Estimated

4 reported rows against measured corpus ladders (33%); global-corpus-prior over 203 held-out anchors (67%); shrunk 6.2 toward the measured corpus

Estimated

global-corpus-prior over 203 held-out anchors (9%); launch card against 4 resolved peers on 12 rows (91%); exceeds every named peer on 2 of 12 rows; optimism haircut 0 from cross-lab-probe; shrunk 0.2 toward the measured corpus