Finance does not represent the desirability of a gamble in terms of a score on a unidimensional scale, but rather in terms of the trade-off or conflict between various statistical features of the payoff. The approach facilitates computations and dramatically simplifies learning. In addition, because it often changes erratically in a financial context, prediction risk is central to finance. Evidence is reviewed here that suggests that the brain analyzes gambles in terms of the statistical language of finance, and that the brain tracks prediction risk like in finance, using prediction risk estimates to optimally update predictions.