The behaviour of participants in electricity markets is complex and is more appropriately studied as an economic game rather than as an optimisation problem. In this paper, a co-evolutionary approach has been developed to study the dynamic behaviour of participants over many trading intervals. Each market participant is represented by a trading agent. The bidding strategy of each agent is modelled as a set of bidding actions, one for each possible discrete state of the state space observed by the agent. Trading agents co-evolve their own populations of bidding strategies using a Genetic Algorithm. Simulation results have shown that in this competitive environment, participants can learn to improve their trading profit and the proposed state-based bidding strategy can help facilitate this learning process.