The addicted2success.com website has three answers.
1. Break down complex problems
Looking at a question as one big issue usually leads to bad predictions. It's better to unpack the question methodically. Divide it into several smaller questions, and then divide each of them again into even smaller questions. Then examine them.
2. Intellectual challenge
The successful forecasters liked to learn. So they picked out quite a few dots which they could connect, like a puzzle.
3. Looking for multiple views
They were extremely open to hearing other people's opinions. They utilized different perspectives and were able to put together ideas from multiple viewpoints and create well-analyzed predictions.
Book
GARDNER, Dan – TETLOCK, Philip E.: Superforecasting: The Art and Science of Prediction, Crown Publishers, New York, 2015, 352 p.
-jk-