InoteveraGaidhi rinotevera
MaModeli Ekukurudzira eMatanho Makumi maviri
Tekinoroji
Nhungamiro yehunyanzvi
GARCH modhi inofembera yenguva-inosiyana-siyana yezvimiso nekuenderana nekusagadzikana kwazvino nekukatyamadzwa kwakapetwa kaviri uye fungidziro yemusiyano yapfuura.
Teerera kugwaro iriAI-yakarondedzerwa · ~3 min
Idzo dzinowanzo shandiswa pakudzoka kwemari ine volatility clustering, asi yakaringana musiyano wekufanotaura zvinoenderana nefungidziro yekugovera, kurongeka kwemuenzaniso uye maitiro ekudzoka achitsanangurwa zvakakwana.
Madzoro emari anowanzo ratidza kusagadzikana kuungana: shanduko huru dzakakura dzinowanzo kuitika pedyo nedzimwe shanduko huru, nepo nguva dzakadzikama dzichienderera mberi. Iyo yekudzosa pachayo inogona kunge iine diki mutsara autocorrelation uku iyo squared magnitude ichiramba ichitsamira. GARCH, kana generalized autoregressive conditional heteroskedasticity, inoenzanisira musiyano wezvimiso sekushanduka nekufamba kwenguva uchishandisa zvinovhundutsa zvekare uye musiyano wekare. A GARCH(1,1) variance equation ndeye h_t = omega + alpha*epsilon_(t-1)^2 + beta*h_(t-1), apo h_t iine conditional musiyano uye epsilon ndiyo tsvakiridzo kubva kunoreva equation. The ARCH izwi alpha rinopindura ichangoburwa squared kuvhunduka; iyo GARCH temu beta inoendesa kumberi musiyano. Kuvhunduka kukuru nekudaro kunogona kuwedzera kufembera kushanduka kunyangwe kana fungidziro isingachinji. Mamodheru emhando yepamusoro anowedzera mamwe anokatyamadza kana misiyano. Iyo modhi inowanzo tora yakamisikidzwa mitsva inotevera kugovera kwakasarudzwa, seyakajairika kana Mudzidzi t. Heavy-tailed returns inogona kuita kuti fungidziro yakajairika ive isina kukwana. Zvipingamupinyi pamiganhu zvinoshandiswa kuchengetedza kusiyana kwakanaka uye kazhinji kukurudzira kumira; chaiyo parameterization zvinoenderana nesoftware. Kusiyana kusingagumi kuripo pasi pemamwe mamiriro ezvinhu, akadai semari yakagadzikana yekuenderera mberi kweGARCH (1,1), asi iyo mamiriro inofanira kuongororwa pane kufungidzirwa. Mushure mekukodzera, ongorora zvakamisikidzwa zvisaririra zvezvasara serial correlation uye masikweya avo ekusara volatility clustering. Ongorora fungidziro ye data yakazotevera uchishandisa yakakodzera dhizaini yemusiyano wakaitika, uchiziva kuti matanho akaonekwa ane ruzha. A GARCH fit haafembere kwakananga kunotevera kudzoka, ratidza chikonzero chekusagadzikana kana vimbiso yekuvhara nguva dzenjodzi. Kutyoka kwemaitiro, kukwidziridza mhedzisiro (apo zvinokatyamadza uye zvinokatyamadza zvinokanganisa kushanduka zvakasiyana), uye intraday periodicity ingangoda kusiyana kana mhando dzakapfuma. Taura fungidziro uye mamiriro ekufungidzira, kunyanya kana fungidziro ichizivisa miganhu yenjodzi.
Yekufungidzira yekudzoka yakatevedzana ine mavhiki akanyarara anoteverwa nemavhiki asina mhirizhonga. Iyo GARCH modhi inogona kutakura ruzivo kubva munguva ichangopfuura squared kuvhunduka kuenda mune ramangwana mamiriro akasiyana, anomiririra kushata kuungana.
MuGARCH(1,1), fungidziro inotevera yemusiyano inoshandisa isingachinjiki, iyo yapfuura yakapetwa kaviri innovation uye yapfuura ine zvisungo musiyano. Kuvhundutsa kukuru kwazvino kunogona kusimudza fungidziro kunyangwe kudzoka kunotarisirwa kukaramba kuri pedyo ne zero.
Muongorori anofananidza zvakajairwa uye zvinorema-tailed innovation fungidziro uye anotarisa zvakamisikidzwa zvisaririra uye squared residual autocorrelation. Musiyano wemodhi inosiya kuungana mumativi ane mativi asara angave asina kukwana.
Chikwata chine njodzi chinoongorora nhanho imwe-nhanho yekusiyana-siyana zvichienderana neakazozivikanwa proxies uye inoaenzanisa neyakareruka inogara-yakasiyana-siyana yekutanga, kunyora chiyero chemumiriri.
Kufembera kwekusagadzikana kunogona kutaurirwa zvine hungwaru nekuratidza kunofungidzirwa kugovera hunyanzvi, kushingirira, horizon uye evaluation proxy padivi pefungidziro yemusiyano. Zvikwata zvine njodzi zvinofanirwa kuenzanisa fungidziro yeGARCH neyakareruka yekutanga uye kuongorora kuvharika kwepakati pasi pekuchinja kwemisika. Kana kuvhundutsa kuine asymmetric mhedzisiro, enzanisa inoruramisa asymmetric musiyano pane kumanikidza symmetric GARCH kuti itsanangure. Tarisa zvakamira zvakasara zvekuchinja kwehurumende uye gadzirisa zvakare zvichienderana nehumbowo. Mamiriro ezvinhu akasiyana-siyana anopfupikisa modhi-yakavakirwa kusava nechokwadi pane imwe nguva; haisi vimbiso yakasungwa pakurasikirwa kweramangwana.
Tsamba yemahara
Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.
Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.
Zviedze iwe pachako
Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.
Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo
GARCH modhi inofembera yenguva-inosiyana-siyana yezvimiso nekuenderana nekusagadzikana kwazvino nekukatyamadzwa kwakapetwa kaviri uye fungidziro yemusiyano yapfuura. Idzo dzinowanzo shandiswa pakudzoka kwemari ine volatility clustering, asi yakaringana musiyano wekufanotaura zvinoenderana nefungidziro yekugovera, kurongeka kwemuenzaniso uye maitiro ekudzoka achitsanangurwa zvakakwana.
Iyo yakajairwa GARCH(1,1) equation inosanganisa omega, yakarembera yakapetwa kaviri kuvhunduka uye kwakasaririra musiyano.
Iyo lagged squared innovation inopinda zvakanaka pasi pezvipingamupinyi zvakajairwa, ichiwedzera fungidziro mushure mekuvhunduka kukuru.
Musiyano equation inotsanangura kupararira kana kusagadzikana, kwete chiratidzo cheutsva hunotevera.
Zvisaririra zvakamisikidzwa zvinotariswa kuti zvakasara zvichitsamira mushure mekuverengera kune yakasarudzika mamiriro akasiyana.
Kutsamira mumasara ane mativi mana kunogona kuratidza kushanduka kwesimba kunoramba kusingatsananguriki.
Ramba uchidzidza
Mamwe madhairekitori akasarudzirwa nyaya iyi
InoteveraGaidhi rinotevera
MaModeli Ekukurudzira eMatanho Makumi maviri
Tekinoroji