Relative performance (RPE) awards have become an important component of executive compensation. We examine whether RPE awards particularly the peer group are structured in a manner consistent with economic theory. For RPE awards using a custom peer group we find that the custom group is significantly more effective than four plausible alternative peer groups at filtering out common shocks lowering the cost of compensation and increasing managerial incentives. For RPE awards using a market index we find some evidence that firms could have selected a custom set of peers with better filtering properties at a lower cost with similar incentives. For example firms could have saved around $118000 in present value terms on average for an RPE award had they chosen a custom group comprising of their product market peers instead of a market index.