I think there’s a human bias where we remember predictions being wrong more strongly than those being right.

The study shows that they weren’t wrong or right in equal measure, they were right more often, considerably so. But that doesn’t mean that they were always right.