Changes to the statistical architecture of relative binding free energy (RBFE) perturbation maps dramatically improve the accuracy of affinity estimates. In perturbation maps, nodes represent ligands in a lead series of size n, and edges are alchemical transformations. For maximum certainty in predicted potencies, a design would include all edge combinations, or n choose 2 transformations. However, the computational cost to traverse all possible edges becomes prohibitive with increasing n. Experimental designs should therefore allocate resources intelligibly while preserving accuracy. The typical design process, such as that carried out by the open-source tool Lead Optimization Mapper, LOMAP, still relies on heuristic decisions for generating designs and parameters such as node selection and edge number are unexplored. To enable quantitative rigor in RBFE designs, I propose to modernize LOMAP, to find statistically optimal graphs over clusters of ligand series. Beyond optimal design generation, Optimal LOMAP will provide users with theoretically determined design space parameters and statistical tools for improved error correction. This poster on FEP design optimization was presented at OpenEye CUP XXI in March 2022.