The search for better and cheaper electrocatalysts is vital in the global transition to renewable energy resources. High-entropy alloys (HEAs) provide a near-infinite number of different alloys with approximately continuous properties such as catalytic activity. In this work, the catalytic activity for the electrochemical oxygen reduction reaction as a function of molar composition of Ag-Ir-Pd-Pt-Ru HEA is treated as a landscape wherein it is shown that the maxima are connected through ridges. By following the ridges, it is possible to navigate between the maxima using a modified nudged elastic band (NEB) model integrated in a machine learning NEB algorithm. These results provide a new understanding of the composition space being similar to an evolutionary landscape. This provides a possible new search and design strategy for new catalysts in which the composition of known catalysts can be optimized by following ridges rather than exploring the whole alloy composition space.