Decent explainer on EX-08 and what is different about it
And an interesting (albeit from 2016) distribution of the same cohort’s scores across all the models (post #2)
715 wasn’t/isn’t even the 50th percentile, a 650 is roughly the 30th percentile.