Scientists have developed a method to accurately estimate covariance matrices in cosmology without needing a large number of simulations. By fitting a theoretical model to a smaller set of simulations, they can determine the parameters of the model. This approach was tested on two examples, showing that the model covariance obtained from a limited number of simulations can be as reliable as one built from a much larger set. The method was also applied to a scenario with a large covariance matrix, resulting in an improved model that outperformed the traditional Gaussian covariance.