This paper proposes a study of different dynamic objectives aggregation methods (DOAMs) in the context of a multi-objective evolutionary approach to portfolio optimisation. Since the incorporation of chaotic rules or behaviour in population-based optimisation algorithms has been shown to possibly enhance their searching ability, this study considers and evaluates also some chaotic rules in the dynamic weights generation process. The ability of the DOAMs to solve the portfolio rebalancing problem is investigated conducting a computational study on a set of instances based on real data. The portfolio model considers a set of realistic constraints and entails the simultaneous optimisation of the risk on portfolio, the expected return and the transaction cost.