Cyclic block copolymers are predicted to assemble into nanostructured domains up to 40% smaller than their linear analogues, making them promising alternatives for nanoscale patterning applications. The limited cyclic block copolymer structures observed experimentally, however, have not met the domain reductions predicted by scaling theory. Through a systematic dissipative particle dynamics simulation study of linear and cyclic block copolymer assembly into lamellar and cylindrical nanostructures, we explore these discrepancies. We find standard implementations of scaling theory, which assume a cyclic BCP behaves as a linear molecule of half the contour length, fail to account for finite chain size effects, resulting in overpredictions of the extent of domain shrinkage upon cyclization. We propose a revised scaling law that clarifies the interplay of chain length, segregation strength, and chain architecture in determining domain spacing reduction attainable by molecular cyclization and offers an explanation for the discrepancies between prior theoretical predictions and experimental results.