According to the fitting-attitude analysis of value (FA-analysis), to be valuable is to be a fitting object of a pro-attitude. A very similar approach can be taken towards probability: To be probable is to be a fitting object of credence - fitting given the available evidence. In other words, to be probable is to be credible. If credence, or at least high credence, is thought of as a kind of pro-attitude, this makes probability a kind of value. Indeed, many probability theorists from Poisson onwards did adopt this ‘epistemic’ interpretation of the concept of probability. In earlier work, setting off from the FA- analysis, I proposed a modelling of value relations which makes room for different types of value incommensurability. In this talk, I first recapitulate this modelling and then move on to develop a structurally similar analysis of probability relations. One of the advantages of the model is that we get a new account of Keynesian incommensurable probabilities, which goes beyond Keynes in distinguishing between incommensurabilities of different types.