Tasirin ChatGPT akan Kadarorin Crypto Masu Alaƙa da AI: Shaida daga Binciken Sarrafa Kama-da-Kwane-kwane

2.1 Bambanci-Bambanci Na Roba

Binciken yana amfani da hanyar bambanci-bambanci na roba (SDID), wanda ke haɗa abubuwa na sarrafa roba da hanyoyin bambanci-bambanci. Wannan hanyar tana gina haɗin ma'auni na sarrafawa wanda ya dace da halayen raunin magani kafin magani.

Ana iya wakiltar ma'aunin SDID kamar haka:

$\hat{\tau}_{sdid} = \left(\sum_{t=T_0+1}^T Y_{1t} - \sum_{t=T_0+1}^T \hat{Y}_{1t}^{syn}\right) - \left(\sum_{t=1}^{T_0} Y_{1t} - \sum_{t=1}^{T_0} \hat{Y}_{1t}^{syn}\right)$

inda $Y_{1t}$ ke wakiltar sakamakon da aka lura don raunin magani, $\hat{Y}_{1t}^{syn}$ shine hasashen sarrafa roba, kuma $T_0$ yana nuna lokacin shiga tsakani (ƙaddamarwar ChatGPT).

2.2 Tattara Bayanai

Binciken ya haɗa da:

• Bayanan farashin yau da kullun na cryptocurrencies masu alaƙa da AI
• Yawan binciken Google don kalmomin da suka shafi AI
• Girman kasuwa da ma'aunin yawan ciniki
• Ƙungiyar kulawa na cryptocurrencies marasa AI

Bayanan sun ƙunshi watanni 6 kafin ƙaddamarwa da watanni 2 bayan ƙaddamarwa don ɗaukar duka tushen tushe da tasirin magani.

3.1 Tasirin ChatGPT akan Ribar

Binciken ya nuna gagarumin tasiri mai kyau akan kadarorin crypto masu alaƙa da AI:

• Wata ɗaya bayan ƙaddamarwa: Matsakaicin riba na 10.7% zuwa 15.6%
• Watanni biyu bayan ƙaddamarwa: Matsakaicin riba na 35.5% zuwa 41.3%
• Muhimmancin ƙididdiga: p < 0.01 a cikin duk samfuran

Wadannan tasirin sun daɗe bayan sarrafa yanayin kasuwa na gabaɗaya da abubuwan da suka shafi cryptocurrency.

3.2 Binciken Yawan Binciken Google

Yawan binciken Google don sharuɗɗan da suka shafi AI ya fito a matsayin muhimmin ma'auni na farashi bayan ƙaddamarwar ChatGPT:

• Yawan bincike ya karu da kashi 247% na "AI cryptocurrency"
• Ƙaƙƙarfan alaƙa tsakanin yawan bincike da haɓakar farashi (r = 0.78)
• Yawan bincike ya annabta kashi 61% na bambance-bambancen riba a cikin lokacin bayan magani

Sakamakon ya nuna hankalin masu saka hannun jari ya shiga tsakani tasirin ChatGPT akan kimantawar kasuwa.

4.1 Tsarin Lissafi

Ana ƙayyade ma'aunin sarrafa roba ta hanyar rage nisa tsakanin halayen kafin magani:

$\min_{w} \sqrt{(X_1 - X_0w)'V(X_1 - X_0w)}$

mai alaƙa da $w_j \geq 0$ da $\sum_{j=2}^{J+1} w_j = 1$, inda $X_1$ ya ƙunshi halayen kafin magani na raunin da aka bi da shi, $X_0$ ya ƙunshi halayen kafin magani na raunonin kulawa, kuma $V$ matrix ne na diagonal tare da ma'aunin fasali.

4.2 Aiwarar Lambar Code

import numpy as np
import pandas as pd
from scipy.optimize import minimize

class SyntheticControl:
    def __init__(self, treatment_unit, control_units, pre_periods):
        self.treatment = treatment_unit
        self.control = control_units
        self.pre_periods = pre_periods
    
    def fit(self):
        # Halayen kafin magani
        X1 = self.treatment[:self.pre_periods].mean()
        X0 = self.control[:self.pre_periods].mean(axis=1)
        
        # Ingantawa don nemo ma'auni
        def objective(w):
            return np.sqrt((X1 - X0 @ w).T @ (X1 - X0 @ w))
        
        constraints = [{'type': 'eq', 'fun': lambda w: np.sum(w) - 1}]
        bounds = [(0, 1) for _ in range(len(self.control))]
        
        result = minimize(objective, 
                         x0=np.ones(len(self.control))/len(self.control),
                         bounds=bounds,
                         constraints=constraints)
        
        self.weights = result.x
        return self.weights
    
    def predict(self, post_periods):
        synthetic_control = self.weights @ self.control[post_periods]
        return synthetic_control