View in context, Information, March 2026
He says the results are feduciary grade, not the unacceptable 95% correct that most models do.