融合时空信息与多样化特征的股票选择方法研究.pdf

融合时空信息与多样化特征的股票选择方法研究.pdf

  1. 1、本文档共68页,可阅读全部内容。
  2. 2、有哪些信誉好的足球投注网站(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。
  3. 3、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载
  4. 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
查看更多

Abstract

Stockselectionreferstoselectingoutstandingstocksbasedonagivensetofhistorical

stockpricesequences,inordertoachievebetteroverallperformanceininvestmenttrading(high

returnandlowrisk).Choosingstockswithhighinvestmentpotentialalsoplaysacrucialrolein

thedailyinvestmentportfolioandriskmanagementoffunds.Inrelevantfields,someexisting

researchbasedondeeplearninghasachievedgratifyingresults,whichhasattractedmoreand

moreresearchinterest.However,duetothehighstochasticityandcontinuousdynamicevolution

offinancialmarkets,thistaskisstillfullofchallenges.

Tocapturespatiotemporalfeaturesinstocksequences,andintegraterelevantmultimodal

auxiliaryinformation,thisarticleproposes:(1)Astockselectionmethodbasedonmultimodal

multi-scalegraphconvolutionalnetwork.Thismethodtakesthepredictionofstockpricefluc-

tuationsasthelearningobjectiveofthemodel,andselectsthetopkstockswithahigherproba-

bilityofriseforinvestmentbypredictingthefutureriseandfallprobabilitiesofeachstock.This

modelintroducesmediatextinformationandeffectivelyminesandalignsstockpriceandtext

datafeaturesthroughmultimodalfusinganddiffusingblocks.Themulti-scalearchitectureof

thismodelcanfullyextractcoarse-grainedglobalfeaturesandfine-grainedlocalfeaturesfrom

stocktimeseries.Atthesametime,theadaptivespatiotemporalgraphconvolutionallayerin

thismodelcanflexiblylearndynamicdependenciesbetweenindustries,companies,andtime

frames.

However,furtherin-depthresearchandanalysishavefoundthatdeterministicmethods

cannotfullyexplorethediversedatapatternsinlong-termstocksequences.Tolearnmultiple

datapatternsinlong-termstockdata,andimprovemodelefficiencyandeffectiveness,amore

robust(2)stockselectionmethodbasedonenhanceddiversevariationalauto-encoderwasex-

ploredanddesigned

文档评论(0)

营销资料库 + 关注
实名认证
内容提供者

本账号发布文档部分来源于互联网,仅用于技术分享交流用,版权为原作者所有。 2,文档内容部分来自网络意见,与本账号立场无关。

1亿VIP精品文档

相关文档