Description: There is a large literature on claims reserving. However most of the stochastic models are too complex and end up being a black box for practitioners. They also tend to be over parameterized, which has the effect that they tend to model historical observations very accurately, but they do not necessarily perform well when predicting future claims. This paper applies a Bayesian method that was developed for forecasting very short series with stable [seasonal] patterns. It is constructed on basic Bayesian concepts and leads to very simple prediction formulas. It is then compared with several other methods as to their predictive effectiveness. In so doing, it departs from the traditional ‘triangle-shaped’ framework for development data.