Antideuterons are a potential messenger for dark matter annihilation or decay in our own galaxy, with very low backgrounds expected from astrophysical processes. The standard coalescence model of antideuteron formation, while simple to implement, is shown to be under considerable strain by recent data from the LHC. We suggest a new empirically based model, with only one free parameter, which is better able to cope with these data, and we explore the consequences of the model for dark matter searches.

19 pages, 15 figures, C++ code with parametrised cross sections can be found in the ancillary folder, v2 fixes important missing references