基于粒子群算法的投資組合優(yōu)化問題研究.pdf

1、寧夏大學碩士學位論文基于粒子群算法的投資組合優(yōu)化問題研究姓名：安曉會申請學位級別：碩士專業(yè)：計算數(shù)學指導教師：高岳林20080401AbstractPortfoliotheoryhasbeenanimportantpartofmodemfinancetheoryIt’ScentralproblemishowtoallocateandutilizecapitalassetsunderriskTheworkofMarkowitzinport

2、folioselectionhasbeenmostinfluentialforthedevelopmentofmodernfinanceanditsapplicationinpractice，whereheappliedvariancetomcasureinvestmentriskandconstructedthemeanvariancemodelAstheresearchonriskmeasuredeepenson，ithasbeen

3、foundthatthereagesomeflawswhichcannotbeavoidedwhenusingvarianceInordertoovercometheseshortcomings，alotofresearchhasbeendoneintheoreticalfieldButuptothepresent，theseproblemshavenotbeensolvedsatisfactorily1nthisdissertatio

4、n，westudyportfolioselectionbyemployingriskvaluetheoryandoptimizationtheoryandconstructthemodelswithdifferentriskmeasuresNeweffectiveParticleSwarmalgorithmsarcpresentedtosolvethesemodelsITwokindsofhybridPSOalgorithmarepre

5、sentedinthispaperOneisEMPSOalgorithm謝tllaexponentialdecreasinginertiaweightandmixedmutationOlXaatorAnotherisLCPSOalgorithmwithalogarithmicdecreasinginertiaweightandchaosmutationstrategyItisshownbytested、Ⅳithwellknownbenc

6、hmarkfimetiomthatthesehybridPSOalgorithmscanOVeTPA3methephenomenonoftheprematureconvergenceandenhanceconvergencespeedandprecisionandimproveperformanceofPSoalgorithm2Thepaperaddsthecorporatecreditrisktransferintothecalcul

7、ationofloanreturnWesetupaloan’Sportfoliohybrid0lmultiobjectiveoptimizationmodel，inwhichthemaximumofportfolioprofitandtheminimumofriskofbanksaretakenasobjectivefunctions，meanwhile，thenumberofassetandtheupperandlowerlimits

8、andloan’SgrossaretakenasconstraintAnewdiscreteParticleSwarmalgorithmwithselfadaptablechanginginertiaweightispresentedtosolvetheproblemandanumericalexampleisgiventotestthemodel3ConsideringthecircumstancesofChinesesecuriti

9、esbusiness，atypicaltransactioncostmodelⅥ，ilhnoconvex—noconcavefunctionandfeveilueareintroducedWeputforwardanewportfoliooptimizationmodelbyusingriskmeasll陀indexCVaRinwhichthemaximumofexpectantportfolioprofitistakenasobjec

10、tivefunctionWetranslatetheoptimizationproblemwimnonlinearconstraintintoanonrestraintoglebyusingdynamicpunishedfunctionmethodandthensolvetheproblemwithEMPSOalgorithmFinallyacasestudy、析thsixstocksofHuandShencityandabankdep

11、ositisineduced4ThepaperaddstheplusskewnessofyielddistributingintoportfolioselectionmodelConsideringthecircumstancesofChinesesecuritiesbusiness。atypicaltransactioncostmodel謝thnonlinearfunctionandrevenueareintroducedWesetu

12、pamuMperiodportfoliooptimizationmodelbyusingriskmeasureindexVaRinwhichthemaximumofcumulateskewnessistakenasobjectivefunctionWrecombinethedynamicpunishedfunctionmethodwithLCPS0algorithmtosolvetheproblemFinallyacasestudy砸t