The article explores how to estimate mean values and parameters in statistical models with heavy-tailed data. It suggests a method to estimate mean values accurately and efficiently for multivariate heavy-tailed data. The study also shows that M-estimators have a faster convergence rate than least square estimators in certain unstable processes. These findings help improve our understanding of how to analyze and interpret data with heavy tails.