MadLab what helped me with anticipation are two things:
1. I would only do the "finish" (when dog moves from front to heel) once in a while - most of the times I'd reward the recall separately (so maybe I'll release midway with the ball between my legs or maybe I'll have him come to front and then reward etc). The less u do the finish, the less anticipation (same with down with recall)
2. I taught two commands that can be called from the front position - fuss (finish) and "under" (figure eights between my legs). This way I can mix things up while still working the front, finish, etc. Dog learns that even if he anticipates a command after the recall, he doesn't know which one (fuss or have him weave and then fuss)
Summary: be predictably unpredictable
As for Bellon's methods - I never really tried it (I have enough equipment to carry with me, I can't bring a bunch of boxes with me) but from what I've seen the end picture is a nice one, very clear to the dog.
Anthony: I agree with others, you can probably go back to basics. Lure very closely into you (exaggerate the position - I lean back and have the dog's nose literally touching my midsection when I first taught) and reward from the front position, then you can increase distance