+Quantile-QuantilePlots》讲Weareinterestedinasking:•Ifdatalooksrightwhenweuseparametricapproach?•Whatwedois./Plotourdataonahistogramandestimatetherelevantsummarystatistics./Considerwhatkindofdistributionmightfitourdata.》Aplotofthequantilesoftheempiricaldistributionagainstthoseofsomespecifieddistribution.Theshageofthe00glottellsusalotabouthowtheemoiricaldistributioncomoarestothesoecifiedone.》Inparticular,ifthe00glotislinear.thenthespecifieddistributionfitsthedata,andwehaveidentifiedthedistributiontowhichourdatabelong.19-210宝山·刨而.ti僵..+BootstrappedHistoricalSimulation》Theb。。tstrapisverxintuitiveandeasxtoSRRly.•Wecreatealargenumberofnewsamples,eachobservationofwhichisobtainedbydrawingatrandomfromouroriginalsampleandreplacingtheobservationafterithasbeendrawn.•Eachnew1resampled'samplegivesusanewVaRestimate,andwecantakeour’best'estimatetobethemeanoftheseresample-basedestimates.ThesameapproachcanalsobeusedtoproduceresamplebasedESestimates-eachoneofwhichwouldbetheaverageofthelossesineachresampleexceedingtheresampleVaR-andour’best’ESestimatewouldbethemeanoftheseestimates.》Abootstrappedestimatewilloftenbemoreaccuratethana’raw’sampleestimate,andbootstrapsarealsousefulforgaugingtheprecisionofourestimates.26-210宝山·刨而.ti僵..+BaselCommitteeRulesforBacl《testing》Fourcateg。riesofcausesforexceptions:•Basicintegrityofthemodelislacking.Exceptionsoccurredbecauseofincorrectdataorerrorsintheprogramming.Thepenaltyshouldappl弘•Modelaccuracyneedsimprovement.Theexceptionsoccurredbecausethemodeldoesnotdescriberisksprecisely.Thepenaltyshouldappl弘•Intradaytrading.Positionschangedduringtheday.Thepenaltyshouldbeconsidered.•Badluck.Marketswereparticularlyvolatileorcorrelationschanged.Theseexceptionsshouldbeexpectedtooccuratleastsomeofthetime.71-210宝山·刨而.ti僵..+Example1》InvestinabondfromSpain.whilepurchaseacreditdefaultswapfromaFrenchbank:•ThevalueoftheCDSisdeterminedbythedefaultprobabilityofthereferenceentitySpainandbythejointdefaultcorrelationofth...