市場(chǎng)的沖擊成本， HowtoestimateTradingEstimatingEffectiveCostsfromDailyData(JoeWhatdoWhatdoweknowabouttradingStatistically:PoorlyHighlyManyextremeStrongEconomically:opentovarietyofMostdirectlyameasureoftradingactivity(notRelationtotradingcostThispaperisaboutTradingcostestimatesforUSformedfromdailyCRSPExpectedreturnspecificationswiththecostRiskAssetpricing/A.Studiesusingmicros