I've never been much of a fan of forecasting approaches to intermittent demand. In situations like intermittent demand (or other areas where we have little hope of reasonably accurate forecasts), my thinking is "why bother?" If we can't expect to solve the problem with forecasting, we need a different approach.