返回
PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS

PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS

ID：40723495

大小：699.37 KB

页数：102页

时间：2019-08-06

PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS_第1页
PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS_第2页
PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS_第3页
PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS_第4页
PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS_第5页
资源描述：

《PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS》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、PREDICTIONOFFINANCIALTIMESERIESWITHHIDDENMARKOVMODELSbyYingjianZhangB.Eng.ShandongUniversity,China,2001athesissubmittedinpartialfulfillmentoftherequirementsforthedegreeofMasterofAppliedScienceintheSchoolofComputingSciencecYingjianZhang2004SIMONFRASERUNIVERSITYMay2004Allrightsreserved.Thi

2、sworkmaynotbereproducedinwholeorinpart,byphotocopyorothermeans,withoutthepermissionoftheauthor.APPROVALName:YingjianZhangDegree:MasterofAppliedScienceTitleofthesis:PredictionofFinancialTimeSerieswithHiddenMarkovModelsExaminingCommittee:Dr.GhassanHamarnehChairDr.AnoopSarkarAssistantProfes

3、sor,ComputingScienceSeniorSupervisorDr.AndreyPavlovAssistantProfessor,BusinessAdministrationCo-SeniorSupervisorDr.OliverSchulteAssistantProfessor,ComputingScienceSupervisorDr.MartinEsterAssociateProfessor,ComputingScienceSFUExaminerDateApproved:iiAbstractInthisthesis,wedevelopanextension

4、oftheHiddenMarkovModel(HMM)thataddressestwoofthemostimportantchallengesof nancialtimeseriesmodeling:non-stationaryandnon-linearity.Speci cally,weextendtheHMMtoincludeanovelexponentiallyweightedExpectation-Maximization(EM)algorithmtohandlethesetwochallenges.Weshowthatthisextensionallowsth

5、eHMMalgorithmtomodelnotonlysequencedatabutalsodynamic nancialtimeseries.WeshowtheupdaterulesfortheHMMparameterscanbewritteninaformofexponentialmovingaveragesofthemodelvariablessothatwecantaketheadvantageofexistingtechnicalanalysistechniques.WefurtherproposeadoubleweightedEMalgorithmthati

6、sabletoadjusttrainingsensitivityautomatically.ConvergenceresultsfortheproposedalgorithmsareprovedusingtechniquesfromtheEMTheorem.ExperimentalresultsshowthatourmodelsconsistentlybeattheS&P500Indexover ve400-daytestingperiodsfrom1994to2002,includingbothbullandbearmarkets.Ourmodelsalsoconsi

7、stentlyoutperformthetop5S&P500mutualfundsintermsoftheSharpeRatio.iiiTomymomivAcknowledgmentsIwouldliketoexpressmydeepgratitudetomyseniorsupervisor,Dr.AnoopSarkarforhisgreatpatience,detailedinstructionsandinsightfulcommentsateverystageofthisthesis.WhatIhavelearnedfromhimgo

当前文档最多预览五页，下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页，下载文档查看全文
温馨提示：
1. 部分包含数学公式或PPT动画的文件，查看预览时可能会显示错乱或异常，文件下载后无此问题，请放心下载。
2. 本文档由用户上传，版权归属用户，天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容，确认文档内容符合您的需求后进行下载，若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误，付费完成后未能成功下载的用户请联系客服处理。