Anonymity depends on how you define it and what kind of possible anonymity attacks do you assume. If you also include - for example - waterboarding in the arsenal of your possible attacks, then not many methods are anonymous. But some might still be kind of resistant to this kind of attack - for example, some methods may offer so called "plausible deniability".

But there may be some methods that get very close to "true anonymity" with some reasonable definition of "anonymity". For example, those methods that use zero knowledge proofs may get very close to this (I haven't actually looked yet at any such coin, but zero knowledge proofs seem to be the appropriate mathematical machinery for this kind of job).