The Constant Proportion Portfolio Insurance (CPPI) technique is a dynamic capital-protection strategy that aims at providing investors with a guaranteed minimum level of wealth at the end of a specified time horizon. A pertinent concern of issuers of CPPI products is when to perform portfolio readjustments. One way of achieving this is through the use of rebalancing triggers; this constitutes the main focus of this paper. We propose a genetic programming (GP) approach to evolve trigger-based rebalancing strategies that rely on some tolerance bounds around the CPPI multiplier, as well as on the time-dependent implied multiplier, to determine the timing sequence of the portfolio readjustments. We carry out experiments using GARCH datasets, and use two different types of fitness functions, namely variants of Tracking Error and Sortino ratio, for multiple scenarios involving different data and/or CPPI settings. We find that the GP-CPPI strategies yield better results than calendar-based rebalancing strategies in general, both in terms of expected returns and shortfall probability, despite the fitness measures having no special functionality that explicitly penalises floor violations. Since the results support the viability and feasibility of the proposed approach, potential extensions and ameliorations of the GP framework are also discussed.