Dynamics na Entropic na Farashin Musayar Kudi da Zaɓuɓɓuka: Sabon Tsari don Ƙirar FX

Tsarin Abubuwan Ciki

1. Gabatarwa

Wannan takarda ta gabatar da tsarin Dynamics na Entropic don ƙirar farashin musayar kudi na waje (FX) da farashin zaɓuɓɓukan Turai. Babban manufa ita ce samar da madadin tushe, tushen ka'idar bayanai ga dynamics na kuɗi, wanda ya wuce lissafin al'ada na stochastic. Marubutan, Mohammad Abedi da Daniel Bartolomeo, sun yi amfani da ka'idojin tunani na entropic—wata hanya don yin tunani a ƙarƙashin cikakkun bayanai—don fitar da sanannun tsarin kuɗi daga ka'idojin farko.

Aikin ya haɗa ra'ayoyin madaidaicin entropy da lissafin bayanai zuwa aikin kuɗi, wanda ya ƙare a fitar da Geometric Brownian Motion (GBM) don farashin musayar kudi da tsarin Garman-Kohlhagen don zaɓuɓɓukan FX. Wannan hanya ta nuna rashin canjin sikelin da ke cikin nau'ikan kuɗi, wanda ya kai ga zaɓin halitta na ƙirar lissafin farashin musayar kudi.

2. Tsarin Ka'idar

2.1. Tunani na Entropic da Matsakaicin Entropy

Tunani na entropic tsari ne na tunani don yanayi tare da cikakkun bayanai. Kayan aikin sa na farko shine ka'idar yiwuwa don wakiltar yanayin imani. Na biyu shine entropy na dangi (ko bambancin Kullback-Leibler), ana amfani dashi don sabunta imani lokacin da sabbin bayanai suka zo, bisa ga Ka'idar Sabuntawa Mafi ƙanƙanta. Ƙara entropy na dangi yana haifar da mafi ƙarancin son zuciya na baya-bayan nan wanda ya haɗa da duk bayanan da ake da su.

Kayan aiki na uku shine lissafin bayanai, wanda ke ba da ma'auni akan sararin rarraba yiwuwa. Duk da ba a bincika shi sosai a nan ba, marubutan sun lura da yuwuwar muhimmancinsa ga sarrafa fayil da dynamics na kadarori da yawa.

2.2. Dynamics na Entropic da Lokaci

Dynamics na Entropic yana amfani da tunani na entropic don ƙirar yadda tsarin ke canzawa. Wani muhimmin ƙirƙira shine gabatar da sigar lokacin entropic, wanda ke fitowa kuma an keɓance shi ga takamaiman tsarin maimakon zama agogo na duniya. Wannan ra'ayi an yi amfani da shi cikin nasara a cikin yanayi daban-daban na kimiyyar lissafi kuma an daidaita shi zuwa kuɗi a nan.

2.3. Rashin Canjin Sikelin a cikin FX

Wani muhimmin daidaito a cikin kasuwannin FX shine rashin canjin sikelin: dynamics bai kamata ya dogara da ko mun faɗi farashin musayar kudi a matsayin USD/EUR ko a cikin sigar sa ba. Wannan daidaito ya ƙayyade cewa ya kamata a tsara tsarin dangane da lissafin farashin musayar kudi, $x = \ln S$, inda $S$ shine farashin FX na lokaci. Canje-canje kamar $S \to \lambda S$ (sikelin mai sauƙi) suna barin dynamics ba canzawa ba lokacin da aka bayyana su cikin sharuddan $x$.

3. Fitar da Tsarin

3.1. Daga Ka'idojin Entropic zuwa GBM

An fara da bayanan farko game da ƙimar FX—musamman, ƙimar farko da saurin sa—marubutan suna amfani da tsarin dynamics na entropic don fitar da juyin halittar lokacinsa. Ta hanyar sanya ƙuntatawa daidai da abubuwan lura na kasuwa (kamar iyakataccen bambanci) da ƙara entropy, an nuna rarraba yiwuwar don lissafin musayar kudi na gaba $x$ yana bin tsarin motsi-diffusion.

Canza komawa zuwa farashin lokaci $S = e^x$, wannan tsari ya zama sanannen Geometric Brownian Motion (GBM): $$ dS_t = \mu S_t dt + \sigma S_t dW_t $$ inda $\mu$ shine motsi, $\sigma$ shine sauri, kuma $W_t$ tsarin Wiener ne. Fitarwa ta nuna girmamawa ga rashin canjin sikelin.

3.2. Ma'aunin Rashin Haɗari da Farashin Zaɓuɓɓuka

Don farashin abubuwan da aka samo asali, ana kiran ka'idar rashin cin riba. Marubutan sun nuna yadda ake fitar da ma'aunin rashin haɗari $\mathbb{Q}$ a cikin tsarin entropic. Wannan ya haɗa da daidaita motsin tsarin GBM zuwa bambancin ƙimar rashin haɗari tsakanin kuɗaɗen biyu, $(r_d - r_f)$.

A ƙarƙashin $\mathbb{Q}$, dynamics ya zama: $$ dS_t = (r_d - r_f) S_t dt + \sigma S_t dW_t^{\mathbb{Q}} $$ Farashin zaɓi na Turai akan ƙimar FX tare da wannan dynamics yana kaiwa kai tsaye zuwa dabarar Garman-Kohlhagen, kwatankwacin dabarar Black-Scholes na FX.

4. Sakamako da Tattaunawa

4.1. Tsarin Garman-Kohlhagen

Sakamako na ƙarshe na fitarwar entropic shine tsarin Garman-Kohlhagen don farashin zaɓi na Turai: $$ C = S_0 e^{-r_f T} \Phi(d_1) - K e^{-r_d T} \Phi(d_2) $$ inda $$ d_1 = \frac{\ln(S_0/K) + (r_d - r_f + \sigma^2/2)T}{\sigma\sqrt{T}}, \quad d_2 = d_1 - \sigma\sqrt{T} $$ $S_0$ shine farashin lokaci, $K$ shine bugun, $T$ shine lokacin cikawa, $r_d$ da $r_f$ su ne ƙimar rashin haɗari na cikin gida da na waje, $\sigma$ shine sauri, kuma $\Phi$ shine daidaitaccen CDF na al'ada.

4.2. Kwatanta da Hanyoyin Al'ada

Babban gudunmawar takardar ita ce hanyar aiki. Ta dawo da tsararrun tsare-tsare (GBM, Garman-Kohlhagen) ba ta hanyar lissafin stochastic da hujjojin shinge ba, amma ta hanyar ka'idar bayanai, tsarin farko wanda ya dogara da ƙara entropy da daidaito. Wannan yana ba da ƙarin zurfi, mafi tushen hujja ga waɗannan tsare-tsare kuma yana buɗe kofa don haɓaka su ta hanyar haɗa ƙuntatawa daban-daban ko mafi rikitarwa.

5. Fahimtar Jigo & Ra'ayin Mai Bincike

Fahimtar Jigo: Wannan takarda ba game da sabon dabarar farashi mafi kyau ba ce; wasa ne na falsafa mai ƙarfi. Tana jayayya cewa gabaɗayan ginin kuɗi na ci gaba da lokaci, daga Bachelier zuwa Black-Scholes, za a iya sake gina shi daga tushe ta amfani da ka'idar bayanai da ka'idar matsakaicin entropy. A zahiri marubutan suna cewa, "Ka manta da lemma na Ito na daƙiƙa guda; halayen kasuwa shine kawai abin da ba zai iya ba da mamaki ba, idan aka yi la'akari da abin da muka sani." Wannan babban sauyi ne daga ƙirar farashin zuwa ƙirar ilimin game da farashin.

Kwararar Hankali: Hujjar tana da kyau kuma tana da sauƙi. 1) Muna da cikakkun bayanai (rarrabuwar farko). 2) Muna da daidaito (rashin canjin sikelin). 3) Muna sabunta imaninmu ta amfani da kayan aikin da ya canza su mafi ƙanƙanta (matsakaicin entropy na dangi). 4) Wannan sabuntawar, wanda aka fassara shi azaman dynamics, yana ba mu GBM. 5) Rashin cin riba ya ƙayyade motsi, yana ba mu ma'aunin rashin haɗari don farashi. Fitarwa ce mai tsabta, wacce aka kora ta hanyar axiom wacce ta sa hujjar PDE/shinge ta al'ada ta yi kama da ƙazanta idan aka kwatanta.

Ƙarfi & Kurakurai: Ƙarfin shine kyawun tushe da yuwuwar haɓakawa. Kamar yadda aka gani a kimiyyar lissafi tare da aikin E.T. Jaynes da daga baya Caticha, hanyoyin entropic suna ƙware wajen fitar da sakamako na al'ada daga ka'idoji masu sauƙi. Laifin, kamar yadda yake da yawancin ka'idoji masu kyau, shine tazarar zuwa gaskiyar rikice-rikice. Tsarin ya fitar da GBM cikin kyau, amma GBM da kansa tsari ne mai aibi don FX (yana ƙima ƙarancin haɗarin wutsiya, yana watsi da tarin sauri). Takardar ta taƙaita ambaton aikin gaba akan tsalle-tsalle da lissafin bayanai, wanda shine inda ainihin gwajin yake. Shin wannan tsarin zai iya haɗa gaskiyar gaskiyar kasuwa (misali, wutsiya mai kitse) ta hanyar ƙara daidaitattun ƙuntatawa kawai, ko zai buƙaci gyare-gyare na ad-hoc waɗanda ke rage tsarkinsa?

Fahimta Mai Aiki: Ga masu ƙididdigewa da masu tabbatar da tsari, wannan takarda wajibi ce karantawa. Tana ba da sabon ruwan tabarau don tantance haɗarin tsari. Maimakon kawai gwada dacewar tsarin, tambayi: "Wane bayani ne wannan tsarin yake ɗauka? Shin wannan saitin bayanin ya cika ko ya dace?" Ga masu ƙirƙira, taswirar tana bayyana. Mataki na gaba shine amfani da wannan tsarin don gina sababbin tsare-tsare. Ƙuntata ƙarar entropy tare da bayanai game da murmushin saurin gani ko mitar tsalle, kamar yadda marubutan suka nuna ta hanyar nuni ga tsarin Bates da Heston. Kyautar ita ce haɗin kai, ka'idar haɗin kai ta farashin abubuwan da aka samo asali wanda baya haɗa tsare-tsaren da ba su dace ba. Aikin Peters da Gell-Mann (2016) akan tattalin arzikin ergodicity yana nuna irin wannan tunani na tushe yana samun karbuwa. Wannan takarda mataki ne mai ƙarfi a wannan hanyar, amma kasuwa za ta zama alkali na ƙarshe na amfaninta fiye da jan hankali na falsafa.

6. Cikakkun Bayanai na Fasaha

Jigon lissafin ya haɗa da ƙara entropy na dangi $\mathcal{S}[P|Q]$ na rarraba baya $P(x'|x)$ dangane da na farko $Q(x'|x)$, ƙarƙashin ƙuntatawa. Muhimmin ƙuntatawa shine matsakaicin murabba'in motsi, wanda ke gabatar da saurin $\sigma$: $$ \langle (\Delta x)^2 \rangle = \kappa dt $$ inda $\kappa$ yana da alaƙa da saurin $\sigma$. Ƙara yana haifar da yiwuwar canzawa na Gaussian: $$ P(x'|x) \propto \exp\left(-\frac{(x' - x - \alpha dt)^2}{2\kappa dt}\right) $$ wanda a cikin iyakar ci gaba yana kaiwa ga SDE na motsi-diffusion don $x_t$. Haɗin kai zuwa PDE na Black-Scholes-Merton an yi shi ta hanyar daidaitaccen hujjar ƙima mara haɗari da aka yi amfani da shi ga tsarin GBM da aka samo.

7. Misalin Tsarin Bincike

Harka: Haɗa Bayanan Murmushin Sauri. Tsarin entropic yana ba da damar haɗa ƙarin bayanan kasuwa. A ce, bayan farashin lokaci da saurin tarihi, muna kuma da bayanai daga kasuwar zaɓuɓɓukan da ke nuna cewa rarraba rashin haɗari na dawowar lissafi ba Gaussian ba ne amma yana da ƙima mara kyau da ƙarin kurtosis (murmushin sauri).

Mataki 1: Ayyana Ƙuntatawa. Baya ga ƙuntatawar bambanci $\langle (\Delta x)^2 \rangle = \sigma^2 dt$, mun ƙara ƙuntatawar lokaci daga fuskar saurin da aka gani: $$ \langle (\Delta x)^3 \rangle = \tilde{S} dt, \quad \langle (\Delta x)^4 \rangle - 3\langle (\Delta x)^2 \rangle^2 = \tilde{K} dt $$ inda $\tilde{S}$ da $\tilde{K}$ suka kama skewness da kurtosis a kowace lokaci.

Mataki 2: Ƙara Entropy. Ƙara entropy na dangi tare da waɗannan ƙuntatawa huɗu (matsakaici, bambanci, skewness, kurtosis) yana kaiwa ga yiwuwar canzawa $P(x'|x)$ wanda aka bayyana ta jerin Gram-Charlier ko ƙarin rarraba iyali mai ƙarfi, ba Gaussian mai sauƙi ba.

Mataki 3: Fitar Dynamics. Sakamakon iyakar ci gaba da lokaci zai zama tsarin diffusion tare da motsi mai dogaro da jiha da sauri, ko yuwuwar tsarin tsalle-diffusion, yana fitar da tsari kamar na Bates ko Heston daga ka'idojin farko na bayanai maimakon ƙayyadaddun tsarin sauri na stochastic.

Wannan misalin yana nuna ikon tsarin don haɓaka tsare-tsare bisa tsari ta hanyar haɗa ƙarin bayanan kasuwa a matsayin ƙuntatawa.

8. Aikace-aikace na Gaba & Hanyoyi

Tsarin dynamics na entropic yana buɗe hanyoyi masu ban sha'awa da yawa don bincike na gaba a cikin kuɗin ƙididdigewa:

• Fayiloli na Kadarori Da Yawa & Lissafin Bayanai: Marubutan sun ambaci amfani da lissafin bayanai ga zaɓin fayil. Wannan na iya haifar da sabbin dabarun rarraba kadarori dangane da "nisa" tsakanin rarraba kasuwa na yanzu da ingantaccen rarraba da aka yi niyya, wanda ya wuce ingantaccen bambanci-matsakaici.
• Ƙirar Gaskiyar Salo: Tsarin ya dace da halitta don haɗa sanannun siffofi na zahiri kamar wutsiya mai kitse, tarin sauri, da tasirin leverage ta hanyar ƙara ƙuntatawa na motsi ko sanya ƙuntatawa da kansu sun dogara da lokaci bisa bayanan da suka gabata.
• Kasuwannin da ba su tsaya ba da Canjin Tsarin Mulki: Rarraba na farko $Q$ a cikin entropy na dangi ana iya sabunta shi da motsi don nuna canje-canjen tsarin kasuwa, yana ba da damar hanyar da ta dace don gina tsare-tsare masu daidaitawa waɗanda ke amsa karyewar tsarin.
• Haɗin Kuɗi na Halayya: Ƙuntatawar "bayanai" za a iya ƙaddara don haɗa ma'auni na ra'ayin mai zuba jari ko hankali, tare da haɗa tazarar tsakanin kuɗin ƙididdigewa na al'ada da tsarin halayya.
• Haɗin Kayan Koyon Injina: Ka'idar matsakaicin entropy ita ce ginshiƙi na yawancin hanyoyin koyon injina. Wannan tsarin zai iya samar da ingantaccen tushe na ka'idar bayanai don tsarin haɗin gwiwar ML-kuɗi, yana bayyana dalilin da yasa wasu gine-ginen hanyar sadarwar jijiyoyi ko dabarun daidaitawa suke aiki da kyau don jerin lokacin kuɗi.

Manufa ta ƙarshe ita ce haɗin kai, ka'idar tushen axiom na dynamics kasuwa wanda duka ingantacce ne a ka'ida kuma daidai ne a zahiri, yana rage buƙatar facin tsarin ad-hoc na gama gari a cikin injiniyanci na kuɗi na yau.

9. Nassoshi

1. Jaynes, E. T. (1957). Ka'idar Bayanai da Lissafin Kididdiga. Bita na Jiki, 106(4), 620–630.
2. Caticha, A. (2012). Tunani na Entropic da Tushen Kimiyyar Lissafi. A cikin Proceedings of the MaxEnt 2012 conference.
3. Garman, M. B., & Kohlhagen, S. W. (1983). Ƙimar zaɓi na kuɗin waje. Jaridar Kuɗin Duniya, 2(3), 231–237.
4. Black, F., & Scholes, M. (1973). Farashin zaɓuɓɓuka da bashin kamfanoni. Jaridar Siyasa ta Tattalin Arziki, 81(3), 637–654.
5. Peters, O., & Gell-Mann, M. (2016). Kimanta caca ta amfani da dynamics. Hargitsi: Jarida ta Kimiyya ta Rashin Layi, 26(2), 023103. https://doi.org/10.1063/1.4940236
6. Amari, S. I. (2016). Lissafin Bayanai da Aikace-aikacensa. Springer.
7. Bachelier, L. (1900). Ka'idar hasashe. Annales scientifiques de l'École Normale Supérieure, 3(17), 21–86.