Scholarlink

This paper aims to investigate the performance of the information criteria among three different GARCH models (GARCH, EGARCH, and GRJ-GARCH), considering whether or not to use Google Trends (GT) time series as an exogenous variable in the conditional variance of the Ibovespa index. Furthermore, the study assessed which of two different probability functions - normal distribution and t-student - is the best option. The results show that the GRJ-GARCH model with Student’s t distribution performed best according to Akaike’s Information Criteria (AIC) and the Schwarz Bayesian Criterion (BIC). However, according to the different information criteria, the results diverged on the effectiveness of including GT data as an exogenous variable.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Incorporating Google Trends Into GARCH Models for the Brazilian Stock Market

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Felipe Gonçalves Pereira,
  • Joao Carlos Felix Souza,
  • Joao Gabriel de Moraes Souza,
  • Ricardo Matos Chain,
  • Mariana Mendes Bastos Gama

摘要

This paper aims to investigate the performance of the information criteria among three different GARCH models (GARCH, EGARCH, and GRJ-GARCH), considering whether or not to use Google Trends (GT) time series as an exogenous variable in the conditional variance of the Ibovespa index. Furthermore, the study assessed which of two different probability functions - normal distribution and t-student - is the best option. The results show that the GRJ-GARCH model with Student’s t distribution performed best according to Akaike’s Information Criteria (AIC) and the Schwarz Bayesian Criterion (BIC). However, according to the different information criteria, the results diverged on the effectiveness of including GT data as an exogenous variable.