CDT doesn't statistically win in these examples: in all cases, if you reason only from what your actions cause you will be in a world where communication is statistically harder, transaction costs are statistically higher, etc. Newcomb's Problem only differs in the certainty of this relationship.