Ukuhlolwa kokutholwa kwama-factor e-futures: kusukela ekufakweni kwe-Docker kuya ekuhlolweni kabusha kwama-factor nasekuhwebeni kokulingisa

Isingeniso sephrojekthi

Ukuhlolwa kokutholwa kwama-factor e-futures kuyinkundla elungele ukusetshenziswa yocwaningo lwe-quantitative kanye nokuhweba kokulingisa kuma-futures. Ihlanganisa ukubuka idatha yemakethe, ukutholwa kwama-factor, ukuhlolwa kabusha kwamasu, ukuhweba nge-akhawunti yephepha, imisebenzi ye-quantitative yesikhathi sangempela kanye nosizo lwe-AI ekuhlaziyeni imakethe ohlelweni olulodwa oluzisingathayo, olufanele ukufunda, ucwaningo kanye nokuhlolwa kokulingisa.

Iphrojekthi inikeza ngokuzenzakalelayo idatha yemakethe yamahhala ye-Sina Finance, evuselelwa cishe njalo ngemizuzwana engu-5, ngaphandle kwesidingo sokubhalisa i-akhawunti yedatha yemakethe. Isizindalwazi, i-cache kanye nedatha yebhizinisi kugcinwa ngumuntu ofaka uhlelo; uma une-front-end ehambisana ne-CTP, ungakwazi nokulungisa ikheli elihambisanayo le-tcp:// kuphaneli yokuphatha.

Le phrojekthi ingeyokufunda, ucwaningo kanye nokuhlolwa kokuhweba kokulingisa kuphela, futhi ayisona iseluleko sokutshala imali. Lapho uxhuma idatha yemakethe yangempela noma izixhumi zokuhweba, umsebenzisi uthwala yedwa ubungozi nomthwalo wokuhambisana nemithetho.

Imisebenzi eyinhloko

• Itheminali yedatha yemakethe: Inikeza amanani esikhathi sangempela amahhala e-Sina kanye nomlando wama-K-line, futhi isekela uhlu lwezinto ozithandayo nesihlahla sezimpahla; amandla afana nebhuku lama-oda lamazinga amahlanu ancike emthonjeni wedatha oxhunyiwe.
• Ilabhorethri yama-factor: Inezici ezingaphezu kuka-60 ezakhelwe ngaphakathi, futhi ivumela ukutholwa kwama-factor ngezisho kanye nokuphindaphinda ukuhlolwa kusetshenziswa i-token encoding.
• Ukuhlolwa kabusha komlando: Isekela ukuhlolwa kabusha kwama-K-line ezikhathi eziningi, ukuhlolwa kabusha kwama-K-line akhiqizwe yi-AI kanye nokuhlolwa kabusha kwamamodeli ezinqumo.
• Ukuhweba kokulingisa: Inikeza i-akhawunti yephepha, ukufaniswa kwama-oda alindile, ukuphathwa kwezikhundla, ijika le-equity kanye nemigomo eminingi yokuvala izikhundla.
• Imisebenzi ye-quantitative: Ibala amasignali wamasu ngesikhathi sangempela, futhi isekela ukuvumelana kwezikhathi eziningi kanye nokumiswa kokulahlekelwa kusetshenziswa ama-K-line esignali.
• Ukuhlaziya imakethe nge-AI: Ikhiqiza imibiko yemakethe yesitayela somsakazi kanye nolwazi olusiza ekuhwebeni ngeziteshi ze-LLM ezilungiselelwe ngumsebenzisi.
• Iphaneli yokuphatha: Iphatha ukuqapha iziteshi zedatha yemakethe nokushintsha phakathi kwesevisi eyinhloko neyesipele, imithombo yama-K-line, abasebenzisi kanye nama-quota.

Ukwakhiwa kobuchwepheshe

Ingxenye yangemuva isekelwe ku-Python 3.12, FastAPI, asynchronous SQLAlchemy, PostgreSQL kanye ne-Redis; ingxenye yangaphambili isebenzisa i-Next.js 15, React 19 kanye ne-Tailwind CSS; zonke izinsiza zihlelwa nge-Docker Compose.

Lokhu kwakhiwa kufaka i-frontend, i-API, isizindalwazi kanye ne-cache ekufakweni okukuma-container, futhi kulungele ukuzisingathela kuseva yomuntu siqu noma endaweni yangaphakathi yocwaningo.

Ukufaka nokuqalisa

Isinyathelo sokuqala: Lungisa indawo yokufaka

Ngaphambi kokuqala, qinisekisa ukuthi i-Docker ifakiwe endaweni yokusebenza futhi i-Docker Compose iyatholakala. Bese uthola ikhodi yomthombo yephrojekthi bese ungena empandeni ye-repository.

Isinyathelo sesibili: Dala ifayela leziguquguquko zemvelo

Kopisha ukucushwa kwesibonelo okunikezwe yi-repository:

cp .env.example .env

Vula i-.env, uhlole noma uguqule ukucushwa kwesizindalwazi, i-Redis kanye nezihluthulelo ze-JWT ngokwendawo yokufaka. Uma ufaka esidlangalaleni, akufanele uqhubeke usebenzisa izihluthulelo okulula ukuziqagela noma ezifana nalezo zesibonelo.

Isinyathelo sesithathu: Qalisa ama-container

Qalisa lokhu empandeni ye-repository:

docker compose up -d

Lo myalo uzoqalisa ngemuva izinsiza ezichazwe ku-Compose. Ngemva kokuqalisa, ungasebenzisa lo myalo ukubuka isimo sama-container:

docker compose ps

Imiyalelo yokwakha amatafula esizindalwazi, ukucushwa kwedatha kanye nencazelo ephelele yeziguquguquko zemvelo kufanele ilandelwe ku-SELF-HOSTING.md yephrojekthi. I-README ayithembisi imbobo yokufinyelela emisiwe, ngakho ikheli langempela kufanele linqunywe ngokwefayela le-Compose kanye neziguquguquko zemvelo zasendaweni.

Ukulungisa umthombo wedatha yemakethe

Sebenzisa ngokuqondile idatha ezenzakalelayo ye-Sina

Ekuzameni kokuqala ungagcina umthombo wedatha ozenzakalelayo we-sina. Lo mthombo awudingi i-akhawunti noma ikheli leseva, uvuselela ngokuzenzakalelayo cishe njalo ngemizuzwana engu-5, futhi ulungele ukuqinisekisa ngokushesha ikhasi ledatha yemakethe kanye nenqubo eyisisekelo yocwaningo.

Xhuma i-front-end yakho ehambisana ne-CTP

Uma usuvele unesevisi ethembekile yedatha yemakethe ehambisana ne-CTP, ungakhetha i-ctp_sim, bese ugcwalisa ikheli lakho ekhasini elithi “Ukuqapha iziteshi zedatha yemakethe” kuphaneli yokuphatha, isibonelo:

tcp://host:port

I-repository yomthombo ovulekile ayifaki amakheli amaseva edatha yemakethe akhokhelwayo noma ayimfihlo; umuntu ofaka uhlelo kufanele azitholele futhi alungise ikheli le-front-end elisemthethweni nelisebenzayo.

Sebenzisa indawo yokulingisa ye-openctp 7×24

openctp_724 ihambisana nendawo yokulingisa engu-7×24 ye-openctp, futhi ilungele ukuhlolwa kwemakethe ebonakalayo. Imithombo yedatha ehlukene inganikeza izinkambu namazinga edatha ahlukene; isibonelo, ibhuku lama-oda lamazinga amahlanu lidinga ukuthi umthombo wedatha ube nalelo khono.

Inqubo eyisisekelo yokusebenzisa

1. Khetha into yocwaningo ku-terminal yemakethe

Ngemva kokungena ku-terminal yemakethe, buka amanani ama-futures kanye nomlando wama-K-line ngohlu lwezinto ozithandayo noma ngesihlahla sezimpahla. Qinisekisa kuqala ukuthi idatha iyavuseleleka ngaphambi kokuqala ukuhlolwa kwama-factor noma ukuhlolwa kabusha, ukuze unciphise iziphetho ezingalungile ezibangelwa izinkinga zomthombo wedatha.

2. Yenza ukuhlolwa elabhorethri yama-factor

Ilabhorethri yama-factor inikeza izici ezakhelwe ngaphakathi ezingaphezu kuka-60 futhi isekela ukuhlanganisa ama-factor ngezisho. Kunconywa ukuthi uqale ngezisho ezilula, ubhale phansi impahla yocwaningo, isikhathi kanye nemingcele, bese ukhuphula ubunkimbinkimbi kancane kancane.

I-token encoding yenkundla ingasetshenziselwa ukuphindaphinda ukuhlolwa. Ngemva kokuqeda ucwaningo oluwusizo lwe-factor, kufanele ugcine i-token ehambisanayo futhi ubhale phansi umthombo wedatha kanye nebanga lesikhathi elisetshenzisiwe. Lokhu kwenza kube lula ukuqinisekisa ukuthi izimo zokuhlola ziyafana lapho uphinda usebenzisa noma uqhathanisa imiphumela.

3. Qalisa ukuhlolwa kabusha komlando

Ngena kumodule yokuhlolwa kabusha, bese ukhetha ukuhlolwa kabusha kwama-K-line ezikhathi eziningi, ukuhlolwa kabusha kwama-K-line akhiqizwe yi-AI noma ukuhlolwa kabusha kwemodeli yesinqumo ngokwenhloso yocwaningo. Ngesikhathi sokuhlolwa kabusha, hlola ikakhulukazi ukuthi isikhathi esikhethiwe, umqondo wesignali kanye nemibandela yokuvala izikhundla kuyahambisana yini nombono wokuqala.

Ungabheki kuphela umphumela wenzuzo owodwa. Njengoba i-README ingacacisi izinkomba ezithile zokuhlola noma imingcele ezenzakalelayo, ukuhlaziya kwangempela kufanele kusekelwe emiphumeleni yokuhlolwa kabusha eboniswa yisistimu futhi kuqhathaniswe nezikhathi nezimo ezahlukene.

4. Dlulela ekuqinisekiseni ngokuhweba kokulingisa

Ngemva kokuhlolwa kabusha komlando, ungasebenzisa i-akhawunti yephepha ukuze uqinisekise ngokulingisa. Imodule yokuhweba kokulingisa isekela ukufaniswa kwama-oda alindile, ukubuka izikhundla kanye nejika le-equity, futhi ingasebenzisa izindinganiso zokuvala izikhundla ezifana ne-death cross yezinkomba, umkhawulo we-factor noma ireferensi yokunyakaza kwentengo.

Kunconywa ukuthi uqale ubuke ukusebenza kwamasignali ezimakethe eziqhubekayo usebenzisa i-akhawunti yokulingisa, kunokuthatha iziphetho zokuhlolwa kabusha komlando njengezibikezelo zokusebenza kwesikhathi esizayo. Kufanele uhlole ngokukhethekile ukuthi isimo sama-oda alindile, izinguquko zezikhundla kanye nejika le-equity kuyahambisana yini nalokho okulindelwe yisu.

5. Dala umsebenzi we-quantitative wesikhathi sangempela

Imisebenzi ye-quantitative ingabala amasignali wamasu ngesikhathi sangempela futhi isekele ukuvumelana kwezikhathi eziningi kanye nokumiswa kokulahlekelwa ngama-K-line esignali. Ekusetshenzisweni kwangempela, ungaqala ngokuqinisekisa isignali ngesikhathi esisodwa, bese wengeza ezinye izikhathi ukuze usebenzise ukuhlunga kokuvumelana, okusiza ukuhlukanisa umthelela wesignali eyisisekelo nowemibandela yezikhathi eziningi.

6. Lungisa umsebenzi we-AI wokuhlaziya imakethe

Umsebenzi we-AI wokuhlaziya imakethe uncike eziteshini zamamodeli ezilungiselelwe ngumuntu ofaka uhlelo. Ngemva kokulungisa isiteshi se-LLM, isistimu inganikeza imibiko yemakethe yesitayela somsakazi kanye nolwazi olusiza ekuhwebeni.

Okukhishwa yi-AI kufanele kube usizo locwaningo, hhayi isiphetho sokuhweba esiqinisekile. Lapho usebenzisa ama-K-line akhiqizwe yi-AI noma incazelo yemakethe, kufanele uqinisekise ngokuzimela usebenzisa idatha yemakethe yangempela, imiphumela yokuhlolwa kabusha kanye nokusebenza kwe-akhawunti yokulingisa.

Izincomo zokusebenzisa ezithuthukile

Gcina ukuhlolwa kuphindaphindeka

• Gcina i-token encoding ehambisana nesisho se-factor.
• Bhala phansi impahla, isikhathi sama-K-line, umthombo wedatha kanye nebanga lesikhathi lokuhlola.
• Lapho uqhathanisa ukuhlolwa, zama ukushintsha i-variable eyinhloko eyodwa kuphela.
• Ngemva kokushintsha umthombo wedatha, hlola kabusha izinkambu zedatha kanye namandla ebhuku lama-oda.

Qinisekisa isu ngezigaba

1. Yakha futhi uhlole ama-factor elabhorethri yama-factor.
2. Qinisekisa ukusebenza ezikhathini ezahlukene ngokuhlolwa kabusha komlando.
3. Hlola inqubo yama-oda alindile, izikhundla kanye nokuvala izikhundla usebenzisa i-akhawunti yephepha.
4. Dala umsebenzi we-quantitative bese ubheka amasignali esikhathi sangempela nokuvumelana kwezikhathi eziningi.
5. Sebenzisa umsebenzi we-AI wokuhlaziya imakethe njengolwazi olwengeziwe, hhayi esikhundleni semithetho yesu.

Thuthukisa ukutholakala ngokuphatha iziteshi zedatha yemakethe

Uma ulungise iziteshi zedatha yemakethe eziningi, ungasebenzisa iphaneli yokuphatha ukuqapha iziteshi nokushintsha phakathi kwesevisi eyinhloko neyesipele. Hlola isimo sokuxhumana njalo futhi uqinisekise ukuthi isiteshi esiyisipele sinekheli lesevisi elilungile nelifinyeleleka.

Vikela indawo ozisingathela yona

Ngaphambi kokufaka, shintsha isihluthulelo se-JWT futhi uvikele kahle imininingwane yokuxhumana ye-PostgreSQL, Redis, i-front-end yemakethe kanye neziteshi zamamodeli. Abasebenzisi nama-quota bangaphathwa ngephaneli yokuphatha ukuze kugwenywe ukusetshenziswa kwezinsiza okungadingekile noma ukufinyelela okungagunyaziwe.

Imiyalo evamile yokusebenza nokugcinwa

Buka isimo samanje sama-container:

docker compose ps

Buka amalogi ezinsiza:

docker compose logs

Landela amalogi ngokuqhubekayo:

docker compose logs -f

Misa izinsiza ze-Compose:

docker compose down

Uma udinga ukulungisa i-database, i-Redis, i-JWT noma okunye ukucushwa, qala ubheke imibhalo yokuzifakela yephrojekthi bese usebenza ngokuvumelana nokucushwa kwamanje kwe-Docker Compose.

Isiphetho

Ukuhlolwa kokutholwa kwama-factor e-futures kunikeza indlela ephelele yocwaningo lokuzisingathela: iqala ngedatha yemakethe yamahhala nokwakha ama-factor, iqhubeke nokuhlolwa kabusha kwezikhathi eziningi nokuhweba kokulingisa, bese inabela emisebenzini ye-quantitative yesikhathi sangempela kanye ne-AI yokuhlaziya imakethe. Kunconywa ukuthi uqale uqede inqubo eyisisekelo usebenzisa idatha ezenzakalelayo ye-Sina, bese uxhuma i-front-end yakho yedatha yemakethe neziteshi zamamodeli ngokwezidingo, futhi uhlale uqinisekisa amasu ngokucophelela ngaphakathi kocwaningo nokulingisa.