Well what's worrying me is that when you power off a stepper motor, it doesn't hold position. What's worse, is that the unpowered detent torque positions are not aligned with the step positions. Perhaps, in some cases, when you power up the motor again, the rotor will move to the previous position. But if you power up ans step without any delay, the chances of losing position increase considerably. And it makes no difference if you increase the current, this is not lost position due to desynchronising at speed, it's the lack of alignment between the step positions and the rest positions.
The solution to this is to use a "wave drive" current profile, which has only one motor winding on at a time, at positions which are more closely aligned with the unpowered detent positions. Whether this is possible with your stepper shield I can't say, it depends on which shield you are using, and whether you have control of the current in each winding, or whether you send it step and direction signals.
Going back to the easy solutions: the easiest may be not to power off at all, and use a hulking big car battery. Sorry the news isn't better, all I seem to bring you is grief 