Due to ever lower cost, investments in renewable electricity generation and storage have become more attractive in recent years to electricity consumers at different scales. At the same time, electricity generation and storage have also become something that can be shared or traded locally in energy communities and microgrid systems. In this context, peer-to-peer (P2P) sharing has gained attention, since it offers a way to optimize the cost-benefits from distributed resources, making them financially more attractive. However, cooperation in practical instances still faces unclear requirements about e.g. how much predictive power is required for significant cost-saving; how many peers to contact to form efficient groups; and then, who to team up with for sharing electricity generation and storage. To answer such questions, we introduce a realistic and comprehensive cost-optimization model for P2P energy sharing communities, making continuous decisions while using only limited forecast for the input data. We provide strong evidence, based on the analysis of real household data, that the financial benefit of cooperation does not require long forecast horizons and even P2P energy sharing in small groups (with only 2–5 participants in this study) can reach a high fraction (96% in our results) of the ideal maximum gain, achievable when all input is known ahead of time. Maintaining such small communities results in much lower associated complexity and better privacy, as each participant only needs to share its data with few other peers. Our findings shed new light and motivate requirements for how to organize locally in an efficient manner prosumers and consumers into energy sharing communities in tomorrow's real implementations.