Guaranteeing the stability of the estimator of the system matrix of a vector autoregression of order one $\operatorname{VAR}(1)$, is of great importance in many applications. However, almost all existing algorithms that do so are iterative, require many tuning parameters and are computationally expensive. Here we extend our recent work to derive two sets of new closed-form estimators that require no tuning parameters and are computationally cheap. We prove their stability and statistical consistency and compare them in simulations.