Passwords have become the first and last line of defense of the many users of computer services that do not take into account policies and strategies against well-equipped computer vandals, this is because very rarely given the importance right to choice and management of passwords, which can easily be improved by setting appropriate policies for the selection and safe use in computer services. Entities and institutions that provide these services are very concerned for the convenience of users when using the service and not by how robust are the passwords they choose, the periodicity in the changes to them and the ease remember. It is at this point that this research has its importance due to the ability to use Bayes Theorem to calculate the conditional probability and then combined to predict the trend of a given user who is teacher and choose a password that is 4 symbols and are numeric, this result shows that it is possible to generate robust policies to reduce the likelihood that users tend to choose passwords easy prediction, allowing administrators to enforce security strategies of users, but at the same time this result will allow the vandal on that set of possible passwords should concentrate their efforts to predict it in the shortest time possible.