The determination of the pilot sample size for simulations

In the techniques of Monte Carlo simulation, one of the key problems is to determine the proper pilot sample size. This paper addresses two aspects when choosing the pilot samples size for a simulation: whether or not the stability of the variance and the normality of the mean are satisfied. Following an introduction and analyzes of methods to deal with these problems, an example of investment projects is presented and analyzed based on repeated simulations. The results show that these two tests are necessary to determine the proper sample size for a simulation in order to estimate the mean of the performance of an investment project.