This chapter examines the two main goals of scientific prediction. In its first role, associated with traditional philosophy of science, prediction is a means to theory confirmation, hence the characteristics of a successful prediction are those that are best suited to testing hypotheses and theories: accuracy and riskiness. Philosophers have argued that the latter is usually achieved through the characteristics of novelty and precision. However, in many applied sciences, predictions are used to intervene on the world. Unfortunately, the success rate of applied predictions tends to be quite low. Therefore, I argue that the standards for success should be modified in the case of applied predictions, sacrificing a level of precision for the sake of greater accuracy.