Esse é um post rápido para comparar os resultados de duas técnicas de matching diferentes: Propensity-score matching (PSM) e Entropy matching.
Uma explicação mais apurada das técnicas está fora do escopo desse post, mas de uma forma simples, você pode pensar da seguinte forma.
This is a post about randomization when defining treatment and control groups in a randomized experiment. The conversation goes along the following lines: you need to be sure the group that receives the treatment and the control group are randomly selected from the population.
Overview A well-known problem with mean-variance portfolio optimization is that it is subject to instability: small changes in the inputs lead to large differences in the optimal portfolios. Since estimates of the required parameters (particularly the expected returns) using historical returns are extremely noisy, this results in unstable portfolios that would required frequent rebalancing.