The tracer-tracer correlation technique (TRAC) has been widely employed to infer chemical ozone loss from observations. Yet, its applicability to chemistry-climate model (CCM) data is disputed. Here, we report the successful application of TRAC on the results of a CCM simulation. By comparing TRAC-calculated ozone loss to ozone loss derived with the passive ozone method in a chemistry transport model we differentiate effects of internal mixing and cross vortex boundary mixing on a TRAC reference correlation. As a test case, we consider results of a cold Arctic winter/spring episode from an E39/C experiment, where typical features, for example, sufficient polar stratospheric cloud formation potential, denitrification and dehydration, and intermittent and final stratospheric warming events, are simulated. We find that internal mixing does not impact the TRAC-derived reference correlation at all. Mixing across the vortex boundary would lead to an underestimation of ozone loss by ~10% when calculated with TRAC. We provide arguments that TRAC is a consistent and conservative method to derive chemical ozone loss and can be used to extract its chemical signature also from CCM simulations. As a consequence, we will be able to provide a lower bound for chemical ozone loss for model simulations where a passive ozone tracer is not available.