1. Gabatarwa
Tsinkayoyin da ake samu daga dandalin tsinkayar jama'a na kan layi kamar Metaculus suna ƙara amfani da su daga cibiyoyi kamar Babban Bankin Turai, kafofin watsa labarai, da masu tsara manufofi a matsayin tushen hangen nesa. Duk da haka, ba a sami isassun shaida game da daidaiton su idan aka kwatanta da ingantattun hanyoyin tsinkaya na gargajiya ba. Wannan binciken ya magance wannan gibi ta hanyar kimanta daidaiton tsinkayar farashin musayar kudi daga Metaculus da wani ma'auni na gargajiya kuma mai wahala a doke shi: tsarin tafiya bazuwar ba tare da karkata ba. Sakamakon yana da muhimman tasiri ga amincin da kuma amfani da hankalin jama'a a cikin tsinkayar kuɗi da tattalin arziki.
2. Bita na Adabi
2.1 Tsinkayar Jama'a
Ra'ayin "hikimar jama'a" yana nuna cewa tsinkayoyin da aka tattara daga rukuni daban-daban na iya zama mafi daidaito fiye da na ƙwararrun mutum ɗaya. Dandamali kamar Metaculus da Aikin Hukunci Mai Kyau suna aiwatar da wannan ta hanyoyin tambaya da haɗawa daban-daban (misali, matsakaita mai sauƙi, ƙa'idodin kasuwa na Bayesian). Duk da cewa shaida ta nuna tsinkayar jama'a ta fi zato bazuwar (Petropoulos et al., 2022), kwatancen kai tsaye da ma'auni na ƙididdiga a cikin fagage masu sarkakiya kamar kuɗi ba su da yawa.
2.2 Tsinkayar Farashin Musayar Kudi
Tsinkayar farashin musayar kudi sanannen abu ne mai wahala. Siririn Meese da Rogoff (1983) ya kafa cewa sauƙaƙan tsarin tafiya bazuwar sau da yawa yana fi ingantattun tsarin ƙididdiga a cikin gwaje-gwajen da ba na samfura ba don manyan nau'ikan kuɗi. Wannan ya sa tafiya bazuwar ta zama ma'auni mai tsauri kuma mai daraja don kimanta kowace sabuwar hanyar tsinkaya, gami da tsinkayar jama'a.
3. Bayanai & Dandali
Binciken ya yi amfani da bayanan tsinkayar farashin musayar kudi daga dandalin Metaculus. Metaculus yana ɗaukar tambayoyin inda masu amfani suke tsinkaya yuwuwar abubuwan da za su faru a gaba. An fitar da tsinkayoyin da suka dace game da motsin farashin musayar kudi (misali, EUR/USD, GBP/USD) ta hanyar API na dandalin. An samo ainihin bayanan farashin musayar kudi masu dacewa don tabbatarwa daga ingantattun rumbunan bayanan kuɗi (misali, Bloomberg, Refinitiv).
4. Hanyar Bincike
Babbar hanyar bincike ta ƙunshi kimanta daidaiton daidaito. An kwatanta tsinkayar jama'a (tsinkayar da aka tattara daga masu amfani na Metaculus) game da matakin farashin musayar kudi na gaba da tsinkayar da aka samar ta hanyar tsarin tafiya bazuwar ba tare da karkata ba. Tsinkayar tafiya bazuwar ita ce kawai farashin musayar kudi na ƙarshe da aka lura: $S_{t+1|t} = S_t$, inda $S_t$ shine farashin lokaci a lokacin $t$. Ana auna daidaiton tsinkaya ta amfani da ma'auni na kuskure na yau da kullun:
- Matsakaicin Kuskure Cikakke (MAE): $MAE = \frac{1}{N}\sum_{i=1}^{N} |F_i - A_i|$
- Tushen Matsakaicin Kuskure Madaidaici (RMSE): $RMSE = \sqrt{\frac{1}{N}\sum_{i=1}^{N} (F_i - A_i)^2}$
Inda $F_i$ shine tsinkaya kuma $A_i$ shine ainihin ƙimar. An gwada mahimmancin ƙididdiga na bambancin kurakurai ta amfani da gwajin Diebold-Mariano.
5. Sakamako
Babban sakamakon a bayyane yake kuma yana ban mamaki: tsarin tafiya bazuwar ba tare da karkata ba yana ba da tsinkayar farashin musayar kudi mafi daidaito sosai fiye da tsinkayoyin da aka tattara daga jama'ar Metaculus. RMSE da MAE na tsinkayar tafiya bazuwar sun kasance ƙasa a ko'ina cikin nau'ikan kuɗin da aka kimanta da kuma tsayin tsinkaya. Gwajin Diebold-Mariano ya tabbatar da cewa wannan fifikon yana da mahimmancin ƙididdiga.
6. Tattaunawa
Wannan sakamakon ya ƙalubalanci farin ciki marar tantancewa wani lokacin da ke kewaye da tsinkayar jama'a. Duk da yake jama'a na iya ƙware a cikin fagage masu iyaka, matsalolin da za a iya raba su (misali, kimanta nauyin sa), kasuwannin kuɗi da ke da babban amo, rashin tsayawa, da juyayi (inda tsinkaya ke tasiri sakamakon) na iya mamaye tsarin "hikima". Jama'a na iya haɗa sigina na ƙarya ko son zuciya wanda sauƙaƙan tafiya bazuwar mara sigina ta guje wa.
7. Ƙarshe
Don tsinkayar farashin musayar kudi, ma'auni na gargajiya kuma mai sauƙi na ƙididdiga (tafiya bazuwar) ya fi tsinkayoyin daga ingantaccen dandalin tsinkayar jama'a na kan layi. Wannan yana jaddada mahimmancin ingantaccen ma'auni kafin a tura sabbin kayan aikin tsinkaya a cikin aikace-aikace masu mahimmanci. Yana nuna cewa ƙimar tsinkayar jama'a na iya zama ta musamman ga yanki kuma bai kamata a ɗauka cewa za a iya faɗaɗa shi zuwa jerin lokutan kuɗi masu sarkakiya ba.
8. Bincike na Asali & Sharhin Kwararru
Fahimta ta Asali: Takardar ta kawo tabbataccen bincike na gaskiya, mai buƙata. Babban binciken - cewa ƙirar da ba ta da hankali ta doke "hikimar jama'a" a cikin kuɗi - bai zama abin mamaki ga ƙwararrun ƙididdiga ba, amma magani ne mai mahimmanci ga tashin hankali. Yana ƙarfafa wani tushe na asali na ƙididdigar kuɗi: doke tafiya bazuwar shine kutsawar mai tsarki, kuma yawancin abubuwa sun kasa. Ainihin gudunmawar takardar ita ce amfani da wannan ma'auni marar tausayi ga wata hanya ta zamani, mai ban sha'awa.
Kwararar Hankali: Hankali yana da inganci kuma na gargajiya: ayyana maƙasudi mai wahala (farashin musayar kudi), zaɓi mafi ƙarfin ma'auni (tafiya bazuwar), kuma a gudanar da tsaftataccen tsere. Amfani da ingantattun ma'auni na kuskure (RMSE, MAE) da gwaje-gwajen ƙididdiga (Diebold-Mariano) yana da ƙarfi ta hanyar bincike. Yana bin samfurin da aka tabbatar na sukar Meese-Rogoff, yana tambaya yadda ya kamata: "Shin wannan sabon abu yana magance tsohuwar matsalar da ba a warware ba?" Amsar ita ce a'a a bayyane.
Ƙarfi & Kurakurai: Ƙarfinsa shine sauƙinsa mai ladabi da bayyanannen sakamako. Kuskuren, wanda aka yarda da shi a cikin tattaunawa, shine iyakacin yaduwa. Wannan bincike ne na yanki ɗaya (FX) akan dandali ɗaya (Metaculus). Ba ya ƙaryata tsinkayar jama'a don, a ce, abubuwan siyasa ko sifofin amfani da fasaha, inda bayanai ba su da yawa kuma ƙirar ba su da ƙarfi. Kamar yadda bincike daga Aikin Hukunci Mai Kyau ya nuna, ƙirar tambaya tare da masu tsinkaya da aka horar za su iya yin fiye da haka a irin waɗannan yankuna (Tetlock & Gardner, 2015). Takardar za ta iya zama mafi ƙarfi ta hanyar hasashen dalilin da ya sa jama'a suka kasa - shin sun wuce gona da iri ga amo, kiwo, ko rashin ƙwarewar yanki a cikin mahalarta?
Fahimta Mai Aiki: Ga masu aiki: Kada ku maye gurbin dandamalin jama'a a makance don ingantattun ma'auni a cikin ƙididdigar kuɗi. Yi amfani da su azaman sigina mai dacewa, mai yuwuwar sabani. Ga masu haɓaka dandali: Binciken umarni ne don ƙirƙira. Shin za a iya inganta algorithms na tattarawa don tace amo? Shin dandamali ya kamata su auna masu tsinkaya ta hanyar ingantattun tarihin yanki da aka tabbatar, kamar ra'ayoyin maganin gaskiya na Bayesian da Prelec (2004) ya bincika? Ga masu bincike: Maimaita wannan! Gwada sauran nau'ikan kadarori, wasu dandamali (misali, Polymarket), da ƙirar gauraye waɗanda suka haɗa ra'ayin jama'a da ƙirar ƙididdiga, kamar yadda aka ba da shawara a cikin tsinkayar annoba (McAndrew et al., 2024). Iyakar ba jama'a da ƙira ba ce, amma haɗin kai mai hankali.
9. Cikakkun Bayanai na Fasaha & Tsarin Lissafi
An ayyana tsarin tafiya bazuwar ba tare da karkata ba don jerin lokaci $S_t$ kamar haka: $S_t = S_{t-1} + \epsilon_t$, inda $\epsilon_t$ shine kalmar kuskuren farin amo tare da $E[\epsilon_t]=0$ da $Var(\epsilon_t)=\sigma^2$. Tsinkayar $h$-mataki-gaba kawai ita ce: $\hat{S}_{t+h|t} = S_t$. Wannan ƙirar tana nuna cewa mafi kyawun tsinkaya na ƙimar gaba shine ƙimar yanzu, kuma canje-canje ba za a iya tsinkaya su ba.
Tsinkayar jama'a daga Metaculus, $C_{t+h|t}$, tarawa ce (sau da yawa matsakaita mai nauyi) na tsinkayoyin mutum ɗaya na masu amfani don farashin musayar kudi a lokacin $t+h$. Kwatancen ya dogara ne akan bambancin kuskuren tsinkaya: $d_t = e_{t}^{RW} - e_{t}^{C}$, inda $e_{t}^{RW} = (S_{t+h} - \hat{S}_{t+h|t}^{RW})^2$ da $e_{t}^{C} = (S_{t+h} - \hat{C}_{t+h|t})^2$. Alamar gwajin Diebold-Mariano ita ce: $DM = \frac{\bar{d}}{\sqrt{\widehat{Var}(\bar{d})/T}} \sim N(0,1)$, inda $\bar{d}$ shine matsakaicin samfurin bambancin asara.
10. Sakamakon Gwaji & Bayanin Ginshiƙi
Bayanin Ginshiƙi (An yi hasashe bisa sakamako): Ginshiƙi mai taken "Kwatancin Kuskuren Tsinkaya: Tafiya Bazuwar da Jama'ar Metaculus." X-axis yana jera nau'ikan kuɗi daban-daban (misali, EUR/USD, GBP/USD, USD/JPY). An naita saiti biyu na sanduna ga kowane nau'i: ɗaya don RMSE na Tafiya Bazuwar (a cikin shuɗi) da ɗaya don RMSE na Jama'ar Metaculus (a cikin ja). A ko'ina cikin nau'ikan, sandunan shuɗi (Tafiya Bazuwar) suna bayyane gajarta fiye da sandunan ja (Jama'a), suna kwatanta daidaiton daidaito na tafiya bazuwar ta hanyar ƙididdiga. Wani layin layi na biyu da aka lulluɓe akan ginshiƙi yana nuna jerin lokutan bambancin asara ($d_t$), wanda ke jujjuyawa a kusa da matsakaici mai kyau, yana nuna ci gaba da fifikon tafiya bazuwar. Taurari da ke sama da sandunan ja suna nuna mahimmancin ƙididdiga a matakin 5% bisa gwajin Diebold-Mariano.
11. Tsarin Bincike: Misali Mai Amfani
Harka: Kimanta Sabon Sigina na "AI-Powered" FX. Manajan kadarori an gabatar masa da sabon ƙirar ML da ke da'awar tsinkayar EUR/USD. Ta yaya ake kimanta shi?
Mataki 1 – Ayyana Ma'auni: Nan da nan saita tafiya bazuwar ($F_{t+1} = S_t$) a matsayin babban ma'auni. Kada a yi amfani da wani ƙirar sarkakiya a matsayin ma'auni kaɗai.
Mataki 2 – Rarraba Bayanai: Yi amfani da lokacin da ba na samfura ba mai tsayi (misali, bayanan yau da kullun na shekaru 3-5 da ba a yi amfani da su a horar da ƙirar ML ba).
Mataki 3 – Lissafin Kuskure: Yi lissafin RMSE don duka ƙirar ML da tsinkayar tafiya bazuwar a cikin lokacin da ba na samfura ba.
Mataki 4 – Gwajin Ƙididdiga: Yi gwajin Diebold-Mariano akan bambance-bambancen kuskure madaidaici. Shin ƙananan kuskuren ƙirar ML yana da mahimmancin ƙididdiga (ƙimar-p < 0.05)?
Mataki 5 – Mahimmancin Tattalin Arziki: Ko da yana da mahimmancin ƙididdiga, shin raguwar kuskure yana da ma'anar tattalin arziki don dabarun ciniki bayan lissafin farashin ma'amala?
Wannan tsarin, wanda aka yi amfani da shi kai tsaye a cikin takardar, gwajin litmus ne na duniya ga kowace sabuwar da'awar tsinkaya a cikin kuɗi.
12. Aikace-aikace na Gaba & Jagororin Bincike
- Ƙirar Tsinkaya ta Gauraye: Maimakon ko dai/ko hanyar, bincike ya kamata ya mayar da hankali kan haɗa ingantaccen kimanta yuwuwar da aka samo daga jama'a da ƙirar jerin lokaci na gargajiya. Matsakaicin ƙirar Bayesian ko hanyoyin haɗin gwiwa na iya amfani da ikon jama'a na kimanta abubuwan da ba kasafai ba da ƙarfin ƙirar a cikin kama dagewa.
- Ƙirar Dandali Ta Musamman ga Yanki: Dandamalin jama'a na gaba don kuɗi na iya buƙatar fasalulluka na musamman: shuka tsinkayoyi tare da fitar da ƙirar ƙididdiga, auna masu tsinkaya bisa ga aikin da suka yi a baya a cikin tambayoyin kuɗi, da kuma tambayar bayyana rarraba tsinkaya maimakon ƙididdiga don kama rashin tabbas da kyau.
- Bayyana Rashin Nasara/Nasarar Jama'a: Ana buƙatar ƙarin bincike don rarraba dalilin da ya sa jama'a suka kasa a wasu yankuna (FX) amma suka yi nasara a wasu (annoba). Shin yanayin bayanan ne, tafkin mahalarta, ko tsarin tambayar? Wannan yana buƙatar aikin tsaka-tsaki wanda ke haɗa ilimin halin ɗan adam, ƙididdiga, da ƙwarewar yanki.
- Aikace-aikace a cikin Filayen Kusa: Ya kamata a ƙaddamar da hanyar ma'auni zuwa wasu yankuna "masu wahala a tsinkaya" kamar sauyin cryptocurrency, farashin kayayyaki, ko abubuwan ban mamaki na nuna alamar tattalin arziki.
13. Nassoshi
- Lehmann, N. V. (2025). Ƙwarewar tsinkaya na dandalin tsinkayar jama'a: Kwatancin tsinkayar farashin musayar kudi. arXiv preprint arXiv:2312.09081v2.
- Meese, R. A., & Rogoff, K. (1983). Ƙirar ƙirar farashin musayar kudi na shekarun saba'in: Shin sun dace da waje da samfura? Journal of International Economics, 14(1-2), 3-24.
- Tetlock, P. E., & Gardner, D. (2015). Superforecasting: Fasaha da Kimiyyar Tsinkaya. Crown Publishers.
- Prelec, D. (2004). Maganin gaskiya na Bayesian don bayanan da ke ƙarƙashin ra'ayi. Kimiyya, 306(5695), 462-466.
- Diebold, F. X., & Mariano, R. S. (1995). Kwatanta daidaiton tsinkaya. Journal of Business & Economic Statistics, 13(3), 253-263.
- McAndrew, T., Gibson, G., et al. (2024). Haɗa tsinkayoyin da jama'a suka samo tare da ƙirar ƙididdiga don tsinkayar annoba. PLOS Computational Biology.
- Atanasov, P., et al. (2022). Distilling the wisdom of crowds: A primer on forecasting tournaments and prediction markets. In The Oxford Handbook of the Economics of Networks.