In this paper, the application of a hybrid coastal morphodynamic model to forecast inter-annual beach change is discussed through the prediction of beach change in a high energy sandy beach over a period of 5 years. The modelling approach combines a ‘reduced-physics’ formulation with a data-driven approach through an inverse technique to form the hybrid coastal morphodynamic model. The beach considered for the demonstration of the model is the Narrabeen Beach, which is a dynamic sand beach located in New South Wales, Australia. Despite its simplicity, we find that the model is able to capture beach change at Narrabeen Beach at inter-annual timescales with root mean square error between measured and computed beach profiles less than 0.4 m on average. Even though the model is used to forecast inter-annual beach change in this study, its ability to predict beach change is not limited to that timescale but depends on the frequency of historic beach profile measurements available to determine key unknown parameters of the model. Also, the length of profile forecasts largely depends on the length of available historic measurements where longer data sets allow longer predictions within a range of beach behaviour contained in the observations. The ability of the model to reliably forecast coastal change at inter-annual and potentially at other timescales, and its high efficiency make it possible to be used in providing multiple simulations required for probabilistic coastal change forecasts which will be very useful for coastal management purposes.