基于ga2fgp技术的期权定价模型及程序化交易的自动优化方法

《基于ga2fgp技术的期权定价模型及程序化交易的自动优化方法》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、山东大学硕士学位论文AutomaticOptimizationMethodonOptionPricingModelandProgramTradingBasedonGA／GPTechnologiesBinTENG(InstituteforFinancialStudies，ShandongUniversity，Jinan，250100)(Advisor：Prof．YufengSHI)CoNTENTSInthispaper，weusegeneticalgorithms(GA)andgeneticprogramming(GP)optimizationtechniques，combi

2、nedwithfinancialtheories，respectively,toachievetheparametersestimationoftheoptionpricingmodelsandautomaticoptimizationoftheprogramtradingstrategies．Geneticalgorithmsandgeneticprogramming，arebothglobaloptimizationalgorithms，Learningfromthetheoryofevolutionsuchastherulesofnaturalselectionand

3、survivalofthefittest．Geneticalgorithmsapplytooptimizetheparametersofaparticularmodel，withhighefficiency,non-linearcharacteristics．Geneticprogramming，whichisbasedonthegeneticalgorithm，achievedtheop-timizationofthemodelstructurethroughthebinarytree，thusgreatlyexpandingthescopeofitsapplicatio

4、ninthefieldofmachinelearningandartificialintelli-gence·CombinedwithvolatilityestimationmodelsandOptionpricingmodelswhicharealmostbasedonBlack-Scholesformula，wecanpricingtheoptionproducts．Volatilityestimationisthecoreofoptionpricing．Inthispaper，westudytheestimationmethodsofExponentiallyWeig

5、htedMovingAveragemodel(EWMA)，GeneralizedAutoregressiveConditionalHeteroscedasticmodel(GARCH)andS—tochasticVolatilitymodel(sv)．WehavedesignedastandardgeneticalgorithmfunctionusingtheMATLABGeneticAlgorithmToolbox，anddevelopedaGAmethodforvolatilitymodelparameterestimationbyprogrammingdifferen

6、tfit—heSSfunctions．Moreover，withthecombinationofgeneticalgorithmandMonteCarlosimulationmethod，wehaveachievedtheparaLIneterestimationofGARCH—MCandSVmodels，andachievedthenumericaloptionpricing，asanimportantpartofthewholeoptionprcingcomputerprogram．一III—山东大学硕士学位论文Toverifythepracticalityoftheg

7、eneticalgorithmmethod，weuseChinawarrantsandHongKong’SHangSengIndexOptionsdataforempiricalanalysis．Consideringfromtheresultsofthevolatilityestimation，GAmethodsforEWMAandGARCHmodelsachievethesameaccuracywithMATLABbuilt—inalgorithm，andbetterportabilitythanit．Empi