Computing an Aggregator's Long Term Profit under Uncertain Behavior of the Agents

In the retail electricity market, consumers can subscribe a contract with a conventional retailer or cooperate through an aggregator who takes forward positions in the wholesale electricity market, modeled as a two-tiered system. We characterize analytically the core of the game and give conditions for its non emptiness. Then we propose a Machine Learning algorithm to forecast the consumers' demand and use these forecasts as inputs to optimize the aggregator's pricing strategy. The viability of the aggregator's pricing strategy is finally evaluated on a case study containing the power consumptions of 370 Portuguese consumers over four years