文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 经管营销 > 金融/投资 > News impact on stock price return via sentiment analysis
侵权投诉

News impact on stock price return via sentiment analysis

News impact on stock price return via sentiment analysis

Newsimpactonstockpricereturnviasentiment

News impact on stock price return via sentiment analysis

analysis

XiaodongLia,HaoranXiea, ,LiChenb,JianpingWanga,XiaotieDengc

a

DepartmentofComputerScience,CityUniversityofHongKong,83TatCheeAvenue,Kowloon,HongKongSpecialAdminstrativeRegionDepartmentofComputerScience,HongKongBaptistUniversity,Kowloon,HongKongSpecialAdminstrativeRegionc

DepartmentofComputerScienceandEngineering,ShanghaiJiaotongUniversity,Shanghai200240,China

b

articleinfoabstract

Financialnewsarticlesarebelievedtohaveimpactsonstockpricereturn.Previousworksmodelnewspiecesinbag-of-wordsspace,whichanalyzesthelatentrelationshipbetweenwordstatisticalpatternsandstockpricemovements.However,newssentiment,whichisanimportantringonthechainofmap-pingfromthewordpatternstothepricemovements,israrelytouched.Inthispaper,we rstimplementagenericstockpricepredictionframework,andpluginsixdifferentmodelswithdifferentanalyzingapproaches.Totakeonestepfurther,weuseHarvardpsychologicaldictionaryandLoughran–McDonald nancialsentimentdictionarytoconstructasentimentspace.Textualnewsarticlesarethenquantita-tivelymeasuredandprojectedontothesentimentspace.Instancelabelingmethodisrigorouslydiscussedandtested.Weevaluatethemodels’predictionaccuracyandempiricallycomparetheirperfor-manceatdifferentmarketclassi cationlevels.Experimentsareconductedon veyearshistoricalHongKongStockExchangepricesandnewsarticles.Resultsshowthat(1)atindividualstock,sectorandindexlevels,themodelswithsentimentanalysisoutperformthebag-of-wordsmodelinbothvalidationsetandindependenttestingset;(2)themodelswhichusesentimentpolaritycannotprovideusefulpredictions;(3)thereisaminordifferencebetweenthemodelsusingtwodifferentsentimentdictionaries.

Ó2014ElsevierB.V.Allrightsreserved.

Articlehistory:

Received31October2013

Receivedinrevisedform15April2014Accepted15April2014

Availableonline26April2014Keywords:Newsimpact

StockpricereturnSentimentPredictionExperiment

1.Introduction

Stockmarketisanimportantandactivepartofnowadays nan-cialmarket.Bothinvestorsandspeculatorsinthemarketwouldliketomakebetterpro tsbyanalyzingmarketinformation.Finan-cialnewsarticles,knownasonemajorsourceofmarketinforma-tion,arewidelyusedandanalyzedbyinvestors.InBigDataera,theamountofnewsarticleshasbeenincreasingtremendously.Infrontofsuchabigvolumeofnewspieces,moreandmoreinsti-tutionsrelyonthehighprocessingpowerofmoderncomputersforanalysis.Predictionsgivenbysupportsystemscouldassistinves-torsto lternoisesandmakewiserdecisions.Therefore,howtomodelandanalyzenewsarticlessoastomakemoreaccuratepre-dictionsbecomesaninterestingproblem.

Bag-of-wordsbasedapproachesmodelnewsarticlesbyvectorspacemodelwhichtranslateseachnewspieceintoavectorofwordstatisticalmeasurements,suchasthenumberofoccurrences,etc.Machinelearningmodelsarethenemployedtocapturetherelationshipbetweenthewordstatisticalpatternsandthestockpricemovements.Althoughmanybag-of-wordsbasedapproachesCorrespondingauthor.Tel.:+85268502340.

E-mailaddress:hrxie2@gmail.com(H.Xie).

News impact on stock price return via sentiment analysis

doc.guandang.net

0950-7051/Ó2014ElsevierB.V.Allrightsreserved.

havebeenreportedtohavepredictionpowerinsomepreviousworks[14,41,42],theymissoneimportantringonthechainofthemappingfromtextualnewsarticlestothe naldirectionalpre-dictions,whichisthesentimentofnews.AsshowninFig.1,oneimportantstepinthe owisthatthenewsarticlesare rstinter-pretedbyinvestorsandtranslatedintomarketsentiments;theinvestorsthenmaketheirdecisionsbasedonthesentimentinter-pretations;andmarketpricesaggregatetheactionsofeachinves-torandre ecttheminthe nalpricemovements.Therefore,integratingthesentimentanalysisintothepredictionframeworkwouldbecomemorecritical.

Sentimentanalysismodelsdocumentsfromsentimentdimen-sions.Insteadofjustmeasuringwordfrequency,eachword(espe-ciallythecolorfulonesthathavesentimentpolarity)isdecomposedandrepresentedbyavectorofsentimentfeatures.Forexample,word‘‘accelerate’’canberepresentedbystrongandactivebyusingHarvardIV-4psychologicaldictionary;usingthesamedictionary,word‘‘accept’’canberepresentedbypositive,submitandsocialrelationfeatures.Thenumberofthesentimentdimensionis xedinthedictionary.Eachdocumentcanberepresentedbyavectorofsentimentvaluesbysummingupthesentimentvectorsofeachwordinthedocument.Makingpredictionsbasedonthesentimentrepresentationshouldhavemanyadvantagesoverbag-of-words:

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email