Why do we use false models?

Summary How can we understand that at the heart of the mathematical tools used daily by financial actors to guide their decisions is the assimilation of two situations - situations of randomness and heterogeneity - that statistical tools very unequally allow to grasp? The article traces the way in which the amalgam between situations of randomness and situations of heterogeneity has come to play the role of a determining, albeit implicit, institution within contemporary financial reasoning. We explore the two social scenes in which this assimilation is likely to be engaged: that of its theoretical uses, in the social space where probability theory is written and developed and the uses made of it in financial mathematics and economics; that of its practical uses, in the organizations and markets where the models it contributes to found are used, focusing more particularly on the insurance sector.