Ukuphathwa Kwephothifoliyo: Zonke Izindlela I-AI Eshintsha Ngayo Amasu Empahla Yesimanje

Ukukhuphuka kwe-AI ngokusobala kube nomthelela ezimbonini ezahlukahlukene, futhi imboni yezezimali iphakathi kwaleyo ethinteke kakhulu . Isibonelo, ukwethulwa esidlangalaleni kwamamodeli afana ne-GPT-3.5 ngonyaka odlule kukhulise intshisekelo yokusebenzisa i-AI ukusiza ukukhulisa amakhono abaphathi bezikhwama ekuhlaziyeni, ekulawuleni ubungozi, nasekuthatheni izinqumo.

Ngakho-ke, amathuluzi e-AI asetshenziswa ukwenza ukuhlolwa kwemakethe kunembe kakhudlwana futhi kulawuleke ubungozi ngempumelelo kakhudlwana. Abaphathi bephothifoliyo kulindeleke ukuthi benze ukuhlaziya okucacile kokuhamba kwemakethe, banciphise ukukhetha okufanelekile kokutshalwa kwezimali, futhi balawule izingozi lapho besebenzisa ama-algorithms okufunda ngomshini, ukucutshungulwa kolimi lwemvelo, namathuluzi obuhlakani bokwenziwa ekuhwebeni kwabo.

Ukuhlanganiswa kwama-algorithms okufunda komshini, kanye namathuluzi okucubungula ulimi lwemvelo kumasu okuhweba abadlali ababalulekile, kubasiza ukuthi bakhulise ukusebenza kahle kwalezi zinqubo futhi bathole inzuzo yokuncintisana ngezinqumo zokutshala izimali ezisheshayo nezinembe kakhulu kanye nezibalo eziqagelayo.

Emashumini eminyaka adlule, i-AI isetshenziswe emikhakheni ehlukene yemboni yezezimali. Ehhovisi elingemuva, ama-algorithms e-ML asetshenziselwa ukuthola okudidayo kumalogi okubulala, ukuthola okwenziwayo okusolisayo, kanye nokuphatha ubungozi, okuholela ekwengezeni ukusebenza kahle nokuvikeleka. Ehhovisi elingaphambili, i-AI isiza ukuhlukanisa amakhasimende, yenze ngokuzenzakalela izinqubo zokusekelwa kwamakhasimende, futhi ithuthukise amanani entengo.

Kodwa-ke, ingxenye yayo ethakazelisa kakhulu amandla e-AI ohlangothi lwezezimali - ukuhlonza amasignali abikezelayo phakathi nomsindo wemakethe ngokuhlaziya amanani abalulekile edatha ngokushesha okukhulu. Isibonelo, izinhlelo zokusebenza ezinjalo zingase zifake ukubikezela kochungechunge lwesikhathi, izimakethe zokuhlukanisa, kanye nokuphatha amaphothifoliyo wempahla. Amathuba e-AI okucubungula nokuhlaziya amasethi edatha asiza ukuthola amaphethini acashile okungenzeka ukuthi izindlela zendabuko zingageja.

Ukwenziwa ngcono kwephothifoliyo kube umkhuba ojwayelekile amashumi eminyaka ambalwa, uvela ngokuphawulekayo ngaphansi kokuthuthukiswa kwesayensi yedatha kanye nokuqaliswa kwamasu okuhlanganisa athuthukile. Izindlela Сlassical, ezifana ne-Markowitz's Modern Portfolio Theory (1952) kanye ne-Capital Asset Pricing Model (1964) zethulwa eminyakeni engaphezu kwengu-50 edlule kodwa zisasebenza. Kodwa-ke, ukulinganiselwa kwabo ekusingatheni ubungozi obungewona umugqa kanye nokuncika kudatha yomlando kuya kuba sobala kakhulu usuku nosuku.

Imikhuba efana nokumodela ubungozi, ukuhlaziya isimo, nokuhweba ngamanani, okusetshenziswa kabanzi ngabadlali ababalulekile, njenge-Renaissance Technologies, i-DE Shaw, kanye ne-Two Sigma Investments kuholele ekusetshenzisweni kwe-algorithms eyinkimbinkimbi futhi ethuthuke kakhulu. Ngaphezu kwalokho, imboni ithinteke kakhulu i-AI eminyakeni yamuva, njengoba ukufunda komshini nobuhlakani bokwenziwa kwenze ukuhlaziya okubikezelayo kwanembe kakhudlwana, futhi kwenza okufanayo kumasu okutshalwa kwezimali komuntu siqu kanye nezinqubo ezizenzakalelayo zokuthatha izinqumo eziyinkimbinkimbi.

Lolu shintsho oluqhutshwa yi-AI lunike amandla abaphathi bephothifoliyo ukuthi bacubungule uxhaxha lwedatha ngesikhathi sangempela futhi baxazulule izinselele ezintathu ezibalulekile:

• Ukuqina: Ukuphatha nokuhlaziya idatha yezinga elikhulu elivela ezimpahleni eziningi nezimakethe zomhlaba manje sekulula kakhulu ukukwenza.

• Ukwenza Izinqumo Eziyinkimbinkimbi: I-AI “ingagcina engqondweni” izici eziningi, okuhlanganisa nokuhlaziya kwengqondo nokuziphatha, ezinqubweni zokwenza izinqumo.

• Ukuzivumelanisa nezimo: Amasistimu e-AI angafunda angami futhi azivumelanise nezimo zemakethe ezintsha, asize abaphathi ukuthi balungise ngokushesha amasu.

Ngokuvumelana ne I-Global Market Insights , I-AI emakethe Yokuphathwa Kwempahla ibilinganiselwa ku-USD 2.5 billion futhi kulindeleke ukuthi ikhule ku-CAGR ka-24% eminyakeni eyi-10 ezayo. Kuyathakazelisa ukuthi i-Portfolio Optimization ihola ekuhlukaniseni imakethe yomhlaba wonke ngokusetshenziswa, kulandelwa Ukuhlaziywa Kwedatha, ukubalwa kwezimali. U-25% wesabelo semakethe .

Ukwenyusa ukwamukelwa kanye nokutshalwa kwezimali ezisombululweni zokuphathwa kwempahla okunikwa amandla yi-AI nokugqamisa ukusetshenziswa okungokoqobo kwe-AI ekwenzeni iphothifoliyo.

Ukutholwa kwe-AI Ekuphathweni Kwephothifoliyo:

Ukwamukelwa kwe-AI ngaphakathi kwemboni yokuphathwa kwempahla akuyona inkambiso entsha; ibone ukukhula eminyakeni yamuva kodwa isakhawulelwe enanini elincane labadlali bezimakethe okuyi-hedge funds, amahhovisi okuphatha amanani, iminyango emikhulu yocwaningo, nezikhungo zezezimali ezisebenzisa izinsizakalo ze-IT.

Kukhona izinkambu eziningi zokufaka isicelo se-AI kakade:

Ukuthuthukisa Iphothifoliyo

I-AI yenza ngcono kakhulu inqubo yokwakhiwa kwephothifoliyo. Isibonelo, indlela yakudala ye-Modern Portfolio Theory kaMarkowitz, encike emicabangweni yokuthuthukisa i-convex, isebenza njengesandulela sezindlela zesimanje eziqhutshwa yi-AI. Isizathu sokuthi le theory yesisekelo ibaluleke kakhulu ukuthi yakha isisekelo lapho ama-algorithms e-AI angaqhubeka nokushintsha futhi acwengisise amasu okutshala imali.

Namuhla, i-AI inweba kulo mbono ngokuhlola ubukhulu obusha bedatha nokuhlanganisa amasu okuhlaziya athuthukile. Leli khono ledatha elinwetshiwe livumela ukuthathwa kwezinqumo okuguquguqukayo nokunolwazi - umkhuba osetshenziswe kabanzi embonini.

Ukuhlaziya Okuyisisekelo

Amasu athile e-AI ahambisana ngokuphelele nokuphathwa komthamo, kusetshenziswa imiqulu emikhulu yedatha emayelana nezisekelo zenkampani, imvelo yomnotho omkhulu, noma izimo zemakethe. Ama-algorithms wokufunda komshini angathola ubudlelwano obuyinkimbinkimbi obungewona umugqa phakathi kokuhlukahluka okuhlukile futhi, vele, athole amathrendi abahlaziyi abangakwazi ukuwathola.

Ukuhlaziywa Kwemibhalo

Ukuhlaziywa kombhalo kungolunye uhlelo lwe-AI ekuhlaziyeni okuyisisekelo. Isebenzisa ukucutshungulwa kolimi lwemvelo (NLP), i-AI icubungula futhi ihlaziye imithombo yombhalo njengemibiko yezinzuzo zezinkampani, ukukhishwa kwabezindaba kwebhange eliphakathi, nezindaba zezimali. Nge-NLP, i-AI ingakhipha ulwazi olubalulekile kwezomnotho nakwezezimali kule datha engahlelekile. Ngokwenza kanjalo, ihlinzeka ngesilinganiso sobuningi nesihlelekile esithuthukisa futhi sisiza ukuhumusha komuntu.

Ukuhweba Imisebenzi

Amandla e-AI awusizo kakhulu ekuhwebeni, lapho ubunkimbinkimbi bokwenziwayo kanye nesidingo sesivinini kubhalansile. I-AI isekela ukuhweba kwe-algorithmic ngokwenza ngokuzenzakalelayo izigaba eziningi zenqubo, ithuthukise ukusebenza kahle kwentengiselwano ephethwe ezimakethe zezimali.

Amasevisi Okweluleka Ngokutshalwa Kwezimali

I-AI ivule ithuba lokunikezwa okubanzi kwezinsizakalo zokweluleka ngokutshalwa kwezimali komuntu siqu ngezindleko eziphansi. Lezi zinhlelo zisebenzisa ama-algorithms ayinkimbinkimbi ukucubungula idatha yemakethe yesikhathi sangempela, eqhamuka namasu afaneleka kakhulu ezidingo zeklayenti ngalinye ngokusekelwe ezinhlosweni zawo zokubuyisela kanye namaphrofayili engcuphe.

Ukulawulwa Kwengozi

Ekulawulweni kobungozi, i-AI isiza ngokwenza imodeli yezimo ezihlukahlukene 'ezingenzeka kodwa ezingafuneki', ezithuthukisa izinqubo zendabuko ezigxile kuphela emiphumeleni engenzeka.

Amasu Namathuluzi Obuhlakani Be-Artificial Intelligence (AI) Ekuphathweni Kwephothifoliyo

Ama-algorithms wokufunda ngomshini:

Izindlela Zokufunda Zomshini Wakudala zisadume kakhulu Ekuphathweni Kwephothifoliyo, futhi yilezi: Amamodeli Awumugqa, afaka Izikwele Ezivamile Ezincane, Ukuhlehla Kwe-Ridge, kanye Ne-Lasso Regression. Lokhu kuvame ukuhlanganiswa nenqubo ye-Mean-Variance Optimization kanye nezindlela zokuwohloka kwe-matrix ezifana nokuwohloka kweValue Eyodwa (SVD) kanye Nokuhlaziywa Kwengxenye Eyinhloko (PCA), okuyisisekelo ekuqondeni ubudlelwano bempahla nokuthuthukisa ukwabiwa kwephothifoliyo.

Ibekwe phakathi kwalezi zindlela zakudala kanye nezindlela zesimanjemanje iMishini Yokusekela Vector (ama-SVM). Nakuba ama-SVM esetshenziswa ekusebenzeni, awavamile ukuthunyelwa kodwa adlala indima ebalulekile, ikakhulukazi, emisebenzini yokuhlukanisa ehloselwe ukubikezela ukusebenza kwesitoko.

Le misebenzi ngokuvamile ihlanganisa ukubikezela ukuthi isitoko sizothola inzuzo noma ukulahlekelwa, kusetshenziswa idatha yomlando yezezimali ehlanganisa ukushintshashintsha kwenani lesitoko namavolumu okuhweba ukuze kubekwe izimpahla ngezigaba futhi zibikezele ukusebenza kwazo.

Uma sikhuluma ngezindlela zesimanjemanje, amanethiwekhi e-neural abonisa intuthuko enkulu ekufundeni komshini ekuphathweni kwephothifoliyo futhi anikezela ngamakhono athuthukisiwe okumodela amaphethini ayinkimbinkimbi angewona alayini okunzima ukuwathwebula ngamamodeli endabuko. Ngaphandle kwamanethiwekhi e-neural, ezinye izindlela zakudala ezifana nokufunda okugadiwe nokungagadiwe ziqhubeka zithuthukisa futhi zicwengisise ukuhlaziya idatha, okwenza ukutholakala nokuxhashazwa kwezimpawu zemakethe ezicashile kwenzeke.

Izindlela ezintsha, ezifana ne-Reinforcement Learning kanye ne-Deep Q-Learning ziletha lezi zimfanelo ezindaweni zokwenza izinqumo ezisheshayo, lapho amaphothifoliyo angalungiswa ngesikhathi sangempela ukuze kuthuthukiswe imiphumela yezezimali ngokusekelwe ekufundeni kwesistimu kusukela kumpendulo yemakethe.

Ukucubungula Ulimi Lwemvelo (NLP):

Amasu Okucubungula Ulimi Lwemvelo njengokuhlaziya imizwa angasiza ukukhetha nokukhetha imibono evamile ezintweni ezifana nezihloko zephephandaba, okuthunyelwe kwenkundla yezokuxhumana, nemibiko yomhlaziyi. Ukwengeza, abaphathi bephothifoliyo bangaphinda bahlaziye ulimi olusetshenziswa emithonjeni yezezimali, okuhlanganisa nemibiko yemali etholwa amafemu, ukuzwa imizwa yabatshalizimali futhi babikezele umnyakazo wezimakethe, konke okuwulwazi olubalulekile enqubweni yokwenza izinqumo.

Amasu Okuhweba Okulinganiselwe:

Amafemu agxile ekuhwebeni kwe-high-frequency trading (HFT), njengalawo asebenzisa ama-algorithms okuhweba amanani anamandla e-AI, enza imali ngokungasebenzi kahle okwenzeka isikhashana nje emakethe. Lezi zinkampani zisebenzisa ubuchwepheshe bokufunda komshini ukuze zihlaziye ulwazi lwemakethe olufanele ngesivinini esiphezulu kakhulu futhi zenze ama-oda anemba isikhathi esifushane njenge-millisecond.

Ukubulawa okusheshayo okunjalo kubavumela ukuthi bazuze emathubeni e-arbitrage futhi bakhulise inzuzo ngokuthatha isinyathelo sokungafani kwamanani ngokushesha kunabaqhudelana nabo. Nakuba i-Renaissance Technologies yaziwa ngezindlela zayo zokuhweba ngobuningi, kubalulekile ukukhumbula isu layo elibanzi elihlanganisa izikhathi zokubamba ezihlukahlukene kusukela kumikhuba yendabuko ye-HFT, egxile kakhulu esivinini.

I-AI echazekayo (XAI):

I-LIME (Izincazelo Zemodeli Ye-agnostic Yasendaweni) iyindlela evelele ye-XAI esetshenziselwa ukwenza okuphumayo kwamamodeli okufunda omshini ayinkimbinkimbi kucace kakhudlwana. Ekulawulweni kwephothifoliyo, le ndlela ingabaluleka kakhulu ekuchazeni ukuthi amamodeli ebhokisi elimnyama enza kanjani izibikezelo. Ngokusebenzisa idatha yokufaka nokuhlaziya umthelela emiphumeleni yemodeli, i-LIME isiza abaphathi bephothifoliyo nososayensi bedatha ukuchaza ukuthi yiziphi izici ezithonya izinqumo zokutshalwa kwezimali ngaphezu kwabanye.

Le nqubo isiza ukuthuthukisa ukubonakala kwezinqumo ezikhuthazwe yi-AI futhi isekela imizamo yokuqinisekisa nokuthuthukisa ukuthi zingaba lula kangakanani ukuqonda lawa mamodeli. Kodwa-ke, ngenkathi i-LIME ithuthukisa ukuqonda kwethu ukuziphatha kwemodeli, ukuhlola ukwethembeka kukonke kwamamodeli kuhilela amasu okuqinisekisa engeziwe.

I-AI Ekuthobeleni Nokuqapha:

Ubuchwepheshe be-AI budlala indima enkulu ekuqinisekiseni ukuhambisana nezinhlaka zokulawula kanye nokuqapha imikhawulo yokutshalwa kwezimali ngaphakathi kwemboni yezezimali. Ngokuzenzakalela lezi zinqubo, amasistimu e-AI asiza amafemu ezezimali ukuthi anamathele ezindinganisweni zomthetho ngempumelelo kakhudlwana, ngokunembe kakhudlwana, futhi angangeni enkingeni. Lobu buchwepheshe bubaluleke kakhulu ekuqapheni ukuthotyelwa kwemithetho kuwo wonke amanani amakhulu okwenziwayo nemisebenzi ehlukahlukene yephothifoliyo, lapho bungakwazi khona ukuhlonza ngokushesha (ngokuphazima kweso, eqinisweni) ukuchezuka ezidingweni zokulawula noma imihlahlandlela yangaphakathi.

Ngaphezu kwalokho, ukusetshenziswa kwe-AI kunciphisa ubungozi bephutha lomuntu, okubalulekile ezindaweni eziphakeme zokulawula lapho amaphutha angaholela emiphumeleni yezomthetho nezezimali.

Ukulinganisa kabusha iphothifoliyo:

Izinhlelo zokusebenza ze-AI ekulinganiseni kabusha okuzenzakalelayo zibalulekile ekugcineni ukwabiwa kwempahla efanele ngokuhamba kwesikhathi. Bangakwazi ukulungisa amaphothifoliyo ekuphenduleni izinguquko zemakethe noma amashifu kuphrofayela yengcuphe yomtshali-zimali, okuqinisekisa ukuhambisana nemigomo yamasu yokutshala imali.

Ngokubuka Okubanzi

Ngokungeziwe ezinhlelweni eziklanyelwe ukutshalwa kwezimali, amandla okuthuthukiswa kobuhlakani bokwenziwa ngaphakathi kwebhizinisi lokuphatha impahla abonakala ebanzi. Kodwa-ke, naphezu kweqiniso lokuthi ngokwemvelo sibona ukuthi kungenzeka ukuzenzela imisebenzi ethile ezigabeni ezihlukahlukene zochungechunge lokusebenza, kusenzima ukulindela ngokugcwele amandla okuphazamisa obuhlakani bokwenziwa. Lokhu kungenxa yokuthi i-AI kulindeleke ukuthi idale imikhakha emisha yokusetshenziswa njengoba kuthuthukiswa ukuthuthukiswa okwengeziwe.

Kufanele siqaphele ukulinganiselwa kobuhlakani bokwenziwa kanye nobungozi obubekayo kwezinye izici zokuphathwa kwephothifoliyo, naphezu kweqiniso lokuthi kwenze kwaba nokwenzeka ukuthuthuka kwezobuchwepheshe kanye nezinzuzo zokukhiqiza kusetshenziswa ubuhlakani bokwenziwa. Okokuqala, ubuhlakani bokwenziwa nezindlela zokufunda komshini zincike kudatha esetshenziselwa ukuphakela ama-algorithms okufunda.

Kudingeka ukuthi le datha ibe yekhwalithi ephezulu ngokwemibandela yezibuyekezo, ukunemba, ukuphelela, nokumelela.

Ngaphezu kwesidingo sevolumu enkulu kakhulu yedatha, engatholakali ngaso sonke isikhathi, kuyindaba yokuthi le datha kufanele ibe yekhwalithi enhle. Kunoma isiphi esinye isimo, okutholakele okutholwa kusetshenziswa amamodeli abikezelayo akuthembekile noma kuqinile.

Ngaphezu kwalokho, ama-algorithms angenza nokuqagela okungamanga ngokukhetha amathrendi angabalulekile kudathasethi ehlaziywayo, okungase kuholele eziphethweni eziyiphutha. Lokhu kungase kubangele ukubamba izikali, ukweqa okucijile kakhulu, kanye nokuphahlazeka okuncane kakhulu okungaba khona. Ukulahlekelwa ukuncintisana kwemakethe kungase kwenzeke ngenxa yokuthi abaqhubi bemakethe abaningi abaphethe ama-algorithms e-AI afanayo bangenza isinqumo esingalungile ngesikhathi esisodwa noma basabele ngendlela efanayo nesimo sesikhathi sangempela. Ingozi enjalo ingaba yingozi.

Ngaphandle kwezinzuzo ezingaba khona ze-AI ekuphathweni kwephothifoliyo, njengakunoma yimuphi umkhakha, kunezinselelo eziningi okufanele sizikhumbule futhi ekugcineni - sizilungise. Obunye bobunzima obuyinhloko ukuntula okungase kube sobala kanye nezindaba zokuhumusha amamodeli e-AI, okungenza kube inselele kubaphathi ukuchaza imiphumela yokusebenzisana kwabo ne-AI. Lokhu kusebenzisa inkimbinkimbi kungase kube esinye sezizathu zokuthi kungani ukwamukelwa kwe-AI ezimalini zaseYurophu kuphansi kakhulu. Kusukela ngoSepthemba 2022, 65 kuphela kwezimali ezingama-22,000 ezinze e-European Union zithi isebenzisa i-AI ezinqubweni zabo zokutshala imali.

I-European Financial Markets Authority (ESMA) ikhombe izici ezingase zibe nomthelela esilinganisweni esiphansi sokutholwa, njengokuntuleka kwezinhlaka ezicacile zokulawula kanye namakhono e-AI phakathi kwabaphathi bezikhwama. Kodwa-ke, inselele yokuchaza imiphumela ye-AI ngenxa yobunkimbinkimbi bemodeli ingase ibe enye yezinto ezithethelela izinga eliphansi lokutholwa. Ngicabanga ukuthi sizothola ngokuhamba kwesikhathi.

Kuleli qophelo, kubonakala sengathi ubuhlakani bokwenziwa kusekude ekuthatheni indawo abantu bangempela embonini yokuphathwa kwempahla. Uma sekushiwo lokho, ukubeka izinto obala, ubudlelwano bokuthembana, nokuxhumana phakathi kwamakhasimende nochwepheshe bokuphatha kuyaqhubeka nokuba yizici ezibalulekile, manje kunangaphambili.

Nokho, asinakuphika ukuthi ubuhlakani bokwenziwa buletha amathuluzi amasha najabulisayo angasetshenziswa ochungechungeni lwenani, futhi amandla alawa mathuluzi angashintsha ngempela indlela imboni ebukeka ngayo namuhla.