Outliers are values that doesn’t converge with the rest of the data series. These values when they arise in financial context can represent problems that have a direct influence on the health of an enterprise and the decision-making by the managers. In view of this, it was intended with this work identify anomalies in financial launches arising from the accounts of real companies. For this, statistical analyzes of the launches were fulfilled in order that outliers detection techniques could be chosen and then compared with the outliers detection of evaluators . Among the great variety of techniques were chosen the methods of Boxplot, Boxplot adjusted, MAD and standard deviation. The results show that most of the series didn’t follow a normal distribution, and the experimental results of the comparisons between the automatic methods and the evaluators showed substantial differences.
