“…15 The discriminative performance of a risk model depends not only on the identification of individual risk factors, but also on how these risk variables interact with other variables, how accurately these factors can be measured, and the appropriateness of the population and statistical techniques used for modeling. 16 Generally, the discriminative performance of risk models in the initial development data set is better than the performance in other data sets. 1,13,17 This self-fulfilling prophecy, otherwise known as optimism, is a statistical phenomenon that is well described in the literature.…”