Reward level evaluation of parallel sytems