Built a factor-model based Long/Short Global Macro Strategy with different Portfolio Beta and different lengths of rolling estimator used for the input covariance matrix and the expected return and evaluated 1) the sensitivity of the strategy to variations of beta 2) the sensitivity to the length of the estimators. The strategy used the Fama-French factor model and took 13 ETFs as invest universe. It was back-tested over the period 2007 – 2016 and categorized under different market environment. The Market Portfolio (S&P 500) and the Mean-variance Long/Short Portfolio built with a yearly return target of 15 were considered as benchmark.