




版權說明:本文檔由用戶提供并上傳,收益歸屬內容提供方,若內容存在侵權,請進行舉報或認領
文檔簡介
Makingbigdataworkforcommodities
FromautomationtoAI,andcodingwithPythonto
democratiseddatasets,commoditytradingisundergoingadigitaltransformation
TheExxonValdezranagroundonMarch24,1989.Asitdidso,250,000barrelsofcrudeoilgushedfromthedamagedshipandimpactedtheAlaskashoreline.
Theaccidentwasanenvironmental,legalandreputationaldisaster–butitalsosetoffachainofeventsthatwouldupendtheglobalshippingindustryforever.
Introduction
Analysisiskeytocreatinginsight–andthere’smoretoanalysethanever
Theyearafterthevesselcrashed,politiciansintheUnitedStatespassedlawsaimingtopreventsimilarsignificantoilspills.Withintheselawswastheobligationthatsometankersshouldusevesselmonitoringandtrackingtechnology.Ifsuchshipscouldbetrackedelectronically,theargumentwent,therewouldbegreatertransparencyabouttheirmovements,alowerchanceofcollisions,andportscouldgainabetterunderstandofaship’sactions.Similarsystemsweresimultaneouslybeingdevelopedandtestedbyotherauthoritiesaroundtheworld.
By2000,everythinghadchanged.TheInternationalMaritimeOrganisation(IMO)adoptedanewrequirementforallvesselsoveracertainsizetouseautomaticidentificationsystemtransponders,knownasAIS.Thetechnology,whoseadoptionbecameeffectivein2004,workslikeGPStrackingforships,andincludesavessel’sidentity,location,speed,directionoftravelandplanneddestination.
Now,AISdatapowerswebsitesthattrackshipsallaroundtheworld,andtheinformationisfedintothecommercialsystemsofmultipleindustries–includingthoseoftradersandcommoditiestraders.Theradicalimprovementsintechnologyhaveenabledthetrackingofshipsinreal-timetobecomeroutine.Forcommoditiestraders,theavailabilityofthisinformationopensupthepotentialtomakesmarter,data-informeddecisionsaboutthemovementofgoods–fromcottontooilandmetal–aroundtheworld.Theavailabilityofthistypeofinformation,amongavarietyofsimilarsources,isincreasinglybecomingthecornerstoneoffundamentalstrading,whichseekstobuildapictureoftheworldusingdata.
AISisonlyonedatasource–marinerscan,ofcourse,turnofftheirAISbroadcaststoevadetracking,buttherearenowfewerwaystohidethaneverbefore.Theintroductionoflowcostsatellitesandhigh-resolutioncamerasallowstheEarth(andseas)tobephotographedatalltimes.“Youcanbeobservedbysatelliteimagery,”saysLeighHenson,theglobalheadofcommoditiestradingdivision,atRefinitiv.“Andevenifthere’scloudcover,theycanlookatyourlocationusingradar”.
Machinelearningisevenmakingitpossibletocombinethesedatasourcesandautomaticallywatchthemovementofvesselsovertime.Thesedatasciencetechniqueshavecaughtshipsillegallyfishing,smugglinggoodsandbreakinginternationallaws.Mostrecently,shipshavebeenfoundtamperingwithAISsignalsandbroadcastingfalseones;combiningdataandmachinelearninghasallowedthemtobecaughtintheact.
Theexplosionofdataisnotuniquetoshipping.Acrossallcommoditiestherehasbeenanincreaseintheavailabilityofdataaboutwhatishappeningintherealworld.Cheapsensors,ubiquitousinternetconnectionsandreal-timemonitoringmakeitpossibletotrackandquantifyvastswathesofthecommoditiesindustry.Oilstoragetankscanbemonitored.Methaneemissionscanbetracked.Theflowsofoilpipelinescanbewatched.“Increasinglysophisticatedtypesofobservationsarejustsproutingupallthetime,”Hensonsays.
Asthispictureoftheworldhasgrowninscope,sotoohasfundamentalstradingbecomeevermoresophisticated.Itisnowbasedonthousandsofdailynewsreports,billionsofdatapointsfromvastnumbersofsensors,almost-instantaneousupdatesfromfinancialmarketsandtheproprietarydataheldbycommoditiesfirms.Increasingly,thatsamedataisbeingprovidedbymultiplesources;themeasurementsofoneoilsilocanbetakenbymultiplecompanieswhoprovidetheirdatatothemarket.
Butsimplyamassingdataisnotenoughtogivetradersanadvantageoverthecompetition.Traderswhohavebeenequippedwiththerightanalyticaltools,includingtheabilitytoautomateresearchandvisualisegoodsbeingmoved,areabletounderstandtheworldinamuchmoregranularway.Beingabletoeffectivelyutilisethisknowledgewillbecomeincreasinglycriticalduringthecomingyears.Astechnologymakestradingmoreautomated,everysecondmatters.Companiesneedtobeabletoquicklyanalyseandvisualisethedatatheyhave,whilestaffneedtheskillstooperateanalyticaltoolsandgaininsights.Andthewholeprocessstartswithdatamanagement.
Thise-book,whichispartofafour-partseries,outlinesthekeydatapracticesandtechniquesthattradersandanalystsshouldbeputtinginplacetogainanedge.Itexploreshowthecommoditiesindustryhasbeenrevolutionisedbydataandthestepsneededtoingest,standardise,analyseandvisualisedata.Withouttheseprocessesinplace–eitherinternallyorbeingprovidedbythird-parties–tradingcompaniesarelikelytofallbehindthecompetitioninanincreasinglydigitisedworld.
Meetingthechallengeofanever-growingdatadeluge
Withannualinputsnowbeingmeasuredinzettabytes,there’smoreinformationavailablethanhumantraderscanrealisticallymanage
Theageofdatahasfundamentallychangedhowcommoditiestradingworks.Tento15yearsago,saysAlessandroSanos,globaldirector,salesstrategyandexecutionforcommoditiesatRefinitiv,traderswouldbasetheirdecisionsaroundexclusiveaccesstoinformation.“Thiswaseitherbecausetheyheldthephysicalassets–mines,plantations,infrastructure–orbecausetheyhadaccesstoinformationthroughanetworkofcontacts,orbootsontheground,”Sanossays.Whenitwastimefortradestobemade,theypickedupthephone.
Now,itisdataandtechnologythatdrivedecisionsandtransactions.Inallareasoflife,colossalamountsofdataarebeingproducedeverysingleday.AnalysisfirmIDChaspredictedthat,by2025,theworldwillbecreating163zettabytesofdataeveryyear,agiganticleapupfromforecastsof33zettabytesofdatabeingcreatedin2018.Toputitintophysicalterms,ifallthedatastoredbyRefinitivalonewasburnedontoDVDs,andthediscsstackedontopofeachother,theywouldtowerthreetimeshigherthanLondon’s310-metretallShardskyscraper.
“Dataisarguablythenewcurrency,”saysHilaryTill,theSolichscholarattheUniversityofColoradoDenver’sJ.P.MorganCenterforCommodities.“Tradersareattemptingtoacquireevermorenoveldatasets,especiallyinnichemarkets”.Thismassiveavailabilityofdatahasessentiallyledtoitsdemocratisation–itiseasierthaneverbeforeforcompaniestogatherdataaboutthemovementofcommodities.“Knowledgeisthemostvaluablecommodity,”Sanossays.“Today,thereareliterallythousandsofsourcesofdataavailable.Thosecompaniesabletoassimilatethisinformationtsunamianddetectthesignalfromthenoisewillemergeasthefutureleaders.”
Broadly,therearefourdifferenttypesofexternaldataproviderswhichcommoditiestradingspecialiststurnto.Eachprovidesadifferentkindofinformation,andinsomecases,maybetheonlysourceof
thatdata.Thefirstarethegovernmentagenciesthatcontroltheflowofinformationfromtheirlocalities.Secondly,therearecommercialorganisationsandbigdataprovidersthatholdcertaindatasets,suchastheweatherornewsdata,instructuredways.Thirdly,thereareinternationalmonitoringorganisationssuchastheInternationalMaritimeOrganisation,thebodythatintroducedAISdatatoshipping.Finally,therearetheexchangesthatdealwithtradingdata.Someofthesedatasourcesareavailableforfreeonline,butothersrequiresubscriptionsandcommercialdeals.
“Inonesense,dataismoredifficulttocomeby,sincedatavendorsfullyappreciateitsvalue,”Tillexplains.“However,inanothersense,dataiseasiertocomebyinthatsomedatasetssimplydidnotexistinthepast–atleastcommercially–thatdosonow.”Collectingallthisdataandmakingitavailabletopeoplehasclearadvantages.Byprovidingdetailedmarketinformationalongsidefundamentaldata,Refinitivisabletobringgreatertransparencytothemarket.Ithasneverbeeneasierforcompaniestounderstandwhereandhowcommoditiesarebeingmovedandsold.
MichaelAdjemian,anassociateprofessorofagriculturalandappliedeconomicsattheUniversityofGeorgia,saysdatacollectionwithinagricultureischangingboththeagriculturalindustryandtheworldoftrading.“Satellitedataonweatherandproductionpatterns,aswellasdatageneratedbyGPS-equippedtractors,dronesandsoil-sensingapplications,areregularlyintegratedintoreal-timedashboardstoimprovetheinformationavailabletodecisionmakersinthefinancialeconomies,”hesays.Adjemianaddsthatthiswillbecrucialgivengrowingworldwidedemandsforfoodandthechallengeoftheclimatecrisis.
Butcollectingmassesofdatacausesitsownchallenges.Thevolumeofdatahasbecomeincomprehensibleforhumans,andbeingabletomakethemostofitrequiresbeingabletoeffectivelyanalyseit.It’snowcrucialthatdatacanworkwithotherdata.“Thecompetitiveadvantageisreallyevolvingfrombeingabletoaccessthoseadditionalsourcesofdata,tohowwellcompaniescanintegrateit,commingleitwiththedatathattheyalreadyproducethemselves,andthenapplytechnologytogenerateinsights,”Refinitiv’sSanossays.
Thechallenge:findinginsightsindatafromawidevarietyofsources
Preparingforsuccess:gettingyourdatareadyforbusiness
Withthepropertools,datafrommultiplesourcescanbestandardisedandmerged,pavingtheawayforautomationandartificialintelligence
Commoditiestradinghasbeentransformedinthelasttwodecades–butthereisevengreaterchangecoming.Artificialintelligence,sophisticatedanalyticstoolsandvastlyimprovedvisualisationmethodspointtoafuturethatincludesincreasedautomation–butthisdoesn’tnecessarilymeanreplacinghumantradersandanalysts.Automationcanincludefarmoreeffectivedataprocessingandasophisticatedapplicationofavailableinformation.Essentially,machinescanminedataforintelligence;humanscanthenactuponit.
Buttheuseofdatacan’tbesuperchargedifitsfundamentalcomponentsaremissing.Companiesandtradersneedtoproperlyingestdataintotheirsystems,makesureitisstandardised,usedatasciencetoanalyseit,andunderstandthetoolsneededtovisualisethisanalysisandmakeitunderstandableforpeopleontheground.
Thetaskofpreparingdataforthefutureisn’taneasyone,Sanossays.Itis,however,achallengethatRefintivhasasolutionfor.Refinitiv’sDataManagementSolution(RDMS)isaplatformthatallowscompaniestocombineRefinitiv’smultiplesourcesofcommoditydataintotheirownprocesses.Datacanbemergedandnormalised,anditactsasawayforclientstostandardisedatafromRefinitivalongsidethird-partydataandtheirowninformation.
Justhavingthedataisn’tenough:italsoneedstobeoptimised
13
Disparatedataunited
Combiningmultiple,continuouslyupdatedsourcesintovastdatabasescanyieldexceptionalinsights–iftheycanbenavigatedandparsed
Thedataexplosionhascreatednewchallengesforcompanies.“Youalsoneedsomewheretoputitall,”saysRefinitiv’sHenson.Companiesneedtomakethemostofthedatatheyarecollecting–oraccessingfromthirdpartiessuchasRefinitiv–andneedtocreatesuitableenvironmentswhereitcanbecollectedandaggregated.However,Hensonexplains,increasinglycompaniesarelookingtoaccessandmanipulatedataremotely,addingthattradersandanalystswanttoconnecttodatafeeds–andtheningestitintoevenbiggerdatabases.
It’sherewherethecloud’squickstartuptimesandseamlessabilitytoexpandasrequiredcomeintoplay.BytakingadvantageofcloudhostingandmakingdataavailableinApplicationProgrammingInterfaces(APIs),itispossiblefortraderstoeasilyaccesshugedatastreamsandusetheminthewaystheydesire.Forinstance,Refinitiv’sRDMSallowsanalyststocreatewaystoseehowmuchoilisbeingmovedfromonelocationtoanother.HensonalsoexplainsthatRefinitivprovidesdataenvironmentsinitsterminalproduct,Eikon,whichholdsmorethan2,000pricingdatasources,with1,300providerssendingreports.Itcanalsobeusedinvirtualofficestoaidremoteuse.ThecompanyestimatesthatEikonDatacanbeaccessedandsharedeasily,andsubsequentlyactedupon.Thisdatacanthenbeaccessedandsharedeasily,andsubsequentlyactedupon.
Combiningdatastreamsallowstraderstobuildcustomdatabases
Buildinginsightsonstandardiseddata
Normaliseddatasetsarethefoundationsofinformeddecision-making
Makingonesetofdataworkwithothersetsofdataisn’taneasytask.Asanexample,datamustfirstbenormalised,andfieldsinonedatabaseneedtomatchthoseinanotherdatabaseiftheyaregoingtobesuccessfullycombined.Ifthatdoesn’thappen,theresultmaybeaninaccuratepictureoftheworld–anditmaysubsequentlybeharderfortraderstomakeaccuratedecisionsasaresult.
Sanossaysthechieftechnologyofficershespeakstoarefrustratedabouttheamountoftimetheirdatascientistsandstaffarespendingcleaningupdatasetsinordertooptimisetheirusefulness.“They’retellingmetheiranalystsspendupto90percentoftheirtimejustdoingthisaggregationandnormalisationofdata,”heexplains.Spendinglongermakingdatacompatiblemeanslesstimeisspentanalysingthedataandmakingdecisionsbasedontheintelligenceitprovides.Formanycompanies,theprocessofamassingdatasourcesandstandardisingtheinformationisbestlefttothirdparties.However,Refinitivhasdedicatedteams–locatedinBangalore,Manila,IndiaandPoland–thatanalyseincomingdatasetsandensuretheyarecompatible.Theytakedisparatesourcesofdataandmakeitpossibleforthemtobeeasilycomparedtootherdata.
Fortraders,takinginalreadystandardiseddatasets(throughAPIsordesktopsoftware)thatarereadytobemixedwithproprietaryorthird-partydatacanvastlycutdownontheamountoftimeneededtomaketradingdecisions.“Previously,ourcustomersmayhavebeenhappywithjustonesourceofoilflowsandcargotracking,”explainsSimonWilson,headofoiltradingatRefinitiv.“Now,theywanttotakemultiplesources.”Headds:“Weintegratetwoorthreesourcessothattheycangetablendedviewofthecargotracking.”Thisapproachcanhelptradersmakemoreinformeddecisions.
Oncedatahasbeeningestedandnormalised,itcanthenbepassedthroughanalyticstools,visualisationprocessesand–increasingly–artificialintelligenceandmachinelearningsystems.
Accuratetrackingofcargoneedsdatathathasbeenstandardised
13
Codingiskeyforanalytics
Beingabletointerrogatedataisavitalskillfortraders,andPythonisthetoolofchoice,offeringversatilityandeaseofuseforidentifyinginsights
GonearethedaysofsolelyusingMicrosoftExcelforanalysis–thedatasetsandsourcesavailabletotradersarenowtoolargeforthespreadsheetsoftware,andsonewertoolsareneededtoeffectivelyanalysebigdata.ThecodinglanguagePython,whichisusedacrosswebdevelopmentandanumberofplatforms,isincreasinglybecomingthesystemofchoiceforanalystssiftingtheirwaythroughdataintheworldoffinancialservices.It’soutstrippingbothJavaandC++asthego-tocodinglanguage.
Python’sgrowthisdowntotheeaseofwritingandthehugeamountofpre-writtencode,availablethroughdatasciencelibraries,thatisaccessibleonline.ThelanguageisalsoflexibleenoughtoenablecodetoberuninthecloudorputintoAPIs,andit’soneofthecornerstonesofemergingmachinelearningandartificialintelligencetools.Itisincreasinglybeingusedtomodelfinancialmarketsandhandlethevastamountofdatathat’sgathered.
TheriseofPythonischangingtheskillsetsoftradingworkforces.Thosewhocancodeareindemandandarebeinghiredbyfinancialservicecompanies–numerouslargebanksareutilisingPython-basedinfrastructurefortheirwork.Infact,somepredictthatmanycommoditiestradinghousesandfinancialserviceswillmostlybehiringonlythosewithPythonskillsinthecomingyears.
UsingPythonhelpsanalyststomanagelargervolumesofdata
Interact.Analysis.Action.
Smarterwaysofpresentingdatarevealsfarmorethanrawnumbersonaspreadsheetcould,meaninginsightsandforecastscanbebetterutilised
Thevalueofanalysingcommoditiesdatacomesfromtheinsightsitcanunlock.Increasinglypowerfulvisualisationtools,alongwithcustominstructioncodewritteninPython,arebeingusedtobringdatatolife.Bymakingtheresultsofanalysisvisual,it’spossiblefortraderstobetterunderstandwhatishappening.
Take,forexample,thetrackingofships.ThroughtheuseofAISdata,eachvesselatseacanbeaccuratelyplacedonmapsoftheworld.It’spossibletoclickonashipandsearchforitsdetails,seeitsposition,movementsanddestination.Divingdeeper,itisevenpossibletounderstandthecargothatitmaybecarryingandinferwhatitsmovementsmightmeanfortradingmarkets.
Butthisisjustthetipoftheiceberg.ToolssuchasPowerBI,TableauandothersareallowingstandardiseddatatohavePythonscriptsrunagainstitandproducenewresults.Thetoolsmakeitpossibletoseebiggerpatternswithinthedataandputthemintoachartforlegibility.Forinstance,aSankeydiagramcouldshowthemovementofindividualgradesofcrudeoilmovingfromWestAfricatoNorthernEurope.Datapresentedinthisfashionisofteneasiertocomprehendthannumbersinatable,andtheprocessallowscurrentandforecastviewsofthemarkettobecreated.Withbetterforecasts,it’sanothertoolthatmakesiteasierfortraderstomakeinformeddecisions.
Visualisationcanenabledeepinsightsfromdensedatasets
Tradingmovestowardsamoreautomated,technologicalfuture
Forbothestablishedcompaniesandrecentupstarts,data-empoweredautomationandvisualisationareanessentialcompetitiveedge
Evengreaterlevelsofdisruptionarecoming.Thelastdecadehasseenthefundamentalscommoditymarketbecomefloodedwiththesenewtypesofdata.Butthedataactsonlyasastartingpoint–whatcanbebuiltontopofitwillprovidenewopportunities.
Entermachinelearningandautomation:since2000,machinelearningtechniques,whichusesophisticatedalgorithmstoexaminedataandidentifypatternswithinit,haveimprovedexponentially,andtheywillonlycontinuetogrowmorepowerfulandadvanced.Muchliketheinfluxofdatafuellingthecommoditiesindustry,thissub-fieldofartificialintelligencehasbeendemocratised.Itispossibleforindividualsorbusinessestofindallthetoolstheyneedtorunmachinelearningapplicationsonlineand,often,forfree.
Throughoutthe2020s,theuseofmachinelearningwillincreaseacrossallindustries.Forcommodities,thiscouldmeanmoreautomatedprocessingofdataandpotentiallyevenchangingthewaythattradeshappen.“Ithinkwe’llseemorealgorithms,”Hensonsays.“We’llseemorecomputerstradinginsteadofpeople.It’sbeengrowingintheequitiesmarketsandit’sdefinitelybeengrowinginthecommoditiesworld,too.”AIcanstepintoperformtasksthataretoocomplexortootimeconsumingforhumans.
Thelevelofdisruptionthatwillactuallyhappenisdifficulttopredict.However,theultimategoalmaybeusingartificialintelligencetoaccuratelyforecastthetradingpriceofindividualcommodities.Itislikelyhumanswon’tbetakenentirelyoutoftheloop,butinsteadreceivegrea
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯系上傳者。文件的所有權益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網頁內容里面會有圖紙預覽,若沒有圖紙預覽就沒有圖紙。
- 4. 未經權益所有人同意不得將文件中的內容挪作商業或盈利用途。
- 5. 人人文庫網僅提供信息存儲空間,僅對用戶上傳內容的表現方式做保護處理,對用戶上傳分享的文檔內容本身不做任何修改或編輯,并不能對任何下載內容負責。
- 6. 下載文件中如有侵權或不適當內容,請與我們聯系,我們立即糾正。
- 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 上海招聘筆試題目及答案
- 上海安裝監理試題及答案
- 輪臺公司招聘合同范本2篇
- 房地產商合同廣東集體建設用地使用權出租合同3篇
- 運輸合同-貨物運輸委托合同2篇
- 日本平面設計師藝術解析
- 電梯安全急救指南
- 智慧停車項目績效評估報告
- 肝疾病病例討論
- 2025瓊臺師范學院輔導員考試試題及答案
- 催化裂化裝置-反再系統冷態開車(石油化工裝置仿真操作課件)
- 【地理】2023年高考真題江蘇卷(解析版)
- 舊電梯拆除施工方案
- 小學生研學旅行展示ppt模板
- 《智慧養老》創新創業大賽ppt
- 冀教版三至四年級《發展柔韌性練習》評課稿
- 漢語拼音聲母韻母拼讀全表打印版
- 運動系統病例分析01
- 天津市南開區南開中學2022-2023學年物理高二下期末復習檢測試題含解析
- 澠池鋁礦礦產資源開采與生態修復方案
- 功與功率 課件高一下學期物理人教版(2019)必修第二冊
評論
0/150
提交評論