Balancing in a distributed generation network is an increasingly difficult task because of the increasing number of residential prosumers on the power network. This paper proposes a framework for the estimation, as well as the prediction of the power flexibility of residential prosumers. In order to quantify the residential buildings’ demand flexibility, a thermoelectric simulation model of a typical residential house was developed based on first engineering principles. Based on the calculated flexibility values, a simple prediction method was used to give a short-term forecast of the prosumer flexibility. The results were validated by simulation experiments incorporating real data for four different scenarios.