![Search icon](https://hackernoon.imgix.net/search-new.png?w=19&h=19)
Ukunyuka kwe-AI ngokucacileyo kube nefuthe kumashishini ahlukeneyo, kwaye ishishini lezemali liphakathi kwezona zichaphazeleke kakhulu . Umzekelo, ukuqaliswa koluntu kweemodeli ezifana ne-GPT-3.5 kulo nyaka uphelileyo kunyuse umdla wokusebenzisa i-AI ukunceda ukwandisa izakhono zabaphathi bengxowa-mali kuhlalutyo, kulawulo lomngcipheko, kunye nokwenza izigqibo.
Ngaloo ndlela, izixhobo ze-AI ziphunyezwa ukwenza uvavanyo lweemarike luchaneke ngakumbi kwaye lulawule umngcipheko ngokufanelekileyo. Abaphathi beepotfoliyo kulindeleke ukuba benze uvavanyo olucacileyo lweentshukumo zemarike, bacuthe ukhetho lotyalo-mali olufanelekileyo, kwaye balawule imingcipheko xa besebenzisa i-algorithms yokufunda koomatshini, ukusetyenzwa kolwimi lwendalo, kunye nezixhobo zobukrelekrele bokwenziwa ekurhwebeni kwabo.
Ukudityaniswa kwe-algorithms yokufunda koomatshini, kunye nezixhobo zokucwangcisa ulwimi lwendalo kwiindlela zokurhweba zabadlali abaphambili, zibanceda ukuba bandise ukusebenza kakuhle kwezi nkqubo kwaye bafumane inzuzo yokukhuphisana ngokukhawuleza kunye nezigqibo ezichanekileyo zotyalo-mali kunye nohlalutyo oluqikelelwayo.
Kumashumi eminyaka adlulileyo, i-AI iphunyezwe kumacandelo ahlukeneyo oshishino lwezemali. Kwiofisi yangasemva, ii-algorithms ze-ML zisetyenziselwa ukufumana i-anomalies kwiilogi zokubulawa, ukufumanisa ukuthengiselana okukrokrelayo, kunye nokulawula umngcipheko, okukhokelela ekwandeni kokusebenza kunye nokhuseleko. Kwi-ofisi engaphambili, i-AI inceda ukwahlula abathengi, ukwenza iinkqubo zokuxhasa abathengi ngokuzenzekelayo, kunye nokwandisa amaxabiso aphumayo.
Nangona kunjalo, eyona nxalenye inomdla kuyo ngamandla e-AI kwicala lokuthenga lemali - ukuchonga imiqondiso exela kwangaphambili phakathi kwengxolo yentengiso ngokuhlalutya izixa ezibalulekileyo zedatha ngokukhawuleza. Umzekelo, ezo zicelo zinokubandakanya uqikelelo lwexesha loqikelelo, ukwahlula iimarike, kwaye kunjalo, nokulawula iipotfoliyo ze-asethi. Amathuba e-AI okuqhuba kunye nokuhlalutya iiseti zedatha ezininzi zinceda ukufumana iipateni ezifihlakeleyo iindlela zemveli ezinokuthi ziphosakele.
Ukulungiswa kwePotfoliyo ibe yinto eqhelekileyo kumashumi eminyaka aliqela, iphuhla kakhulu phantsi kophuhliso lwenzululwazi yedatha kunye nokuphunyezwa kobuchule obuphezulu bokubala. Iindlela ze-lassical, ezifana ne-Markowitz's Modern Portfolio Theory (1952) kunye neModeli ye-Asethi ye-Asethi ye-Capital (1964) zaziswa ngaphezu kweminyaka engama-50 eyadlulayo kodwa zisasebenza. Nangona kunjalo, ukulinganiselwa kwabo ekuphatheni umngcipheko ongeyena umgca kunye nokuxhomekeka kwidatha yembali kuya kucaca ngakumbi ngemini.
Izenzo ezifana nemodeli yomngcipheko, uhlalutyo lwemeko, kunye norhwebo lobungakanani, oluphunyezwe ngokubanzi ngabadlali abaphambili, njenge-Renaissance Technologies, i-DE Shaw, kunye noTyalo-mali oluBini lweSigma luye lwakhokelela ekuphunyezweni kwe-algorithms enzima kunye neyokuhamba phambili. Ukongeza, ishishini liye lachatshazelwa kakhulu yi-AI kwiminyaka yakutshanje, njengoko ukufundwa koomatshini kunye nobukrelekrele bokwenziwa buye benza ukuba uhlalutyo oluqikelelweyo luchaneke ngakumbi, kwaye lwenze okufanayo kwizicwangciso zotyalo-mali lobuqu kunye neenkqubo zokuthatha izigqibo ezintsonkothileyo.
Olu tshintsho luqhutywa yi-AI lwenze ukuba abaphathi bepotfoliyo baqhubele phambili uluhlu oluninzi lwedatha ngexesha lokwenyani kunye nokusombulula imingeni emithathu engundoqo:
Ngoku ka
Ukwandisa ukwamkelwa kunye notyalo-mali kwizisombululo zolawulo lwe-asethi ezixhaswe yi-AI kunye nokuqaqambisa ukusetyenziswa okubonakalayo kwe-AI ekuphuculeni ipotfoliyo.
Ukwamkelwa kwe-AI kushishino lolawulo lwe-asethi ayisiyonto intsha; ibone ukukhula kwiminyaka yakutshanje kodwa isanqunyelwe kwinani elincinci labadlali beemarike ezizezi mali zehedge, iiofisi zolawulo lobungakanani, amasebe amakhulu ophando, kunye namaziko emali asebenzisa iinkonzo ze-IT.
Mininzi imimandla yesicelo se-AI esele ikhona:
I-AI iphucula kakhulu inkqubo yokwakhiwa kwepotfoliyo. Umzekelo, indlela yakudala ye-Markowitz's Modern Portfolio Theory, exhomekeke kwiikhonsepthi zokwandisa i-convex, isebenza njengesandulela kwiindlela zangoku eziqhutywa yi-AI. Isizathu sokuba le thiyori yesiseko ibaluleke kakhulu kukuba yenza isiseko apho i-algorithms ye-AI inokutshintsha ngakumbi kwaye icokise izicwangciso zotyalo-mali.
Kule mihla, i-AI iyanda kule thiyori ngokuphonononga ubungakanani obutsha bedatha kunye nokudibanisa iindlela eziphambili zohlalutyo. Esi sakhono sedatha sandisiwe sivumela ukuthathwa kwezigqibo ezintsonkothileyo nezinolwazi - isenzo esiye sasetyenziswa ngokubanzi kushishino.
Ubuchule obuthile be-AI buhambelana ngokugqibeleleyo nolawulo lobungakanani, kusetyenziswa amanani amakhulu edatha malunga neziseko zenkampani, imeko yezoqoqosho olukhulu, okanye iimeko zentengiso. Ii-algorithms zokufunda koomatshini zinokufumana ubudlelwane obuntsonkothileyo obungeyomda phakathi kwezinto ezahlukeneyo kwaye, ewe, ukubona iindlela abahlalutyi abangakwaziyo ukuzifumana.
Uhlalutyo lombhalo lolunye usetyenziso lwe-AI kuhlalutyo olusisiseko. Ukusebenzisa ukusetyenziswa kolwimi lwendalo (NLP), iinkqubo ze-AI kunye nokuhlalutya imithombo yombhalo efana neengxelo zengeniso yenkampani, ukukhutshwa kweendaba zebhanki ephakathi, kunye neendaba zemali. Nge-NLP, i-AI inokukhupha ulwazi olubalulekileyo lwezoqoqosho kunye nezezimali kule datha engacwangciswanga. Ngokwenza njalo, ibonelela ngomlinganiselo wobuninzi kunye nocwangcisiweyo ophucula kwaye uncede ukutolika komntu.
Amandla e-AI aluncedo kakhulu ekurhwebeni, apho ubunzima bentengiselwano kunye nesidingo sesantya silingana. I-AI isekela ukurhweba kwe-algorithmic ngokuzenzekelayo izigaba ezininzi zenkqubo, ukuphucula ukusebenza kakuhle kweentengiselwano ezilawulwa kwiimarike zemali.
I-AI ivule ithuba lonikezelo olubanzi lweenkonzo zokucebisa ngotyalo-mali olulolwakho ngexabiso eliphantsi. Ezi nkqubo zisebenzisa i-algorithms entsonkothileyo yokucubungula idatha yemarike yexesha langempela, iza nezona zicwangciso zifanelekileyo kwiimfuno zomthengi ngamnye ngokusekelwe kwiinjongo zabo zokubuya kunye neeprofayili zomngcipheko.
Kulawulo lomngcipheko, i-AI inceda ngokumisela iimeko ezahlukeneyo 'ezinokwenzeka kodwa ezingafunekiyo', ezithi nazo ziphucule izenzo zemveli ezigxile kuphela kwiziphumo ezinokwenzeka.
Iindlela zokufunda zoomatshini beClassical zisathandwa kakhulu kuLawulo lwePotfoliyo, kwaye zezi: IiModeli zeLinear, ezibandakanya i-Ordinary Least Squares, iRidge Regression, kunye neLasso Regression. Ezi zidityaniswa rhoqo kunye nenkqubo yokuPhucula iMean-Variance Optimization kunye neendlela zokuchithwa kwe-matrix ezifana nokuNcitshiswa kweXabiso eSingular (SVD) kunye noHlahlelo lweCandelo eliyiNtloko (PCA), olusisiseko ekuqondeni ubudlelwane be-asethi kunye nokwandisa ulwabiwo lwepotfoliyo.
Ibekwe phakathi kwezi ndlela zamandulo kunye neendlela zangoku ngakumbi ngooMatshini beVector yeNkxaso (SVMs). Nangona ii-SVM zisetyenziselwa ukuziqhelanisa, aziqhelekanga ukusasazwa kodwa zidlala indima ebalulekileyo, ngakumbi, kwimisebenzi yokuhlela ejolise ekuqikeleleni ukusebenza kwesitokhwe.
Le misebenzi idla ngokubandakanya ukuxela kwangaphambili ukuba isitokhwe siya kufumana inzuzo okanye ilahleko, sisebenzisa idatha yezemali yembali kubandakanya ukuguquguquka kwexabiso lesitokhwe kunye nomthamo wokurhweba ukubeka iimpahla kwiindidi kunye nokubikezela ukusebenza kwazo.
Ukuthetha malunga neendlela zale mihla, uthungelwano lwe-neural lubonisa inkqubela phambili enkulu kumatshini wokufunda kulawulo lwepotfoliyo kwaye inikezela ngezakhono eziphuculweyo zokubonisa iipateni ezintsonkothileyo ezingeyomigca ekunzima ukuzibamba ngeemodeli zemveli. Ngaphandle kothungelwano lwe-neural, ezinye iindlela zeklasiki ezifana nokufunda okungajongwanga kunye nokuphucula ngakumbi kwaye kusulungekiswe uhlalutyo lwedatha, ukwenza ukufunyanwa kunye nokusetyenziswa kweempawu zentengiso ezifihlakeleyo zenzeke.
Iindlela ezintsha, ezifana nokuQiniswa kokuFunda kunye ne-Deep Q-Learning kuzisa ezi mpawu kwiindawo zokuthatha izigqibo ezikhawulezayo, apho iiphothifoliyo zinokuhlengahlengiswa ngexesha langempela ukuze kulungiswe iziphumo zemali ngokusekelwe kwinkqubo yokufunda kwiingxelo zeemarike.
Ubuchwephesha bokuCwangciswa koLwimi lweNdalo njengohlalutyo lweemvakalelo lunokunceda ukukhetha nokukhetha izimvo eziqhelekileyo kwizinto ezifana namanqaku ephephandaba, izithuba zeendaba zoluntu, kunye neengxelo zohlalutyi. Ukongeza, abaphathi beepotfoliyo banokuhlalutya ulwimi olusetyenziswa kumajelo eendaba zezimali, kubandakanywa neengxelo zengeniso yeefemu, ukuva iimvakalelo zabatyali-zimali kwaye baqikelele iintshukumo zemarike, zonke ezo ezilulwazi olubalulekileyo kwinkqubo yokwenziwa kwezigqibo.
Iifemu ezigxile ekurhwebeni kwe-high-frequency trading (HFT), njengalezo zisebenzisa i-AI-powered quantitative algorithms yorhwebo, yenza imali ngokungasebenzi kakuhle okwenzeka okomzuzwana nje kwimarike. Ezi femu zisebenzisa itekhnoloji yokufunda koomatshini ukuhlalutya ulwazi lwentengiso olufanelekileyo ngesantya esiphezulu kakhulu kwaye zenze iiodolo ngokuchaneka kwexesha ngokufutshane njenge-millisecond.
Ukuphunyezwa ngokukhawuleza okunjalo kubavumela ukuba bazuze kumathuba e-arbitrage kunye nokwandisa inzuzo ngokuthatha inyathelo malunga nokungafani kwexabiso ngokukhawuleza kunabakhuphisana nabo. Ngelixa i-Renaissance Technologies isaziwa ngeendlela zayo zokurhweba ngobungakanani, kubalulekile ukugcina engqondweni isicwangciso sayo esibanzi esiquka amaxesha ahlukeneyo okubamba ukusuka kwiinkqubo ze-HFT zemveli, ezigxile kakhulu kwisantya.
I-LIME (IiNkcazo zeNdawo eziTolika iModeli-agnostic) yindlela ebalaseleyo ye-XAI esetyenziselwa ukwenza iziphumo zeemodeli zokufunda ezintsonkothileyo ziqondeke ngakumbi. Kulawulo lweepotfoliyo, le ndlela inokubaluleka kakhulu ekutolikeni indlela iimodeli zebhokisi ezimnyama zenza izibikezelo. Ngokusebenzisa idatha yokufaka kunye nokuhlalutya impembelelo kwiziphumo zemodeli, i-LIME inceda abaphathi beepotfoliyo kunye nososayensi bedatha bachaze ukuba zeziphi iimpawu ezichaphazela izigqibo zotyalo-mali ngaphezu kwabanye.
Le nkqubo inceda ekwandiseni ukucaca kwezigqibo ezonyuswe yi-AI kwaye ixhasa iinzame zokuqinisekisa nokuphucula indlela ekulula ngayo ukuyiqonda le mifuziselo. Nangona kunjalo, ngelixa i-LIME iphucula ukuqonda kwethu imodeli yokuziphatha, ukuvavanya ukuthembeka kukonke kweemodeli kubandakanya iindlela zokuqinisekisa ezongezelelweyo.
I-AI tech idlala indima enkulu ekuqinisekiseni ukuthotyelwa kwemigaqo yolawulo kunye nokubeka iliso kwizithintelo zotyalo-mali ngaphakathi kwishishini lezemali. Ngokuzenzekelayo ezi nkqubo, iinkqubo ze-AI zinceda iifemu zemali zibambelele kwimigangatho yomthetho ngokufanelekileyo, ngokuchanekileyo, kwaye zingangeni engxakini. Le teknoloji ibaluleke kakhulu ekubekeni iliso kuthotyelo kwimiqulu emikhulu yentengiselwano kunye nemisebenzi eyahlukeneyo yeepotfoliyo, apho inokuthi ngokukhawuleza (ngoko nangoko, enyanisweni) ichonge ukutenxa kwiimfuno zolawulo okanye izikhokelo zangaphakathi.
Ngaphezu koko, ukusetyenziswa kwe-AI kunciphisa umngcipheko wempazamo yomntu, ebaluleke kakhulu kwiindawo eziphakamileyo zokulawula apho iimpazamo zingakhokelela kwimiphumo yezomthetho neyemali.
Usetyenziso lwe-AI kuhlengahlengiso oluzenzekelayo lubalulekile ekugcineni ulwabiwo olufanelekileyo lwe-asethi ekuhambeni kwexesha. Bangakwazi ukulungelelanisa iipotfoliyo ekuphenduleni utshintsho lwemarike okanye utshintsho kwiprofayili yomngcipheko womtyali-mali, oqinisekisa ukulungelelaniswa neenjongo zobuchule botyalo-mali.
Ukongeza kwizicelo ezilungiselelwe ngokukodwa utyalo-mali, amandla okuphuhliswa kobukrelekrele bokwenziwa ngaphakathi kwishishini lolawulo lwee-asethi lubonakala lukhulu. Nangona kunjalo, nangona sibona ngethuku ukuba nokwenzeka kokuzenzekelayo imisebenzi ethile kwizigaba ezahlukeneyo zekhonkco lokusebenza, kusenzima ukulindela ngokupheleleyo amandla okuphazamisa ubukrelekrele bokwenziwa. Oku kungenxa yokuba i-AI kulindeleke ukuba ivelise amacandelo amatsha ezicelo njengoko kuphuhliswa inkqubela phambili eyongezelelweyo.
Kufuneka siyilumkele imida yobukrelekrele bokwenziwa kunye neengozi obuzibangelayo kwimiba ethile yolawulo lwepotfoliyo, nangona yenze ukuba kube nokwenzeka ukuqhubela phambili kobuchwepheshe kunye neenzuzo zemveliso kusetyenziswa ubukrelekrele bokwenziwa. Kwindawo yokuqala, ubukrelekrele bokwenziwa kunye neendlela zokufunda koomatshini zixhomekeke kwidatha esetyenziselwa ukondla i-algorithms yokufunda.
Kuyimfuneko ukuba le datha ikumgangatho ophezulu ngokwemigaqo yohlaziyo, ukuchaneka, ukugqibelela, kunye nokumelwa.
Ukongeza kwimfuno yomthamo omkhulu kakhulu wedatha, engasoloko ikhona, yimeko yokuba le datha kufuneka ibe semgangathweni. Nakweyiphi na enye imeko, ukufunyaniswa okufunyenwe kusetyenziswa iimodeli zokuxela kwangaphambili akuthembekanga okanye kunokomelela.
Ngaphezu koko, ii-algorithms zinokwenza neengqikelelo ezingeyonyani ngokukhetha iintsingiselo ezingabalulekanga kwiseti yedatha ehlalutywayo, enokukhokelela kwizigqibo eziphosakeleyo. Oku kunokubangela ukubanjwa okugqithisileyo, imitsi ebukhali kakhulu, kunye nokona kuntlitheka kuncinci kunokwenzeka. Ukulahleka kokhuphiswano lweemarike kunokuthi kwenzeke ngenxa yokuba abaninzi abaqhubi beemarike abalawula i-algorithms ye-AI efanayo banokwenza isigqibo esingalunganga ngaxeshanye okanye baphendule ngendlela efanayo kwimeko yexesha langempela. Umngcipheko onjalo unokufa.
Ngaphandle kweenzuzo ezinokubakho ze-AI kulawulo lweepotfoliyo, njengakuyo nayiphi na intsimi, kukho imingeni eninzi ekufuneka siyikhumbule kwaye ekugqibeleni - idilesi. Obunye ubunzima obuphambili kukunqongophala okunokwenzeka kokungafihli kunye nokutolika imiba ye-AI, enokwenza kube nzima kubaphathi ukuchaza iziphumo zentsebenziswano yabo ne-AI. Olu bunzima bokusebenzisa lunokuba sesinye sezizathu zokuba ukwamkelwa kwe-AI kwiimali zaseYurophu kuphantsi. Ukusukela ngoSeptemba 2022,
I-European Financial Markets Authority (ESMA)
Okwangoku, kubonakala ngathi ubukrelekrele bokwenziwa kusekude ekuthatheni indawo yabantu bokwenyani kushishino lolawulo lwempahla. Oko kuthethiweyo, ukungafihli, ubudlelwane bokuthembana, kunye noqhagamshelwano phakathi kwabathengi kunye neengcali zolawulo ziyaqhubeka ukuba ziimpawu ezibalulekileyo, ngoku kunanini na ngaphambili.
Nangona kunjalo, asinakukhanyela ukuba ubukrelekrele bokwenziwa buzisa izixhobo ezitsha nezinika umdla ezinokuthi zisetyenziswe kwikhonkco lexabiso, kwaye amandla ezi zixhobo anokutshintsha ngokwenene indlela elijongeka ngayo ishishini namhlanje.