For the deterministic investigation of extreme events like capsizing, broaching or wave impacts, methods for the generation of deterministic wave sequences are required. These wave sequences can be derived from full scale measurements, numerical simulations or other sources. Most methods for the generation of deterministic wave sequences rely as a backbone on linear wave theory for the backwards transformation of the wave train from the target position in the wave basin to the position of the wave maker. This implies that nonlinear wave effects are not covered to full extend or they are completely neglected. This paper presents a method to improve the quality of the generated wave train via an experimental optimization. Based on a first wave sequence generated with linear wave theory and measured in the wave basin, the phases and amplitudes of the wave maker control signal are modified in frequency domain. The iteration scheme corrects both, shifts in time and in location, resulting in an improved deterministic wave train at the target location. The paper includes results of this method from three different basins with different types of wave generators, water depth and model scales. In addition, this method is applied to a numerical wave tank where the waves can be optimized before the actual basin testing.