I don't do forecasts or predictions. Indeed, the strategy work I've done using Scenario Mapping / Future Mapping consciously avoids making forecasts, preferring instead to consider alternative and divergent outcomes and what would have to happen for each outcome to become reality.
Nevertheless, I've almost always liked Forecaster / Futurist Paul Saffo's work. His recent article on Six Rules for Effective Forcasting in the Harvard Business Review is no exception. The article is short, accessible, and to the point, so I won't try to summarize.
One idea and ensuing discussion concerns Rule 5: "Look Back Twice As Far As You Look Forward:"