The article introduces a new model called mixed copula-based VAR to study relationships between random variables. They use a method called one-step maximum likelihood estimation to accurately estimate parameters. By combining different copula forms, they show that their model outperforms traditional models in terms of accuracy and reliability. Ignoring complex correlations between errors can lead to less efficient parameter estimation. The mixed copula-based VAR is found to be the best fitting model in their real data study.